Physics-based modeling of FinFET RF variability

Abstract

This paper presents the physics-based variability analysis of multi-fin double-gate (DG) MOSFETs, representing the core structure of FinFETs for RF applications. The variability of the AC parameters as a function of relevant geometrical and physical parameters, such as the fin width, the fin separation, the source (drain)-gate distance and the doping level is investigated. The analysis exploits a numerically efficient Green's Function technique [1]-[2], extending to the RF case the linearized approach well known from DC variability analysis. The variability of a single fin DG-MOS transistor is compared to a more realistic structure with two fins and raised source/drain contacts, i.e. including both the active part of the FinFET and a significant amount of passive (parasitic) components at the device level. Although presently implemented in a 2D in-house software, the technique can be easily exported to standard 3D TCAD tools, e.g. for tri-gate FinFETs analysis.

Variability of FinFET AC parameters: A physics‐based insight

“…The overall simulation time for the FinFET (∼ 5700 nodes) analysis, including 8 bias points and WF, DOP and LDE sampled over 11 values, is approximately 1 hour for a single frequency on a PC with 8 GB RAM and 2.9 GHz processor. As reported in Bughio et al, 7 for each bias point, the GF approach allows for a saving in simulation time around 5% to 10% with respect to the INC one. above threshold.…”

“…This technique allows for the evaluation of the sensitivity of the AC admittance Y matrix with negligible numerical effort with respect to the ordinary device simulation time. While in Donati Guerrieri et al only test case structures are reported, in Bughio et al this technique was applied to double‐gate (DG) devices (FinFETs), but limited to the validation of the new technique against Monte Carlo simulations for a specific application (single bias point) and with emphasis on the comparison of single and multifin structures. In this paper, instead, we present a comprehensive analysis of the RF variability of realistic DG devices, focused on the role of parasitics.…”

Concurrent Efficient Evaluation of Small-Change Parameters and Green’s Functions for TCAD Device Noise and Variability Analysis

“…As a first case we consider the SS admittance matrix of a double gate 54 nm gate length Si FinFET device (the structure shown in [10]), with 1 nm equivalent gate oxide, Si fin (10 nm width and 10 15 cm −3 bulk acceptor doping) and source/drain extensions of 54 nm length and 5 × 10 18 cm −3 donor doping. Figs.…”

Bridging the Gap between Physical and Circuit Analysis for Variability-Aware Microwave Design: Modeling Approaches

“…According to the sideband analysis, a sideband conversion matrix (SCM) describes frequency conversion among sidebands in terms of a matrix product, i.e., a linear superposition [45]. Sideband conversion analysis was introduced within the framework of TCAD simulations in [35], and was later extended to device variability and sensitivity analyses [38,39,46] and to assist the PIV-aware microwave circuit design [47][48][49]. In our in-house TCAD simulator, the admittance SCM Y is calculated with short-circuited device ports and converted into the scattering SCM S by…”