The broader impact/commercial potential of this Partnerships for Innovation - Technology Translation (PFI-TT) project is to enable better cooling for more efficient and powerful computing. This technology development and commercialization project will investigate the potential benefit of 3D printing metal fins directly onto silicon to enable a 10X improvement in cooling compared to standard existing technologies. This technology may enable faster and higher power computing, especially for supercomputers conducting high demand processes, such as training artificial intelligence systems. The research objective is to demonstrate a performance benefit resulting from the optimization of the printing process, mechanical bonding/reliability, and the heat sink design. The education objective is to train a postdoctoral and student team in the essentials of entrepreneurship and leadership through industry mentorship and participation in Binghamton University’s XCEED and regional I-Corps program. Participation will be broadened by inclusion of an LSAMP summer student in the research, and having the student and postdoc team lead a seminar explaining their innovation and entrepreneurial quest at our campus LSAMP’s seminar series.
The proposed project focuses on translating research on laser printing a molten metal alloy, called Sn3Ag4Ti, on a silicon substrate into a minimum viable product for demonstration to potential development and strategic partners. The process includes depositing a thin powder layer of Sn3Ag4Ti onto silicon and exposing the areas where cooling fins are desired to laser heating. Subsequent layers of copper are then deposited onto this layer to form high-aspect ratio cooling fins using selective laser melting. The rapid laser heating forms thin silicide and intermetallic films at the interface. The manufacture of the entire fin assembly using selective laser melting will be researched, specifically how to deposit high thermal conductivity copper metal onto the Sn3Ag4Ti layer. We will also measure the interfacial mechanical properties and thermal cyclability of the device, which are critical to electronic packaging reliability. The design of the heat sink at high heat fluxes, expected in the next generation of packages, will be optimized using neural network techniques. Our research will solve key translational challenges by refining the processing method, demonstrating long-term mechanical reliability, and optimizing the printed heat sink for minimal thermal resistance at high heat fluxes.
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.
