A Custom Iii-v Heterojunction Bipolar Transistor Model

Communication systems today comprise the major use of GaAs technology with the highest volumes found in the cellular handset front-end. Here, heterojunction bipolar transistor (HBT) power amplifiers (PA) and Pseudomorphic High Electron Mobility Transistor (PHEMT) switches enjoy a comfortable market share. In this environment, the demands of production design require a robust CAD system with accurate and verified compact models for both active and passive devices. In this paper, we will outline some of the work done at RFMD to develop and support scalable HBT models suitable for handset power amplifier design. In the first section, a brief outline is given of the evolution of a custom HBT model from a basic Gummel-Poon formulation to one encompassing more GaAs physics. It is written in Verilog-A and and runs in multiple environments. In the second section, a model of a single device is shown that can be scaled to simulate the behavior of large output arrays. This will include both electrical and thermal aspects. Finally, we will present validation data to illustrate the performance of the model. A cross-section of a simple two-finger HBT is shown in Figure 1. In these devices, the emitter is made of a wider bandgap material such as AlGaAs or InGaP while the base has a narrower bandgap, typically GaAs. In a single-heterojunction device, the base, collector and sub-collector will all be of the same material, while in a doubleheterojunction bipolar transistor (DHBT) the collector will use a wider material. The energy band diagram for the more general DHBT is shown in Figure 2. In these systems, the potential barrier seen by base holes in the valence band is higher than that seen by emitter electrons in the conduction band. This results in higher emitter injection efficiency, leading to higher gain. In a conventional homojunction bipolar junction transistor (BJT), high injection efficiency requires a highly doped emitter and a thicker, lower doped base, increasing the base resistance and base transit time. The theory behind this dates back to the early days of the transistor.1