Thermal design and management of micro-pin fin heat sinks for energy-efficient three-dimensional stacked integrated circuits

Abstract

Three-dimensional stacked electronics have substantially improved the electrical performance of integrated circuits. However, given the geometrical complexity and high pressure drop they entail, thermal management difficulties and energy requirements are exacerbated owing to the inapplicability of thermal management schemes. In this study, the thermal and hydrodynamic characteristics of various micro-pin fin arrays were investigated to maximize heat dissipation while minimizing the energy consumption. Specifically, a 10 x 10 mm(2) micro-pin fin array was fabricated on an eight-inch silicon wafer via microfabrication. A Pyrex cover was bonded anodically with the top side of the micro-pin fins to prevent leakage, and a titanium/gold thin film serpentine heater was used to supply uniform heat flux on the backside of the micro-pin fin array. Subsequently, the heat transfer and pressure drop in the micro-pin fin heat sinks were obtained experimentally with various micro-pin fin geometries having pin diameter D-f = 38-100 mu m, transverse pin spacing S-T = 74-301 mu m, longitudinal pin spacing S-L = 74-301 mu m and pin height H-f = 90-207 mu m. Thereafter, the geometrical and operational effects on heat transfer and pressure drop were investigated based on a consolidated database cumulated from the literature. Altogether, 256 data points from 21 geometrical combinations were explored from existing relevant studies to obtain optimized geometric and operating conditions in the micro-pin fin arrays over a wide range of geometrical and operating conditions: Reynolds number Re = 35-491.3, heat flux q" = 0-114 W/cm(2), pin diameter D-f = 38-559 mu m, pin spacing S = 74-800 mu m, and pin height H-f = 90-845 mu m. Subsequently, new empirical correlations based on the consolidated database were formulated to describe the Nusselt number and fanning friction factor in the micro-pin fin arrays. These correlations provide suitable predictions in comparison with those based on extant correlations. (C) 2021 Elsevier Ltd. All rights reserved.