Mr. Yu-Sheng Tai

Abstract:

The third-generation semiconductors, gallium nitride (GaN) and silicon carbide (SiC), have attracted extensive attention due to their exceptional material properties and potential in high-performance electronic and optoelectronic devices. Among them, GaN is especially fascinating for high-frequency applications. This advantage arises from the formation of a two-dimensional electron gas (2DEG) at the AlGaN/GaN heterojunction interface, which enables high electron mobility and supports device architectures such as high electron mobility transistors (HEMTs). By contrast, SiC is renowned for its superior thermal conductivity, wide bandgap, and high critical electric field, which together inspire its capability for handling high power density and ensuring robust thermal management. When these two materials are combined—namely, GaN epitaxially grown on SiC substrates—the resulting heterostructure becomes a highly attractive platform for realizing high-power and high-frequency HEMTs. Such devices are promised to play a critical role in next-generation wireless communication systems, power electronics, and radar applications.

Although the lattice constant mismatch between GaN and SiC is relatively small, the dislocation density in conventional GaN-on-SiC epitaxial structures remains on the order of 10^9 cm^−2. This high density of threading dislocations and related crystalline defects significantly degrades device performance, limiting the reliability, efficiency, and lifetime of HEMTs. Traditional approaches to mitigate defect density mainly rely on optimizing epitaxial growth parameters through methods such as buffer layer engineering, substrate miscut angle control, and multi-step growth processes. However, these approaches are both time-consuming and highly costly, demanding extensive iterative experimentation that is impractical for large-scale production. Consequently, innovative substrate engineering concepts have been explored as an alternative pathway to address these challenges.

One promising strategy is the use of patterned SiC substrates to improve GaN epitaxial quality. By introducing micro-/nano-sized structures onto the SiC surface, epitaxial strain can be more effectively distributed, and dislocation propagation can be hindered, leading to a significant reduction in defect density within the active GaN layer. Building upon this concept, our team has recently developed a new class of engineered substrates, referred to as meta-substrates. These are fabricated by creating periodic meta-structures directly on 4H-SiC substrates, designed to enhance defect annihilation, suppress dislocation propagation, and tailor strain relaxation during GaN epitaxy. Unlike conventional patterned substrates, the meta-substrate approach offers new degrees of freedom for crystal quality optimization.

In this presentation, we report on the complete HEMT device fabrication processes carried out on these SiC meta-substrates, including ohmic and Schottky contact formation, passivation strategies, and gate metallization. Also, we evaluate the radio-frequency (RF) performance of the fabricated devices, highlighting their advantages in terms of cut-off frequency and device performance when compared with conventional GaN-on-SiC HEMTs.

The results confirm that meta-substrates provide a viable pathway for advancing GaN-on-SiC technologies, enabling scalable and cost-effective solutions for high-power and high-frequency electronics. This work demonstrates how advanced substrate design can unlock new performance benchmarks in GaN-based HEMTs. We believe that the meta-substrate concept represents not only a significant advancement in semiconductor device fabrication but also a strategic opportunity to accelerate the deployment of GaN/SiC HEMTs in emerging applications such as 5G/6G communications, electric vehicles, and next-generation radar systems.