Anjan Chakravorty

In this part, we propose a step-by-step strategy to model the static thermal coupling factors between the fingers in a silicon based multifinger bipolar transistor structure. First we provide a physics-based formulation to find out the coupling factors in a multifinger structure having no-trench isolation (cij,nt). As a second step, using the value of cij,nt, we propose a formulation to estimate the coupling factor in a multifinger structure having only shallow trench isolations (cij,st). Finally, the coupling factor model for a deep and shallow trench isolated multifinger device (cij,dt) is presented. The proposed modeling technique takes as inputs the dimensions of emitter fingers, shallow and deep trench isolations, their relative locations and the temperature dependent material thermal conductivity. Coupling coefficients obtained from the model are validated against 3D TCAD simulations of multifinger bipolar transistors with and without trench isolations. Geometry scalability of the model is also demonstrated.

This paper investigates alternative topologies of silicon germanium heterojunction bipolar transistors designed and fabricated in the state-of-the-art BiCMOS process from STMicroelectronics for improved safe-operating characteristics. Electrical and thermal behaviors of various structures are analyzed and compared, along with a detailed discussion on drawbacks and advantages. The test structures under study are different in terms of emitter-finger layouts as well as the metal stacks in the back-end-of-line. It is observed that the multifinger transistor structures having nonuniform finger lengths with wider area enclosed by the deep trench and higher metallization stacks yield an improved thermal behavior. Therefore, the safe-operating area of multifinger transistors can be extended without degrading the RF performances.

In this paper, we present extensive characterization of low-frequency noise in advanced silicon germanium heterojunction bipolar transistors. We demonstrate the extraction methodology of base and collector noise spectral densities for a wide range of transistor geometries. In addition to 1/f noise, generation-recombination (G-R) mechanisms are observed at low bias in the base current noise. Their existence is confirmed by Random Telegraph Signal (RTS) noise measurements. 1/f and G-R components are extracted from the base current noise spectra and their bias dependencies are studied. Finally, base current noise spectral densities measured at the same base current density in different geometries are compared to study the individual contribution of 1/f noise from the periphery as well as the intrinsic device. Part II of this paper will discuss the modeling aspects and noise correlation.

In this brief, we propose a step-by-step strategy to accurately estimate the finger temperature in a silicon-based multifinger bipolar transistor structure from conventional measurements. First we extract the nearly zero-power self-heating resistances (Rth,ii (Ta)) and thermal coupling factors (cij (Ta)) at a given ambient temperature. Now, by applying the superposition principle on these variables at nearly zero-power, where the linearity of the heat diffusion equation is preserved, we estimate an effective thermal resistance (Rth,i (Ta)) and the corresponding revised finger temperature Ti (Ta). Finally, the Kirchhoff's transformation on Ti (Ta) yields the true temperature at each finger (Ti (Ta,Pd)). The proposed extraction technique automatically includes the effects of back-end-of-line metal layers and different types of trenches present within the transistor structure. The technique is first validated against 3-D TCAD simulation results of bipolar transistors with different emitter dimensions and then applied on actual measured data obtained from the state-of-the-art multifinger SiGe HBT from STMicroelectronics B5T technology. It is observed that the superposition of raw measured data at around 40 mW power underestimates the true finger temperature by around 10%.

In this paper, we have assessed the RF measurements of SiGe HBTs upto 500 GHz using TCAD simulation for the first time. In order to bring confidence in simulation, the device geometries and doping profiles are captured in the simulation deck. Then all the basic DC and RF properties are calibrated with the measured data for two different geometries. Additionally the simulated unilateral gain and small signal current gain are also brought in agreement with the corresponding measured data at different bias voltages for both the devices. Finally bias and frequency dependent S- parameter measurements are compared with the TCAD simulation and resulting issues are discussed.

Terahertz (THz) silicon-based electronics is undergoing rapid developments. In order to keep this momentum high, an accurate and optimized on-wafer characterization procedure needs to be developed. While evaluating passive elements, the measured s-parameter data can be verified by a direct use of EM simulation tools. However, this verification requires to precisely introduce part of the measurement environment such as the probes, pads, and access lines to accurately predict the impact of calibration and layout for on-wafer measurements. Unfortunately, this procedure is limited to passive elements. Hence, in this work, we propose a new procedure to emulate the measurement of active devices using an electromagnetic SPICE cosimulation. By this method, one can clearly highlight that a measurement artifact that was observed for the transistor measurement can be reproduced. One of the most representative examples of measurement artifact involves the measurement and estimation of ${f}_{\text {MAX}}$ which is not constant over all frequency bands. Also, the measurement is difficult to perform above 40 GHz. This typical problem is now undoubtedly attributed to the probe-to-substrate coupling and probe-to-probe coupling which are strongly dependent on the probe geometry. Finally, this cosimulation procedure evidently underlines the need for an optimized deembedding procedure above 200 GHz.

The undesired behavior of the substrate significantly affects the output impedance of the device; hence degrades circuit performance mainly in the high frequency regime. Therefore, for high-speed and RF circuits, collector-substrate modeling has to be sufficiently accurate. In this paper, an improved collector-substrate equivalent circuit model is proposed. The circuit model elements are physics based and are calculated from technological data. The validity of the equivalent circuit has been verified by on-wafer measurements of an SiGe HBT fabricated in B55 technology up to 330 GHz, the highest frequency reported so far for collector-substrate modeling. The proposed substrate network can be considered as an extension of the latest large-signal HICUM model (L2v2.4).

This paper proposes a novel 3D nanowire (NW) based SiGe HBT for the first time. The overall purpose is to estimate the RF performance of an NW device based on the state-of-the-art B55 technology from STMicroelectronics. The challenges associated with the device fabrication and corresponding solutions are briefed. The proposed NW-HBT array predicts an fMAX(> 900GHz) that is more than twice of that obtainable from the corresponding bulk SiGe HBT.

From the perspectives of characterized data, calibrated TCAD simulations and compact modeling, we present a deeper investigation of the very high frequency behavior of state-of-the-art sub-THz silicon germanium heterojunction bipolar transistors (SiGe HBTs) fabricated with 55-nm BiC-MOS process technology from STMicroelectronics. The TCAD simulation platform is appropriately calibrated with the measurements in order to aid the extraction of a few selected high-frequency (HF) parameters of the state-of-the-art compact model HICUM, which are otherwise difficult to extract from traditionally prepared test-structures. Physics-based strategies of extracting the HF parameters are elaborately presented followed by a sensitivity study to see the effects of the variations of HF parameters on certain frequency-dependent characteristics until 500 GHz. Finally, the deployed HICUM model is evaluated against the measured s-parameters of the investigated SiGe HBT until 500 GHz.

The applicability of a partitioned-charge-based non-quasi-static (NQS) model is investigated using exact solutions for 1-D bipolar transistors. Limitations of the model to accurately predict y21(ω) are overcome by further delaying the partitioned minority charge associated with the collector. A corresponding small-signal time-domain model is derived. The proposed model effectively includes the delays within both the quasi-neutral base and base-collector space-charge regions. Using only two extra nodes over the quasi-static bipolar transistor model HICUM, the NQS model is implemented using Verilog-A. The simplicity of the model yields straightforward parameter extraction techniques. Modeling results show excellent agreement with numerically simulated data. © 2012 IEEE.