A Hybrid Evolutionary Modeling/Optimization Technique for Collector-Up/Down HBTs in RFIC and OEIC Modules

Abstract

A hybrid computer-aided design (CAD) tool for automatic simulation and characterization of advanced collector-up/down heterojunction bipolar transistors (HBTs), which are expected to be used as the high-power-amplifier (HPA) module in radio-frequency integrated circuit (RFIC) and the blue laser diode (LD)/heterojunction phototransistor (HPT) module in optoelectronic integrated circuit (OEIC), has been successfully established. By expanding the essence of reported techniques, the newly-developed equivalent-circuit modeling approach focuses on combining the genetic algorithm (GA) with refined analytical-modeling equations yielding reliable initial values of model parameters. This SPICE-like simulator reliably and efficiently relates device/circuit performances directly to the direct current/radio frequency (dc/RF) measurements since it implicitly accounts for physical correlations among model parameters and does not require ad hoc de-embedding test structures as well as special simplifying assumptions in the optimization period. Evolutionary programming is performed to achieve tedious circuit-level simulations in shortest turnaround times and the computational efficiency permits execution of an optimization involving random generation of circuit parameters. In this study, a detailed comparison of direct-measured, analytical-derived, and numerical-simulated results, for the pnp InGaAs as well as npn InGaP collector-up HBTs as the HPA module in RFIC and the npn ZnSe-Ge collector-down HBT as the blue LD/HPT module in OEIC, is presented. To further demonstrate the accuracy and robustness of this approach, the npn AlGaAs-InGaAs collector-down HBT, which can be incorporated in a preamplifier for optical communication, is investigated.