Design Principles of Thermoelectric-Microchannel Hybrid Cooling Modules for Hotspot Thermal Management

Abstract

Hotspot thermal management is crucial for microprocessors, radial-frequency electronics, and power electronics. Hybrid cooling combining thermoelectrics and microchannels (TEC-MC) offers an effective solution for active and precise temperature control. This work develops an analytical model for predicting the heat flux, hotspot temperatures, and coefficient of performance ($COP$) of TEC-MC hybrid coolers, by treating thermoelectrics as an adjustable thermal resistor. The model incorporates heat spreading resistances to account for both the in-plane and the cross-plane heat conduction from the hotspot, enabling computation of hotspot temperatures four orders of magnitude faster than three-dimensional finite element simulations. Our method can be seamlessly interfaced with multi-objective optimization algorithms for the co-design of TEC and MC. Results revealed intricate correlations among different parameters. An optimal thickness of thermoelectric legs is identified which scales linearly with the filling ratio of TEC when optimizing the cooling power. On the other hand, thinner thermoelectric legs are favored when optimizing $COP$. Moreover, as the heat transfer performance of the MC heat sink improves, the reduced hot-side temperature of the TEC allows for a further decrease in TEC thickness, leading to higher $COP$. Finally, the Pareto front is identified to quantify the trade-offs between the maximum cooling power and the optimal $COP$. We proposed a co-design workflow and showed that simultaneously decreasing the thickness of thermoelectric legs and the thermal resistance of the MC is pivotal for achieving both high cooling power and improved $COP$. This study offers a guideline for developing hybrid cooling systems for hotspot thermal management.