Parametric Analysis of Microfluidic Cooling Systems for Three-Dimensional-Stacked Silicon Microelectronics by Inferential Statistic Approaches

Abstract

This work is a numerical study of microfluidic cooling of integrated circuits (ICs), using embedded micropin-fins on a silicon chip. The study considers non-uniform chip heat fluxes (250–500 W/cm2) and variable pin fin density using DI water as coolant. A parametric analysis was performed, using the theory of design of experiments (DoE) in order to find the best performing configurations. The proposed factorial design considers six geometrical parameters resulting in 64 microfluidic cooling configurations. The pressure drop and average chip temperatures were obtained for each model to determine the importance of input parameters utilizing a statistical approach. Results from this optimization point to different suitable configurations in which the maximum device temperature is below 60 °C, under moderate pressure drops below 80 kPa. This work takes advantage of numerical models and statistical approaches to seek optimal designs of microfluidic cooling systems and to identify key parameters that have influence on their global performance. In addition, alternative configurations are also assessed for cases in which thermal or hydraulic parameters could be traded-off depending on the application. The results from this study are helpful for the design of chip thermal management with nonuniform power distribution.