Search

The demand for electronic circuits with increased performance and reliability is causing industry to focus its attention more and more on thermal-aware design. Because of the higher values of dissipated power characterizing the semiconductor devices in modern technologies, the adoption of efficient cooling solutions represents the current challenge for companies and research centers. In this context, numerical thermal simulations are a valuable instrument since they can be employed to estimate the thermal behavior of the domain under analysis through temperature and heat maps, thus supporting the development of efficient cooling strategies. Devices operating at high temperatures are also subjected to detrimental electrothermal phenomena, which affect the system performance and, consequently, deserve to be examined. This thesis investigates the thermal and electrothermal behavior of state-of-the-art technologies in radiofrequency applications and power circuits. The features of modern semiconductor devices and their main thermal issues are both thoroughly presented. The technology overview covers GaAs-based heterojunction bipolar transistors, SiC-based power MOSFETs, and GaN-based high electron mobility transistors. At the same time, packaging techniques for RF amplifiers, high-voltage power modules, and printed circuit board are explored by highlighting their effect on the electrothermal behavior of devices. A routine devised by the Author to perform finite-element method thermal simulations is presented; such a tool exploits the livelink between MATLAB and COMSOL Multiphysics to obtain exceptionally accurate 3-D representations of the domains under analysis. The comparison between wire-bonding and flip-chip packaging techniques for RF devices, the electrothermal analyses of HBTs arrays, the double-sided cooled technology for power modules, the thermal modeling of cooling solutions for circuits on PCB, and the effect of technology fluctuation of SiC MOSFETs represent only a few examples of the contributions proposed in this work.

An accurate and efficient algorithm is presented, which allows deriving the structure function of a discretized 3-D heat conduction problem. In this approach, the partial thermal conductances and capacitances in the structure function are computed in terms of weighted spatial averages of thermal resistivity and volumetric heat capacity. As a result, the exact influence of all materials and geometric details on each part of the structure function is determined. The approach is validated on a state-of-the-art SiGe heterojunction bipolar transistor (HBT) for high-frequency applications.

The development of efficient, cheap, compact, and reliable storage systems is getting increasingly crucial. Special attention must be given to Li Ion batteries, which stand out for the stability of their open circuit voltage (VOC), current capability, and compactness [1]. Such features make them suitable for a large scale of applications. Unfortunately, the design of electric/electronic systems embedding batteries as power supply/storage is still challenging because of the lack of simple and trustworthy circuital models accounting for complex thermo electrochemical mechanisms. In this abstract, a SPICE compatible compact electrical model of batteries including thermal and chemical effects (hereinafter referred to as macrocircuit) is proposed (Fig. 1). The macrocircuit was customized on a Li ion pouch cell battery in the NMC technology (schematically depicted in Fig. 2), its capacity being C=20 Ah; however, the model can be adapted to a generic product. First, in both the charge and discharge phases, self heating effects are considered along with their impact on the overall electrical behavior of the system. Furthermore, for each simulation, the initial conditions of the state of charge (SoC) of the battery can be defined; the SoC can be then monitored during the simulation run. The macrocircuit is composed by: (i) a core including a voltage generator VOC and a passive network, the values of which are allowed to vary with both the SoC and the cell average temperature (Tavg); (ii) a thermal feedback block (TFB) based on Foster I equivalent network, which provides Tavg as an output; (iii) some behavioral modeling blocks devoted to evaluate the SoC and the dissipated power (PD). It must be remarked that PD is given by the algebraic sum of two components, namely, the irreversible (PD,irr) and the reversible (PD,rev) one. While the former is always positive being it given by the Joule heating dissipation over the core, the latter accounts for the endothermic/exothermic chemical reactions dictated by the entropic coefficient dU/dT occurring in the battery; therefore, PD,rev can also be negative, thus reducing the overall PD and providing cooling effects. The modeling of the above quantities was based on the meticulous experimental campaign conducted in [2]. In order to prove the accuracy of the macrocircuit, simulative results obtained in the OrCAD SPICE software package [3] were compared with the counterpart performed in the FEM environment, which was used as a reference. The 3 D structure and the corresponding multiphysics FEM problem was (i) calibrated by means of the experimental data shown in [4], and simulations were run in COMSOL [5]. Given the initial state of charge SoC0=80% and Tamb=25°C, the discharge of the battery was emulated by defining an outgoing current at 4.5 C (i.e., 90 A). A good agreement between results carried out through the macrocircuit and FEM simulations was achieved in terms of dissipated power (Fig. 3), cell average temperature (Fig. 4), and voltage at the battery terminals (Fig. 5). It must be remarked that each FEM simulation required approximately 10 minutes, while a few seconds were needed by making use of the proposed model. As a further advantage, the model is prone to be adopted in complex SPICE simulations of circuits for applications in power electronics (such as DC/DC converters, inverters, and rectifiers) as well as to study the electrical connection of several pouch cells forming a battery pack.

This article focuses on the moment matching methodology for deriving dynamic compact thermal models (DCTMs) of electronic components. For the first time, an a priori error bound is theoretically established both in the time and frequency domains, which is suited to accurately estimate the approximation error introduced by DCTMs in practical electronics applications. By exploiting this result, it is also shown that the near-optimal choice of matching points previously proposed by one of the authors minimizes the error bound here derived, among all the possible choices of matching points. The introduced error bound is verified by applying the moment matching algorithm to a state-of-the-art packaged InGaP/GaAs heterojunction bipolar transistor (HBT) used in output stages of power amplifiers for handset applications.