Very High Speed SiGe Heterojunction Bipolar Transistor Reliability Overview

Abstract

We discuss the major reliability mechanisms and the implications arising from the structural changes required for the implementation of state of the art SiGe HBT's. The current gain shift under forward and reverse device operating conditions has been characterized for 120, 200 and 350 GHz transistors. The effects of electromigration and self-heating versus fT have also been studied. In general, it should be possible to continue scaling beyond today's 350 GHz devices, with straightforward extension of the mechanisms and concerns outlined in this paper.

Physical and Electrical Performance Limits of High-Speed Si GeC HBTs—Part II: Lateral Scaling

“…right axis). This appears reasonable since, according to [9] for constant current scaling, 1 one expects in general only a marginal increase in device temperature compared with, e.g., a 350-GHz process with ∆T ≈ 16 • C at peak f T . Compared with the isothermal case, self-heating causes the peak for both f T and f max as well as the corresponding current density to decrease by about 7%, mainly due to the mobility and carrier velocity reduction with temperature in the collector-base region.…”

“…The current density at peaks f T and f max can exceed 100 mA/µm 2 , which causes reliability concerns for the E and C metallization as well as possibly the poly-and monosilicon emitter contact region. Reducing both emitter width and length is typically attempted to maintain the same EM limits between process generations by keeping the ratio of the metal cross-sectional area A M to the maximum emitter window area A E0,max constant [9]. For the discussion below, a slot via and a first-metal (M1) layer, which are strongly recommended for ultra-HF applications, are assumed as well as W as contact material, with J V,max = (16 .…”