Scalable distributed HEMT model for millimetre-wave applications

A scalable HEMT model is highly desirable for device engineers. Modern electronic devices require scalability up to high frequencies and device engineers need more flexibility to design complex circuits. For high-frequency applications the scalable model should be able to accurately predict behaviour of the device. The major drawback is lack of a proper topology to represent a full device of any size and inaccurate techniques to separate the parameters from the measured data. This thesis explores a distributed model of a HEMT and gives an error-free linear scaling technique for the extrinsic parameters of the device. A new technique was introduced to separate the access network from the active device, and the active device was modelled using a distributed technique. A new topology of unit cell is proposed, which can be used to analyse different HEMT devices and predict the behaviour of unmeasured devices sizes for high frequencies. This model also includes the distributed effects for various widths of devices, where a signal is propagated from the gate to drain metallisation. The effect of the manifold in a distributed model is explored for various widths of devices and a figure of merit of the model is discussed. The linear scalable model was applied to compound devices, varying in the number of fingers and widths, and showed excellent agreement. The effects of the manifold on these HEMTs are explained with respect to the gain of the devices. The model is also used to analyse the drain-current behaviour of a HEMT and establishes linear relationships for the drain current in terms of the unit cell and device metallisation. The model is applied to the determination of the maximum achievable gain-bandwidth product as a function of gate width. This defines the upper limit of achievable performance for further work, and a new and improved manifold and finger layouts.