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

Abstract

We present here an efficient numerical approach for the concurrent evaluation of the small-change deterministic device parameters and of the relevant Green's functions exploited in the simulation of device small-signal, small-signal largesignal (conversion matrix), stationary and cyclostationary noise, and variability properties of semiconductor devices through the solution of physics-based models based on a partial-differential equation description of charged carrier transport. The proposed technique guarantees a significant advantage in computation time with respect to the currently implemented solutions. The accuracy is the same as for the standard technique.

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

“…Active devices, instead, require TCAD simulations, e.g., through the drift-diffusion model or higher-order non-stationary transport models, solved over a domain scale of a few hundredths of nanometers, with a discretization grid fine enough to include all relevant device features like doping distribution, material layers, and contact properties. Physics-based simulation may resort to general-purpose physical simulators, like Comsol Multiphysics, to more specific device TCAD commercial simulators, like Synopsys Sentaurus [27] or Silvaco Victory Device [28], or, finally, to ad hoc developed codes, like our TCAD simulator [29], which has been used for this work. MMIC thermal analysis is not included in this work, as it features manifold aspects (e.g., the coupling between the TCAD thermal model and circuit-level analysis through self-consistent electro-thermal solutions [30], or the integration with FEM-3D thermal analysis tools like Keysigth PathWave [31] or Cap-Sym SYMMIC [32]), which would fall outside the scope of this paper, but is the object of future developments, as it is gaining an increasingly important role in a wide range of applications.…”

“…Vk can be added to the nominal device ports to describe parametric variations; see Figure 15 The equivalent PIV generators can be calculated by TCAD concurrently with the device S SCM [29] by means of a Green's function approach: localized technological variations inside the device are transferred by the GFs to equivalent port wave variations. The GF approach greatly reduces the computational cost of TCAD variability analysis, since the relevant propagation quantities must be calculated only once for the nominal parameters γ 0 , thus avoiding repeated simulations.…”

“…The GF approach greatly reduces the computational cost of TCAD variability analysis, since the relevant propagation quantities must be calculated only once for the nominal parameters γ 0 , thus avoiding repeated simulations. Furthermore, the same analysis allows us to calculate the linearized electrical model through the Xpars, also requiring the same S SCM [29]. The resulting model is therefore a doubly linear model, both in terms of port waves and in terms of parameter variations.…”

“…Nonetheless, the efficient calculation of GFs in TCAD is not at all conventional. Up to now, GF-based variability analysis within the harmonic balance approach (i.e., for nonlinear large signal applications) is unique to our in-house software [29], and it is not available in any commercial simulator (indeed Synopsys Sentaurus [27] does implement a GF-based variability analysis, but limited to the DC case only). As a conclusion, a careful choice between case (1NL) and (2NL) must be considered for each specific technology.…”

Multi-Gate FinFET Mixer Variability Assessment Through Physics-Based Simulation

“…In this work, we demonstrate how accurate physics-based simulations can be applied to the numerically efficient variability analysis of a new promising mixer topology based on IG FinFETs [9], [11], linking its RF performance in terms of conversion gain with the corresponding process variations. We exploit a recently developed technique, based on a Green's function approach, allowing for concurrent physics-based DC, AC and sensitivity analysis of electron devices [16]. We show that, due to the peculiar structure of the mixer, the Local Oscillator (LO) and RF bias can be adjusted to achieve maximum conversion gain and virtually null sensitivity to parameter variations.…”