Process-Dependent Performance of Recessed-Gate Enhancement-Mode GaN p-FET

Teng Li, Meng Zhang*, Jingjing Yu, Jiawei Cui, Junjie Yang, Maojun Wang, Yamin Zhang, Shiwei Feng, Jin Wei

*Corresponding author for this work

Research output: Contribution to journalJournal Articlepeer-review

Abstract

There is a strong demand for GaN complementary logic (CL) circuits and power integrated circuits (PICs). The E-mode p-GaN gate high electron mobility transistor (HEMT) platform is a promising solution for GaN CL circuits. Compared to the commercialized E-mode GaN n-FET (i.e., the p-GaN gate HEMT), the E-mode GaN p-FET has fallen far behind. To achieve E-mode operation, gate recess is a common approach, but the channel thickness (tch) and the surface roughness (Rq) of etched p-GaN significantly influence the performance of devices. This work investigates GaN p-FETs with different gate recess depths. A pre-treatment of the etched p-GaN surface is used to smooth the etched p-GaN surface. During the fabrication, tch is found to be a critical process parameter that influences the performance of devices. As the tch decreases, a more robust E-mode operation (i.e., a more negative Vth) is achieved. However, this leads to an increase in on-resistance (Ron). When tch = 8.5 nm, the fabricated E-mode GaN p-FET exhibits a Vth of −1.1 V and a high ID-max of 5.1 mA mm−1. Compared to the E-modep-FET without pre-treatment, the current density more than doubles. This work advances the development of p-FETs on the standard E-mode p-GaN gate HEMT platform.

Original languageEnglish
Article number2400969
JournalPhysica Status Solidi (A) Applications and Materials Science
DOIs
Publication statusE-pub ahead of print - 20 Apr 2025
Externally publishedYes

Bibliographical note

Publisher Copyright:
© 2025 Wiley-VCH GmbH.

Keywords

  • E-modes
  • GaN
  • p-FET
  • threshold voltages
  • wet treatments

Fingerprint

Dive into the research topics of 'Process-Dependent Performance of Recessed-Gate Enhancement-Mode GaN p-FET'. Together they form a unique fingerprint.

Cite this