Modeling of non-linear large-signal dynamic effects in SiGe heterojunction bipolar transistors

超高速硅锗(SiGe)异质结双极晶体管(HBT)工艺技术的出现，激发了人们对利用毫米波和太赫兹频谱进行电子应用的兴趣迅速增加。在如此高的频率下进行集成电路设计和优化需要具有几百GHz截止频率的HBT的精确紧凑(即集总)模型。遗憾的是，由于缺乏足够的测量设备，现有紧凑型模型的大信号(LS)行为无法在67 GHz以上进行验证。数值器件模拟(所谓的TCAD)表明，在高频LS开关操作过程中，特别是在进入跨导峰值的大电流区域时，存在显著的紧凑模型误差。这些误差主要是由晶体管内移动电荷分布在纵向和横向上的分布特性(即非准静态(NQS)效应)引起的，而目前的紧凑模型没有充分考虑这一点。在这个项目的第一阶段，已经从TCAD和晶体管理论中获得了对垂直和横向NQS效应的基于物理的理解。基于一维输运和连续性方程的封闭解，为分析电荷模型简化的影响提供了参考。此外，对二维HBT结构的LS操作的研究表明，用集总模式捕获横向NQS效应是有希望的结果。最后，LS测量揭示了到目前为止无法解释的三次谐波失真行为。拟议的第二个项目阶段解决了以下观察到的问题：(I)从一维晶体管理论推导出垂直NQS效应的基于物理的公式，特别强调对高电流区进行更准确的建模。(Ii)将一维晶体管理论扩展到二维晶体管理论，以实现紧凑内部基极阻抗与转移电流的完全一致的定义，并推导出集总晶体管模型中与横向NQS效应相关的电荷分配因子的基于物理的电流依赖公式。(Iii)将模型应用于先进的SiGe HBT结构，并通过混合模式器件模拟和测量验证LS模型，不仅使用负载拉动，而且使用一种新的基于无源输出滤波器的方法来获得高达325 GHz的谐波幅度。拟议的工作细分为明确定义的工作包，还包括设计和制造测试结构及其高达325 GHz的晶片上试验性表征。

The availability of ultra-high-speed silicon-germanium (SiGe) heterojunction bipolar transistor (HBT) process technology has spurred rapidly increasing interest in utilizing the millimeter-wave and THz frequency spectrum for electronic applications. Integrated circuit design and optimization at such high frequencies requires accurate compact (i.e. lumped) models for HBTs with cut-off frequencies of several hundred GHz. Unfortunately, the large-signal (LS) behavior of existing compact models cannot be verified beyond 67 GHz due to the lack of adequate measurement equipment. Numerical device simulation (so-called TCAD) has indicated significant compact model errors during LS switching operation at high frequencies, especially when entering the high-current region where the transconductance peaks. These errors are caused mainly by the distributed character (i.e. non-quasi-static (NQS) effects) of the mobile charge distribution within the transistor in both vertical and lateral dimension, which is not adequately taken into account in present compact models. In the first phase of this project, a physics-based understanding of the NQS effects in vertical and lateral direction has been obtained from TCAD and transistor theory. Based on a closed-form solution of the one-dimensional (1D) transport and continuity equation, a reference for evaluating the impact of simplifications for analytical charge modeling has been established. Furthermore, investigations of the LS operation of two-dimensional (2D) HBT structures show promising results for capturing lateral NQS effects with a lumped mode. Finally, LS measurements revealed so far unexplained third-harmonic distortion behavior. The proposed second project phase addresses the observed problems as follows: (i) Derivation of a physics-based formulation for vertical NQS effects from 1D transistor theory with special emphasis on a more accurate modeling of the high-current region. (ii) Extension of 1D to 2D transistor theory for enabling a fully consistent definition of a compact internal base impedance with the transfer current and the derivation of a physics-based current dependent formulation of the lateral NQS effect related charge partitioning factor in a lumped transistor model. (iii) Model application to advanced SiGe HBT structures and verification of the LS model with both mixed-mode device simulation and measurements using not only load-pull but also a novel passive output filter based approach for obtaining the amplitudes of harmonics up to 325 GHz. The proposed work, subdivided into clearly defined work packages, also includes the design and fabrication of test structures and their experimental on-wafer characterization up to 325 GHz.