1528
(Invited) AlGaN/GaN HEMT Reliability and Trap Detection Using Optical Pumping
The light sources used in the optical pumping technique consist of four low power lasers: red, green, blue, and violet. The response of the drain current to the illumination is measured using high-speed data acquisition equipment with sampling rates up to 100k samples/s. In the initial experiments a device was exposed to each laser independently from longest to shortest wavelength, i.e., red then green then blue then violet.
AlGaN/GaN HEMTs with gates ranging from 0.125x150 to 0.170x150 microns were pumped before and after high temperature dc stress. The dc stress was at the bias point of VG=-2 and the channel power was maintained at one watt by adjusting the drain voltage at constant temperature of 150⁰C. Devices that degraded gradually have significant increases in the blue energy range which suggests an increase oxygen on nitrogen substitutional defects [M.A. Reshchikov and H. Morkoc; JAP 97, 061301 (2005)]. Other devices showed degradation with noisy fluctuations in drain current (up to 20 mApp), but when optically pumped, showed very little change from the pre-stress response. Trap creation is not suspected in the degradation in these parts, but may be a result of contact or final metal failure.
Trap energies can be determined by analyzing the decay of the drain current after the part has been illuminated. Curve fitting techniques determine the time constants of the decaying curve and when measured at multiple temperatures, activation energies are determined with Arrhenius analysis. An example using this technique with the blue laser (450nm) at 10 temperatures showed a trap of 0.14eV which the literature suggests a carbon on gallium substitutional [Levinshtein, et al., Properties of Advanced Semiconductor Materials : Gan, Aln, InN, BN, SiC, SiGe, 1st ed. New York: Wiley-Interscience, 2001].
The author gratefully acknowleges the support of a DOD MURI monitored by the AFOSR and colleagues who contributed to this work: Rick Deist, Dr. Brent Gila, and Prof. Steve Pearton.