Nonlinear modelling of GaAs and GaN high electron mobility transistors

Accurate modelling of transistors is considered to be an invaluable foundation for successful circuits. This dissertation presents a new approach to scalable, large-signal modelling of GaAs and GaN microwave HEMT devices, that can be used to accurately synthesise arbitrary device geometries. The modelling technique involves an extraction of frequency-independent small-signal intrinsic model parameters that linearly scale with zero offset in proportion to device width. These parameters are extracted from multi-bias steady-state small-signal measurements, i.e. thermal and trap states, are embedded in the extracted small-signal model parameters. Local temperature rise and trap-state of the measured device are separately characterised. A lumped-element thermal model is formulated that can accurately model the local temperature rise, i.e. at the individual gate finger, and scales correctly with number of fingers. A newly- developed trap-state characterisation technique that is used to fit an adequate trap circuit model over a wide bias range is presented. The thermal and trap circuit models form part of the complete large-signal model, in which the interaction of the thermal and trap models are critical to the model completeness, scalability and accuracy, as well as the consistency with the small-signal model. 

The complete model is fitted to the extracted multi-bias small-signal intrinsic parameters. The model is shown to accurately predict the small-signal and large-signal performance of arbitrary synthesised device geometries. The model is then used to design a GaN PA which benefited from the scalability and built-in electrothermal simulation capability of the model to achieve best electrical and thermal performance.