Experimental and Numerical Investigation Ultra-Thin and Multi-Layer Microchannel Oscillating Heat Pipes

The demand for high-performance electronic chips in applications such as artificial intelligence and data centers, has driven the development of three-dimensional integrated circuits (3D-ICs). They are fabricated by vertically stacking multiple dies, which significantly reduces interconnect length and improves signal delay. However, 3D-ICs introduce severe thermal challenges due to the increased power density and limited heat dissipation paths between stacked layers.

To address these challenges, various cooling techniques have been proposed, such as embedded single-phase or two-phase microchannel cooling. While these methods offer high heat transfer performance, they often suffer from high pressure drop and the need for external pumping.

Oscillating heat pipes (OHPs), as a passive cooling technique, could be a potential candidate for 3D-ICs cooling applications. OHPs consist of serpentine channels partially filled with working fluid and vapor bubbles. Self-sustained flow oscillations are induced by the pressure fluctuations caused by liquid evaporation and vapor condensation. OHPs can offer a high heat transfer capability due to the phase change dynamics, yet it does not require external pumping like two-phase embedded microchannel cooling. This makes OHP a competitive solution for future 3D-IC cooling.

Despite the advantages, OHPs have a channel size on the scale of a few mm, which is much larger than the interconnection pitches and die-spacing. OHPs must be scaled down to channel sizes around 100 µm for 3D-ICs cooling applications, yet no studies are available regarding OHPs with 100 µm channel sizes as the very high flow resistance could hinder the startup process of OHPs.

This dissertation investigates the feasibility of ultra-thin and multi-layer microchannel OHPs (micro-OHPs) for 3D-ICs cooling. In Chapter 2, we experimentally studied the effect of channel height on OHP performance using MEMS-fabricated devices with heights from 500 μm to 100 μm. The experimental results support that smaller channel sizes will impede the startup. With modifications in the evaporator locations and channel geometries, a micro-OHP with 100 μm channel height operates successfully to a power input level above 200 W, demonstrating the possibility of miniaturization of OHP channels to 100 μm scale. To better understand the dynamics and aid the design optimization of OHPs, we developed an unsteady one-dimensional (1D) homogeneous model in Chapter 3. The model solves three coupled governing equations, mass, momentum and energy conservation. It is a first-of-its-kind model in that it considers vapor compressibility as the main driving mechanism and successfully reproduces key OHP operating regimes, including startup, oscillation, and dryout. Finally, in Chapter 4, we studied inter-layer thermal interactions in two-layer OHPs through both direct experiments and a thermal-equivalent test structure. These results provide insight into the thermal coupling between stacked OHPs and future refinements to the 1D model for multi-layer integration.

This dissertation provides a foundation for future integration of micro-OHPs in 3D-ICs cooling applications. Fabrication, modeling and experimental validation are discussed for the development of micro-OHPs as a compact, passive thermal solution.