Data-Driven Leaning and Controlling Metallurgy Matters in Dissimilar Metal Joints

This grant will support fundamental research on controlling phases generated at the welded interface between dissimilar metals. Joining dissimilar metals has become increasingly important for creating lightweight, high-performance, and economic structures in various industries. However, the chemical reactions between two dissimilar metals during welding can generate harmful phases like brittle intermetallic compounds. This award aims to address scientific and technical challenges in controlling the metallurgical phase formation in the weld by introducing a suitable interposing metal to create a nonlinear alloy composition pathway through the joint thickness. The situation is illustrated by the laser welding (LW) of aluminum and copper. These two metals are major materials in the assembly of battery cells, which are in high demand for electric vehicles. This research will enable engineers to design and transform the metallic phases in the weld in a controllable fashion with more freedom than when limited to the two base metals. In turn, this award can broaden the adoption of dissimilar metal joints in industries such as automotive, aerospace, power generation, marine application, medical devices, and information technology. This research involves several disciplines, including manufacturing, materials science, multiscale simulations, and machine learning. The multi-disciplinary approach will broaden the participation of underrepresented groups in research and positively impact both undergraduate and graduate education. The investigators will design and realize optimal bonding phases with a data-driven paradigm to learn and control metallurgic phases in dissimilar metal joints. The research team will conduct multiscale simulations for data generation to establish data-driven models which provide high-fidelity welding predictions. Metallurgical reactions at the bonding interface will be explained using calculation of phase diagrams (CALPHAD)-based analysis, correlation analysis and molecular dynamics simulations. Machine learning will be used to provide inverse design of the nonlinear modification and laser welding processes, and simulation and designs will be validated experimentally. This research will fill the knowledge gap in understanding the interactions between LW energy inputs, keyhole dynamics, phase formation, and transition from liquids to solids under different LW conditions. It will build an efficient methodology, via thermodynamics and kinetics, to predict preferred phases and properties to meet a joint’s requirements.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.