Applications and processing of SiGe and SiGe:C for high-speed HBT devices

The continued growth of high-speed-digital data transmission and wireless communications technology has motivated increased integration levels for ICs serving these markets. Further, the increasing use of portable wireless communications tools requiring long battery lifetimes necessitates low power consumption by the semiconductor devices within these tools. The SiGe and SiGe:C materials systems provide solutions to both of these market needs in that they are fully monolithically integratible with Si BiCMOS technology. Also, the use of SiGe or SiGe:C HBTs for the high-frequency bipolar elements in the BiCMOS circuits results in greatly decreased power consumption when compared to Si BJT devices.Either a DFT (graded Ge content across the base) or a true HBT (constant Ge content across the base) bipolar transistor can be fabricated using SiGe or SiGe:C. Historically, the graded profile has been favored in the industry since the average Ge content in the pseudomorphic base is less than that of a true HBT and, therefore, the DFT is tolerant of higher thermal budget processing after deposition of the base. The inclusion of small amounts of C (e.g. <0.5%) in SiGe is effective in suppressing the diffusion of B such that very narrow extremely heavily doped base regions can be built. Thus the f_T and f_max of a SiGe:C HBT/DFT are capable of being much higher than that of a SiGe HBT/DFT.The growth of the base region can be accomplished by either nonselective mixed deposition or by selective epitaxy. The nonselective process has the advantage of reduced complexity, higher deposition rate and, therefore, higher productivity than the selective epitaxy process. The selective epi process, however, requires fewer changes to an existing fabrication sequence in order to accommodate SiGe or SiGe:C HBT/DFT devices into the BiCMOS circuit.     

Performance comparison of state-of-the-art heterojunction bipolar devices (HBT) based on AlGaAs/GaAs, Si/SiGe and InGaAs/InP

This paper presents a comprehensive comparison of three state-of-the-art heterojunction bipolar transistors (HBTs); the AlGaAs/GaAs HBT, the Si/SiGe HBT and the InGaAs/InP HBT. Our aim in this paper is to find the potentials and limitations of these devices and analyze them under common Figure of Merit (FOM) definitions as well as to make a meaningful comparison which is necessary for a technology choice especially in RF-circuit and system level applications such as power amplifier, low noise amplifier circuits and transceiver/receiver systems. Simulation of an HBT device with an HBT model instead of traditional BJT models is also presented for the AlGaAs/GaAs HBT. To the best of our knowledge, this work covers the most extensive FOM analysis for these devices such as I-V behavior, stability, power gain analysis, characteristic frequencies and minimum noise figure. DC and bias point simulations of the devices are performed using Agilent's ADS design tool and a comparison is given for a wide range of FOM specifications. Based on our literature survey and simulation results, we have concluded that GaAs based HBTs are suitable for high-power applications due to their high-breakdown voltages, SiGe based HBTs are promising for low noise applications due to their low noise figures and InP will be the choice if very high-data rates is of primary importance since InP based HBT transistors have superior material properties leading to Terahertz frequency operation.     

Simulation of SiGe:C HBTs using neural network and adaptive neuro-fuzzy inference system for RF applications

In this article, the small-signal equivalent circuit model of SiGe:C heterojunction bipolar transistors (HBTs) has directly been extracted from S-parameter data. Moreover, in this article, we present a new modelling approach using ANFIS (adaptive neuro-fuzzy inference system), which in general has a high degree of accuracy, simplicity and novelty (independent approach). Then measured and model-calculated data show an excellent agreement with less than 1.68?×?10^?5% discrepancy in the frequency range of higher than 300 GHz over a wide range of bias points in ANFIS. The results show ANFIS model is better than ANN (artificial neural network) for redeveloping the model and increasing the input parameters.