Use this URL to cite or link to this record in EThOS:
Title: Multiphysics modelling and measurements of microwave-frequency power transistors
Author: Gillespie, Sean J.
ISNI:       0000 0004 7967 3004
Awarding Body: University of Surrey
Current Institution: University of Surrey
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Restricted access.
Access from Institution:
The introduction of next generation mobile systems and alternative applications demands significant improvement in efficiency and linearity of the power amplifier. It is essential to understand the limitations inherent to the power transistor, and improve it for emerging applications. The package containing the power transistor is built up of various components; MOS capacitors, bond wire arrays and ESD protection circuitry common in commercial power transistors in addition to the semiconductor transistor die. The objectives of this work are to develop multiphysics modelling components which are needed for CAD-based design optimisation. An electrothermal bond wire array model is proposed which combines the electrical and thermal characteristics at high-frequency, reproducing the non-uniform distribution seen in measurements. The model returns the temperature distribution over 2000-times faster, in a fraction of a second, compared to an equivalent 3D multiphysics simulation which takes at least 23 minutes, making it well suited to circuit modelling for CAD-design and optimisation. Fundamentally, the foundations of the power transistor is the semiconductor technology that the package holds. Extraction of compact transistor models is time consuming, requiring a multitude of measurement steps and specialised equipment. Behavioural modelling techniques simplify the measurement extraction procedure, while conserving the accuracy and convergence needed for circuit modelling. A primary limiting factor of the X-parameter behavioural model is that it lacks thermal memory effects that are prevalent in most applicable transistor technologies. This work extends the extraction of X-parameters to include dynamic thermal memory effects. Thermal hysteresis is demonstrated and shown to introduce spectral asymmetry during simulations using two-tone and QAM-16 modulation stimulus. Furthermore, distributed multiphysics modelling techniques co-simulate electromagnetics, thermodynamics and semiconductor device physics - the coupled solution providing enhanced insight into the performance and limitations of the transistor compared to lumped modelling approaches. A distributed electrothermal model was developed which enables significant design improvements by optimisation of the thermal and electrical layouts. This is demonstrated at 28~GHz for a GaN pHEMT transistor where a 12% increase in linear output power and 54% increase in PAE is achieved relative to the peak efficiency of the transistor (24%). The multiphysics modelling components developed in this work aim to improve power transistors for next generation mobile systems and applications requiring the efficient application of high-power microwaves.
Supervisor: Aaen, Peter Sponsor: Engineering and Physical Sciences Research Council (EPSRC)
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral