Experimental Characterization of the Thermal Time Constants of GaN HEMTs Via Micro-Raman Thermometry

• Main Authors: Bagnall, Kevin Robert (Contributor), Saadat, Omair Irfan (Contributor), Jayanta Joglekar, Sameer (Contributor), Palacios, Tomas (Contributor), Wang, Evelyn (Contributor)
• Other Authors: Massachusetts Institute of Technology. Department of Mechanical Engineering (Contributor), Massachusetts Institute of Technology. Microsystems Technology Laboratories (Contributor)
• Format: Article
• Language: English
• Published: Institute of Electrical and Electronics Engineers (IEEE), 2017-07-21T16:00:14Z.
• Subjects: Article
• Online Access: Get fulltext

 Gallium nitride (GaN) high-electron mobility transistors (HEMTs) are a key technology for realizing next generation high-power RF amplifiers and high-efficiency power converters. However, elevated channel temperatures due to self-heating often severely limit their power handling capability. Although the steady-state thermal behavior of GaN HEMTs has been studied extensively, significantly fewer studies have considered their transient thermal response. In this paper, we report a methodology for measuring the transient temperature rise and thermal time constant spectrum of GaN HEMTs via time-resolved micro-Raman thermometry with a temporal resolution of 30 ns. We measured a broad spectrum of time constants from ≈130 ns to ≈3.2 ms that contribute to the temperature rise of an ungated GaN-on-SiC HEMT due to aggressive, multidimensional heat spreading in the die and die-attach. Our findings confirm previous theoretical analysis showing that one or two thermal time constants cannot adequately describe the transient temperature rise and that the temperature reaches steady-state at 16L²/π²α, where L and α are the thickness and thermal diffusivity of the substrate. This paper provides a practical methodology for validating transient thermal models of GaN HEMTs and for obtaining experimental values of the thermal resistances and capacitances for compact electrothermal modeling.