Experimentally Based HFET Modeling for Microwave and Millimeter Wave Applications
Doctoral thesis, 1998
Transistor models are very important in the design of Monolithic Microwave Integrated Circuits (MMICs). Circuit simulations based on accurate transistor models are one of the keys to high circuit yield. Transistor models can also be used to trace problems in the device fabrication processes. Hence, transistor models are important in the development of new transistors as well. This thesis considers important modeling aspects for Heterostructure Field-Effect Transistors (HFETs) operated in the 1-110 GHz range. Three categories of experimentally based HFET models are discussed: small-signal models, noise models, and large-signal models. Special attention is paid to efficient direct extraction procedures for small-signal models. Models extracted both for commercially available and in-house fabricated HFET devices are presented. An easy-to-use computer program for automatic data acquisition and device characterization is also presented. In fact, this program has drastically decreased the HFET characterization time. Concerning noise models, extraction methods are proposed for both the single-parameter and the two-parameter FET temperature noise model (also known as the Pospieszalski noise model). In addition, it is theoretically shown that, for every frequency, it is possible to find a pair of source admittances that cancel (balance out) the noise contributions from the two internal noise sources in this model. In this work these admittances are successfully used to obtain simple and exact direct extraction formulas for the two-parameter noise model. Moreover, a new method is presented for noise analysis of linear circuits. The method is easy to implement in general purpose CAD-programs. However, it is also shown that the method can be used to derive analytical expressions for the noise parameters of smaller circuits. The well known drain-current control equation used in the Chalmers nonlinear model is extended to account for temperature, dispersion and soft-breakdown effects. Finally, the importance of monitoring the light-sensitivity of on-wafer characterized devices is addressed. It is shown that light-sensitive devices can have a significant minimum in the drain-current for a specific light-intensity.
temperature noise model