8/23/2010

Agilent-Documentation of the FBH HBT Model

http://www.cadfamily.com/downinfo/300783.html

1 IntroductionIn recent years, heterobipolar transistors became available in commercial GaAs MMICtechnology. Due to their ability to operate at high current densities they are the devicesof choice for power amplifiers, e.g. in mobile phones. Also InP-based HBT technologies forhigh-speed circuits are available. The lower 1/f noise compared to HEMTs also qualifiesthe devices for oscillator applications.

While technology is mature and industry alreadyships large numbers of HBT-based MMICs, the model development for circuit design lacksbehind. The designer has the choice between about a dozen built-in models for GaAs-basedFETs in standard circuit simulators, but models for GaAs-based HBTs are rare. This onlyleaves the choice to use either sophisticated models taylored for Si devices, the simpleSPICE-type Gummel-Poon (GP) model, or to do the full design with a linear S-parameterbased model.

None of the approaches is satisfactory.This document attempts to fill this gap. It deals with extensions of the GP model thatare necessary for the description of HBTs, and eventually lead to the development of theFBH model.Before going into details, it is necessary to point out that state-of-the-art GaAs-basedHBTs are quite ideal devices.
• The semi-insulating substrate prevents parasitic substrate effects that have to beaccounted for on silicon.
• In a good technology, surface or interface related problems such as parasitic currentsare negligible, and even thermal runaway can be supressed by proper emitter or basefeedback or by thermal shunt technology.
Two effects, however, are most important to simulate HBTs. The first one is self-heating, the second is the current dependence of the transit frequency, caused by highcurrent injection into the collector. These effects will be presented in greater detail in thefollowing.All exampes in this paper are measured at GaInP/GaAs HBTs fabricated on the 400process line of the Ferdinand-Braun-Institut [1].

No comments: