Preprint / Version 3

Semiconductor Switching Interpretability Theory: Establishing Interpretability for Semiconductor Switching Behaviors by Bridging Circuit Theory, Conservation Laws and Semiconductor Physics

##article.authors##

  • Wucheng Ying Electrical Engineering Division, Department of Engineering, University of Cambridge
  • Hui Zhao College of Intelligent Robotics and Advanced Manufacturing, Fudan University
  • Jinwei Qi School of Microelectronics, Xi'an Jiaotong University, Xi'an, China
  • Ameer Janabi Electrical Engineering Division, Department of Engineering, University of Cambridge
  • Biao Zhao
  • Teng Long Electrical Engineering Division, Department of Engineering, University of Cambridge

DOI:

https://doi.org/10.31224/5474

Keywords:

power electronics, switching analysis, Transistors, electrical and electronic engineering

Abstract

Global electricity demand is expected to more than double by 20501-8, driven by emerging loads including AI data centers,3,4 electrified transport5,6, heat pumps7,  electrolytic hydrogen production8 and robotics2. The global sustainability urgently demands green electrical and electronic engineering (EEE), where semiconductors are core components. However, semiconductor switching behaviors have been treated as a “black box” lacking interpretability since 19479, placing fundamental limits on semiconductor sciences, engineering and downstream applications. Here we present the Semiconductor Switching Interpretability Theory (SSIT), which provides interpretability for semiconductor switching behaviors and interactions between semiconductors and other circuit elements, by bridging circuit theory10, conservation laws and semiconductor physics. SSIT carries profound implications, including establishing a new fundamental groundwork and a wealth of foundations and directions for future investigation. As examples, SSIT yields a switching-energy-loss prediction model (errors: 0.88–11.60%), achieving a 17-fold average error reduction compared to the existing model (errors: 34.41–80.05%); it enables unprecedented causal-mechanism interpretations for switching waveforms and interactions between the semiconductor and other circuit elements across scenarios. It informs semiconductor engineering and downstream applications individually, and further enables future-generation co-design between them—currently treated separately—for optimal system performance such as maximized energy and emission savings and reliability. It also opens directions across disciplines including semiconductor materials,11,12 device structure engineering,13,14 packaging,15,16 reliability,17,18 thermal management;19,20 and downstream applications such as power electronics (an EEE’s sub-field handling electric power using semiconductors), and potentially across EEE sub-fields, including higher-frequency communication and computation devices and integrated circuits.21,22

Downloads

Download data is not yet available.

Downloads

Posted

2025-09-29 — Updated on 2026-04-27

Versions

Version justification

revised version R22