Nonlinear Characterization of Wideband Microwave Devices and Dispersive Effects in GaN HEMTs

Abstract: Measurements play a key role in the development of microwave hardware as they allow engineers to test and verify the RF performance on a system, circuit, and component level. Since modern cellular standards employ complex modulation formats with wider signal bandwidths to cope with the growing demand of higher datarates, nonlinear characterization using wideband stimuli is becoming increasingly important. Furthermore, III-N semiconductor technologies such as gallium nitride (GaN) are to a larger extent utilized to enable higher performance in microwave circuits. However, GaN is highly frequency-dispersive due to trapping phenomena and thermal effects. This thesis deals with the development of nonlinear measurement instruments as well as characterization of dispersive effects in GaN high-electron-mobility transistors (HEMTs). A measurement setup for wideband, nonlinear characterization of microwave devices has been designed and verified. The setup allows for simultaneous acquisition of low-frequency and radio-frequency signals from DC up to 4~GHz through the use of wideband signal generators and measurement receivers. This enables measurement scenarios such as multi-band load-pull and large-signal characterization of IQ-mixers, which can give useful insight into how to optimize the performance in a RF transmitter. Electrothermal characterization of GaN devices has been carried out using conventional measurement methods such as pulsed I-V, and it is shown that trapping phenomena and thermal effects due to self-heating or mutual coupling are challenging to separate. Multiple methods must be utilized to fully characterize both the large-signal and small-signal impact on device performance. A new characterization method has been developed, for extraction of thermal transfer functions between mutually coupled devices on e.g. a semiconductor wafer. The method does not require any DC-bias on the measured devices, which can potentially reduce the influence of trapping during characterization of thermal properties in materials.