Introduction to Modeling HBTs

The UCSD model (or ARPA-UCSD model, since it was founded by ARPA) was the first comprehensive GaAs HBT model publicly available. It was developed in the mid-1990s as a collaborative work of numerous semiconductor and circuit simulation companies, coordinated by the High Speed Devices Group at the University of California San Diego (UCSD). The model documentation, including the full code, is published on the Web [13].
It takes into account the following effects:
Early and Webster effects;
Excess storage of charges at heterojunction;
Depletion capacitance model accounting for reach-through;
Transit time and collector capacitance variation due to velocity modulation;
Excess transit time due to base pushout;
Partition of the base-emitter junction into intrinsic and extrinsic parts;
Partition of the base-collector capacitance into intrinsic and extrinsic parts;
Self-heating;
Base-collector breakdown;
Parasitic substrate branch (intended for SiGe-HBTs).
The contribution of this model to the modeling of HBTs in general can hardly be overestimated. Especially, it contains a description of the velocity modulation, or f t peaking, effect based on an analytical physical model [1, 14]. At least this part of the model built the basis for all the models that followed.
On the other hand, the model suffers from certain shortcomings regarding the formulation of the large-signal model. These will be addressed in the following section. It also has rather too many features in some respects, for example, regarding the parasitic base-emitter currents, or concerning the influence of the spike in the conduction band at the...