Search

Thermoelectric coolers (TECs) offer a promising solution for direct cooling of local hotspots and active thermal management in advanced electronic systems. However, TECs present significant trade-offs among spatial cooling, heating and power consumption. The optimization of TECs requires extensive simulations, which are impractical for managing actual systems with multiple hotspots under spatial and temporal variations. In this study, we present a novel machine learning-assisted optimization algorithm for thermoelectric coolers that can achieve global optimal temperature by individually controlling TEC units based on real-time multi-hotspot conditions across the entire domain. We train a convolutional neural network (CNN) with a combination of the Inception module and multi-task learning (MTL) approach to comprehend the coupled thermal-electrical physics underlying the system and attain accurate predictions for both temperature and power consumption with and without TECs. Due to the intricate interaction among passive thermal gradient, Peltier effect and Joule effect, a local optimal TEC control experiences spatial temperature trade-off which may not lead to a global optimal solution. To address this issue, we develop a backtracking-based optimization algorithm using the machine learning model to iterate all possible TEC assignments for attaining global optimal solutions. For any m by n matrix with NHS hotspots (n, m <= 10, 0<= NHS <= 20), our algorithm is capable of providing 52.4% peak temperature reduction and its corresponding TEC array control within an average of 1.64 seconds while iterating through tens of temperature predictions behind-the-scenes. This represents a speed increase of over three orders of magnitude compared to traditional FEM strategies which take approximately 27 minutes., Comment: This article has been submitted to Journal of Applied Physics under review

While the state-of-the-art commercial microelectronic device has achieved a technology node as small as 5nm and even below 5nm, most conventional thermal management techniques are limited to the macroscale level. The serious self-heating effect in micro- and nano-scale transistors necessitates the development of advanced cooling techniques to address the increasingly demanding yet delicate thermal management challenges. In this thesis, we present a novel lateral thermoelectric cooler based on holey silicon that enables dynamic thermal management in microelectronic systems. This holey silicon-based lateral TEC exhibits exceptional TEC cooling performance and is compatible with the conventional microfabrication process, allowing for direct integration into chip architecture and creating an all-in-one package system. In the subsequent chapters, a comprehensive investigation of the transistor-TEC system is conducted through 2D and 3D simulations using the COMSOL Multiphysics platform. The power transistor (LDMOS) is considered as the cooling object to explore the feasibility of utilizing holey silicon-based lateral TECs for dynamic thermal management in both spatial and temporal domains. Initially, a parametric study under steady-state conditions examines the influence of various geometrical parameters, boundary conditions, and material properties on TEC cooling performance. Subsequently, a transient TEC analysis investigates its cooling capabilities when employing transient TEC current pulses for managing constant heat fluxes and transient heat pulses. Finally, a comparative study of three different array designs is performed to evaluate their effectiveness in achieving efficient spatial thermal management. In summary, this thesis demonstrates significant advancements in holey silicon-based lateral TECs' cooling performance while showcasing their potential for dynamic thermal management in space and time. Additionally, it provides valuable insights into optimizing the