Research on GaN HEMTs for emerging electronic applications

Abstract

The escalating global demand for energy-efficient power electronics has driven the urgent need for high-power density, compact designs, and cost-effective solutions. While conventional silicon-based devices are approaching fundamental performance limits, gallium nitride (GaN) has emerged as a promising wide-bandgap semiconductor. Leveraging its superior material properties including high critical breakdown field, high electron mobility, and high thermal stability, GaN high electron mobility transistors (HEMTs) have revolutionized power conversion systems across diverse sectors: consumer electronics (fast chargers, domestic appliances), automotive industries (electric vehicles), and industrial infrastructures (data centers, smart grids). Emerging applications in artificial intelligence (AI) server power supplies, humanoid robots, and space exploration further underscore their transformative potential in next-generation energy systems. To fully exploit GaN HEMTs in emerging electronics, several critical challenges demand sustained investigation: 1) Contact engineering for optimized metal stacks to minimize on-resistance while maintaining high current capacity. 2) Leakage suppression techniques enhancing breakdown performance and ensuring long-term operational reliability. 3) Monolithic integration methodologies to balance high-frequency operation with logic functionality for advanced power integrated circuits (ICs). 4) AI-driven hardware-software co-design strategies to accelerate Internet of Things (IoT) and edge intelligence implementation. Addressing these challenges will accelerate the development of GaN-based power devices with enhanced power density, multi-domain scalability, and fault-tolerant architectures. This thesis presents a comprehensive study of GaN HEMTs for emerging electronic applications. The first work introduces innovative ohmic contact engineering through silicon-incorporated metal stacks, achieving ultra-low contact resistance via barrier doping optimization and nitrogen vacancy modulation, significantly improving the output characteristics. The second work develops a plasma-assisted surface treatment protocol that reconstructs gate interfaces via thin oxide layers formation, fundamentally suppressing leakage paths while enabling optimal electric field redistribution for enhanced breakdown performance. The third work pioneers a charge trapping layer enabled monolithic integration power ICs platform, overcoming traditional E-mode operation limitations through adaptive threshold modulation for high-frequency logic circuits. The fourth work establishes a neuromorphic in-sensor computing paradigm via GaN HEMT-based reservoir networks, merging gas sensing with edge-optimized machine learning architectures. In summary, this thesis establishes a comprehensive innovation research framework spanning from fundamental fabrication engineering to device performance optimization, enabling diverse application systems. The developed solutions: contact resistance minimization, leakage current suppression, logic circuits design, and sensing-computing convergence, collectively address performance trade-offs that have constrained III-nitride device applications. By enabling GaN HEMTs to simultaneously deliver high power density, MHz-range switching frequencies, and embedded intelligence, this thesis aims to pave the way for transformative applications in two strategic domains: ultra-efficient power converters for renewable energy systems and electrified transportation; and self-adaptive sensor-processor networks for smart factories and IoT ecosystems. These advancements position GaN technology as a cornerstone for global energy transition and AI-driven industrial evolution.