A particle erosion model of monocrystalline silicon for high heat flux microchannel heat exchangers

Abstract

As package-level heat generation pushes past 1 kW/cm3 in various military, aerospace, and commercial applications, new thermal management technologies are needed to maximize efficiency and permit advanced power electronic devices to operate closer to their inherent electrical limit. In an effort to align with the size, weight and performance optimization of high temperature electronics, cooling channels embedded directly into the backside of the chip or substrate significantly reduce thermal resistances by minimizing the number of thermal interfaces and distance the heat must travel. One implementation of embedded cooling considers microfluidic jets that directly cool the backside of the substrate. However, as fluid velocities exceed 20 m/s the potential for particle erosion becomes a significant reliability threat. While numerous particle erosion models exist, seldom are the velocities, particle sizes, materials and testing times in alignment with those present in embedded cooling systems. This research fills the above-stated gaps and culminates in a calibrated particle-based erosion model for single crystal silicon. In this type of model the mass of material removed due to a single impacting particle of known velocity and impact angle is calculated. Including this model in commercial computational fluid dynamics (CFD) codes, such as ANSYS FLUENT, can enable erosion predictions in a variety of different microfluidic geometries. First, a CFD model was constructed of a quarter-symmetry impinging jet. Lagrangian particle tracking was used to identify localized particle impact characteristics such as impact velocity, impact angle and the percentage of entrained particle that reach the surface. Next, a slurry erosion jet-impingement test apparatus was constructed to gain insight into the primary material removal mechanisms of silicon under slurry flow conditions. A series of 14 different experiments were performed to identify the effect of jet velocity, particle size, particulate concentration, fluid viscosity and time on maximum erosion depth and volume of material removed. Combining the experimental erosion efforts with the localized particle impact characteristics from the CFD model enabled the previously developed Huang et al. cutting erosion model to be extended to new parameter and application ranges. The model was validated by performing CFD erosion simulations that matched with the experimental test cases in order to compare one-dimensional erosion rates. An impact dampening coefficient was additionally proposed to account for slight deviations between the CFD erosion predictions and experimental erosion rates. The product of this research will ultimately enable high fidelity erosion predictions specifically in mission-critical military, commercial and aerospace applications.