Fast and Accurate Temperature-Dependent Current Modeling of HBTs Using the Dimension Reduction Method

Abstract

Empirical models have been widely and successfully used in device modeling in the past few decades. However, they are becoming increasingly intricate to accurately capture the complex thermal effects in semiconductor devices. Therefore, the aim of this work is to utilize a general dimension-reduction method to quickly and accurately construct large-signal models of semiconductor devices with consideration of thermal effects. In general, the junction voltage dimension is represented by empirical functions, whereas the junction temperature dimension is described by the first-order Taylor series approximation. The final analytical current model is a combination of two independent sets of empirical functions. These functions are constructed from pulsed I-V measurements at different ambient temperatures. The percentages of different components are controlled by the thermal level. Two commercial InGaP/GaAs heterojunction bipolar transistors are investigated to verify the effectiveness of this method. The large-signal models are implemented in Advanced Design System. Excellent agreement is achieved between measurement and simulation of the I-V characteristics, S-parameters, and power sweeps. The dimension reduction method is able to effectively reduce the number of equations and parameters because the temperature dimension is expanded by using a Taylor series. In addition, this method would be applied to the thermal modeling of various devices based on new materials or process technologies. Accordingly, the dimension reduction method is powerful and useful in fast and accurate thermal modeling of microwave semiconductor devices.