Variability analysis of FinFET AC/RF performances through efficient physics-based simulations for the optimization of RF CMOS stages

A nearly insatiable appetite for the latest electronic device enables the electronic technology sector to maintain research momentum. The necessity for advancement with miniaturization of electronic devices is the need of the day. Aggressive downscaling of electronic devices face some fundamental limits and thus, buoy up the change in device geometry. MOSFETs have been the leading contender in the electronics industry for years, but the dire need for miniaturization is forcing MOSFET to be scaled to nano-scale and in sub-50 nm scale. Short channel effects (SCE) become dominant and adversely affect the performance of the MOSFET. So, the need for a novel structure was felt to suppress SCE to an acceptable level. Among the proposed devices, FinFETs (Fin Field Effect Transistors) were found to be most effective to counter-act SCE in electronic devices. Today, many industries are working on electronic circuits with FinFETs as their primary element.One of limitation which FinFET faces is device variability. The purpose of this work was to study the effect that different sources of parameter fluctuations have on the behavior and characteristics of FinFETs. With deep literature review, we have gained insight into key sources of variability. Different sources of variations, like random dopant fluctuation, line edge roughness, fin variations, workfunction variations, oxide thickness variation, and source/drain doping variations, were studied and their impact on the performance of the device was studied as well. The adverse effect of these variations fosters the great amount of research towards variability modeling. A proper modeling of these variations is required to address the device performance metric before the fabrication of any new generation of the device on the commercial scale. The conventional methods to address the characteristics of a device under variability are Monte-Carlo-like techniques. In Monte Carlo analysis, all process parameters can be varied individually or simultaneously in a more realistic approach. The Monte Carlo algorithm takes a random value within the range of each process parameter and performs circuit simulations repeatedly. The statistical characteristics are estimated from the responses. This technique is accurate but requires high computational resources and time. Thus, efforts are being put by different research groups to find alternative tools. If the variations are small, Green’s Function (GF) approach can be seen as a breakthrough methodology. One of the most open research fields regards "Variability of FinFET AC performances". One reason for the limited AC variability investigations is the lack of commercially available efficient simulation tools, especially those based on accurate physics-based analysis: in fact, the only way to perform AC variability analysis through commercial TCAD tools like Synopsys Sentaurus is through the so-called Monte Carlo approach, that when variations are deterministic, is more properly referred to as incremental analysis, i.e., repeated solutions of the device model with varying physical parameters. For each selected parameter, the model must be solved first in DC operating condition (working point, WP) and then linearized around the WP, hence increasing severely the simulation time. In this work, instead, we used GF approach, using our in-house Simulator "POLITO", to perform AC variability analysis, provided that variations are small, alleviating the requirement of double linearization and reducing the simulation time significantly with a slight trade-off in accuracy. Using this tool we have, for the first time addressed the dependency of FinFET AC parameters on the most relevant process variations, opening the way to its application to RF circuits. This work is ultimately dedicated to the successful implementation of RF stages in commercial applications by incorporating variability effects and controlling the degradation of AC parameters due to variability. We exploited the POLITO (in-house simulator) limited to 2D structures, but this work can be extended to the variability analysis of 3D FinFET structure. Also variability analysis of III-V Group structures can be addressed. There is also potentiality to carry out the sensitivity analysis for the other source of variations, e.g., thermal variations.