Although nitride electronics have matured rapidly, the performance and reliability of nitride high electron mobility transistors (HEMT) and other electronic devices have been hampered by electrically active defects that manifest as deep levels in the bandgap and/or as trap states. To alleviate these problems, not only is a fundamental understanding of the defects in GaN and AlGaN necessary for the continued development of nitride electronics, but also correlations of these defects to the performance and reliability limiting problems is required. Using a multipronged effort, defects were quantitatively studied at the component layer level (i.e. GaN and AlGaN) and also in operational HEMTs using techniques uniquely designed to quantitatively characterize defects as deep levels and traps in these devices.

Deep level optical spectroscopy and related methods of trap spectroscopy are applied to several sets of systematically varied GaN and AlGaN materials. Traps in GaN were typically located at E_C−0.25, E_C−0.60, E_C−0.90, E_C−(1.28−1.35), E_C−2.6, and E_C−3.22/3.28 and for AlGaN at E_C−0.87, E_C−1.5, E_C−3.10, and E_C−3.93. It was determined that several traps showed specific dependencies on variations in growth parameters, substrate orientation, dislocation density, and growth method. Physical sources were attributed to most of these states for the first time, and this taxonomy is essential for analysis of trap effects in working AlGaN/GaN transistors, which constitutes the second focus of this research.

To relate defect incorporation with HEMT performance and reliability, constant drain-current deep level optical/transient spectroscopies using gate or drain voltage as the feedback mechanism are developed. This enables simultaneous and quantitative measurement of defect energies and concentrations of individual defects throughout the bandgap in HEMTs, measurement of device relevant parameters (threshold voltage shift and the change in gate-drain access resistance), separation of gate and access regions, and comparison of other HEMT results because of the absolute nature of these measurements. This novel set of techniques is applied to a HEMT for the first time revealing a 105 mV shift in the threshold voltage corresponding to a ∼3.5×10¹¹ cm⁻² trap concentration under the gate and 6 Ω variation in access resistance in which initial calculations reveal a total access region defect concentration of ∼8×10¹² cm⁻². States detected under the gate at E_C−0.59 and E_C−3.3 eV correlate with known dislocation-related and carbon acceptor defects in GaN, respectively, while the ∼ E_C−2.3 and AlGaN-related E_C−3.7 eV levels observed under the gate and in the access region are not localized only to the surface. The access region also exhibited a distinct level at E_C−0.47 not previously seen in thick GaN or AlGaN. The conclusion of this work is that the time dependencies of these deep levels are likely sources of dispersion in AlGaN/GaN HEMTs.

To further refine the capabilities to quantitatively measure defect energies and concentrations in the access regions of HEMTs, an atomic force microscope is adapted to perform nanometer-scale defect characterization. Using scanning Kelvin probe microscopy, evidence of the spatial and time-dependent measurement capabilities is demonstrated. Initial HEMT results are presented and suggest the total trap concentration of ∼10¹² cm⁻² consistent with previous results.