Ultra-scaled SiGeC HBTs beyond the existing roadmap - A simulation based study

点击翻译按钮获取中文摘要

Silicon-germanium (SiGe) heterojunction bipolar transistors (HBTs) have found widespread use in high-frequency (HF) applications, such as communications and automotive radar, due to their co-integration with CMOS. This allows combining high-rate data transfer with digital signal processing on a single chip. So far, the fabrication of SiGeHBTs with cut-off frequencies up to 700 GHz has been demonstrated, and recent simulations predict cut-off frequencies up to 2 THz as physical limit. With such performance, SiGe BiCMOS technology is becoming the enabler for the rapidly emerging field of millimeter-wave and THz electronics with applications in the fields of, e.g., health,security and science. The prediction of the HF performance limit mentioned above was based on a hierarchy of (semi-)classical transport simulation tools. The device optimization resulted in a SiGe base layer thickness of 5 nm, which corresponds to about 36 atom layers. With a peak boron concentration at the solubility limit, thereare statistically just 0.36 doping atoms in a 5 nm stack of <100> lattice unit cells. The impact of carbon in the base, used to prevent boron outdiffusion in fabricated HBTs, was taken into account only phenomenologically. Also, the random arrangement of the material composition and doping atoms in such thin layers leads to presentlyunknown statistical fluctuations of the electrical properties. It is obvious that the assumptions made so far in semi-classical simulations are questionable, and a more detailed investigation is required at the atomistic level. Such studies are currently lacking for SiGeC HBTs, in which carrier transport perpendicular (out-of-plane) tothe surface is the dominant mechanism. This research proposal addresses carrier transport and the resulting HF performance in highly scaled SiGeC HBTs with a boron doped base layer, for the first time, by applying atomistic simulation approaches. Since the complete HBT structure cannot be simulated atomistically, a multiscalemodeling approach will be pursued to assess HF characteristics. First, based on atomistic simulations, the transport related material properties in extremely scaled base layers will be determined for a large variation of compositional and doping atomarrangements. Methods will be developed for extracting the material parameters relevant for incorporation into Boltzmann transport simulations and for calibrating classical drift-diffusion transport models. The latter are needed for structural optimization and generating the data needed for compact models, which in turn enable obtaining the realistic HF characteristics of actual HBT structures and circuits. Finally, being able to bridge the gap between material science and electrical engineering, the impact of random atomic arrangement on the electrical characteristics and further scaling of the vertical HBT structure will be explored. The investigations will be supported by measurements of specially fabricated SiGeC HBTs.