Litcius/Paper detail

Achieving Ultralow Specific Contact Resistivity in Ti/n<sup>+</sup>-GaN Ohmic Contacts by Mitigating the FLP Effect with a Gallium Oxide Interlayer

Shujie Xie, Jiaheng He, Xuankun Wu, Zhe Cheng, Lian Zhang, Changxin Mi, Qiao Xie, Yun Zhang

2025ACS Applied Electronic Materials5 citationsDOI

Abstract

As gallium nitride (GaN) devices are scaled for higher-frequency performance, their advancement is increasingly limited by parasitic delays due to elevated Ohmic contact resistance. To mitigate this, selective-area growth n-type doped GaN (n + -GaN) with titanium (Ti) as the Ohmic contact metal has been widely used, achieving specific contact resistivity in the range of 1 × 10 –7 Ω·cm 2 . However, further reductions of Ti/n + -GaN interfacial specific contact resistivity are constrained by the Fermi-level pinning (FLP) effect that originated from the metal-induced gap states and interfacial dangling bonding states. In this study, we propose an approach to relieve the FLP effect and achieve ultralow contact resistivity by forming an approximately 2 nm gallium oxide passivation layer at the Ti/n + -GaN interface through air annealing of the n + -GaN surface. This passivation method yields 0.24 eV Schottky barrier height and a low specific contact resistivity of 3 × 10 –8 Ω·cm 2 for GaN Ohmic contact. Atomic force microscopy (AFM), transmission electron microscopy (TEM), and energy-dispersive X-ray spectroscopy (EDX) confirm the formation of various oxide layers under different annealing conditions. This study demonstrates an effective strategy for reducing Ohmic contact resistance, addressing parasitic resistance, and enabling further scaling of GaN devices for enhanced performance.

Topics & Concepts

Ohmic contactGalliumElectrical resistivity and conductivityMaterials scienceContact resistanceOxideElectrical contactsOptoelectronicsNanotechnologyMetallurgyPhysicsLayer (electronics)Quantum mechanicsSemiconductor materials and devicesGaN-based semiconductor devices and materialsAdvanced Memory and Neural Computing