Digital Twin Technology for Microelectronics Reliability

Abstract

Electronic components are getting increasingly integrated into a diversity of applications, products, and industries and are becoming an essential part of them. In some cases, they are responsible for handling critical tasks (e.g., the perception system in autonomous driving) and are exposed to harsh environmental conditions (e.g., elevated temperatures). Thus, the reliable functioning of electronics is more significant than ever before. Traditional reliability qualification methods rely on the tests in specification manuals and handbooks and, therefore, hold little significance today. New methods of reliability estimation have emerged and evolved quite a lot over the past six decades. The recent ones focus on product-specific reliability, as it is often the case that identical electronic components experience non-identical operating conditions and environmental loads and, thus, have a variation in their lifetimes. Implementing Prognostics and Health Management (PHM) for electronic components is a promising way to address this challenge. Demonstrating this approach using a Digital Twin-based framework is the primary goal of this dissertation. Organised into six main chapters, this thesis lays out a generalised framework for PHM and its building blocks (in Chapter 1); presents a systematic review of the term 'Digital Twin', its state-of-the-art, definitions, and architectural models (in Chapter 2); explores the physics-of-degradation for electronics packaging and encapsulating materials and identifies two commonly observed package-associated mechanical failure mechanisms (in Chapter 3); showcases a systematic procedure to prepare a six-parameter material model reflecting thermo-oxidative ageing of Epoxy Moulding Compounds (EMC) and its effects on the thermomechanical behaviour of an electronic package (in Chapter 4); focuses on developing and testing in-situ monitoring for package-to-PCB solder interconnects of a wafer-level chip-scale package using a high-resolution piezoresistive sensor (in Chapter 5); and demonstrates a superelement-based Finite Element reduced-order modelling technique and optimises for its accuracy and efficiency (in Chapter 6).