Synthesis of Ag-Sn Phases Predicted with Machine Learning

Authors: Jeff Lam, Olesya Gorbunova, Md Ariful Islam, Ganesh Tiwari and Tibendra Adhikari

Field of Study: Science, Technology, Engineering, and/or Math

Program Affiliation: Scalable Asymmetric Lifecycle Engagement (SCALE)

Faculty Mentors: Zhong Lin, Aleksey Kolmogorov

Easel: 13

Timeslot: Morning

Abstract: Machine learning has the potential to accelerate materials discovery for targeted applications. In electronics packaging, continued miniaturization demands new solutions, with Sn-rich alloys emerging as promising lead-free interconnects. The Ag-Sn binary was recently explored with a combination of machine learning, evolutionary optimization, and ab initio modeling, revealing previously unknown Sn-rich intermetallics stable at elevated temperatures. In this project, these predictions have been tested experimentally using flux growth synthesis, where Sn, due to its low melting point, acts as flux for dissolving raw materials at high temperatures before crystallizing target phases through controlled cooling. To ensure safe processing conditions, Python-based software was developed to monitor and manage vapor pressure during high-temperature treatments. Energy dispersive X-Ray spectroscopy (EDX) analysis performed on synthesized alloys suggests binary compounds with compositions close to the predicted 1:2 and 1:4 stoichiometries are present. This work was supported by the Department of Energy through grant DE-SC0021202.