Comprehensive Capacitance–Voltage Simulation and Extraction Tool Including Quantum Effects for High- $k$ on SixGe1−x and InxGa1−xAs: Part II—Fits and Extraction From Experimental Data

Capacitance-voltage ( $C$ – $V$ ) measurement and analysis is highly useful for determining important information about MOS gate stacks. Parameters such as the equivalent oxide thickness (EOT), substrate doping density, flatband voltage, fixed oxide charge, density of interface traps ( ${D}_{ \rm {it}}$ ), and effective gate work function can all be extracted from experimental $C$ – $V$ curves. However, to extract these gate-stack parameters accurately, the correct models must be utilized. In Part I, we described the modeling and implementation of a $C$ – $V$ code that can be used for alternative channel semiconductors in conjunction with high- $k$ gate dielectrics and metal gates. Importantly, this new code (CV ACE) includes the effects of nonparabolic bands and quantum capacitance, enabling accurate models to be applied to experimental $C$ – $V$ curves. In this paper, we demonstrate the capabilities of this new code to extract accurate parameters, including EOT and ${D}_{ \rm {it}}$ profiles from experimental high- $k$ on Ge and In_0.53Ga_0.47As gate stacks.

I. Introduction

Capacitance–voltage (–) simulation and extraction tools have been previously developed for Si-based MOS gate-stack analysis including those with various models to account for quantization effects. The most widely used among these is the North Carolina State University (NCSU) CVC Program [1]. This program has been heavily used, since its publication to aid with gate-stack process development by enabling the extraction of equivalent oxide thickness (EOT), doping density, and the flat band voltage. However, CVC was implemented assuming silicon’s band structure only and is therefore inaccurate when used for MOS devices based on any other semiconductor. Additionally, because Si has such a high-quality native oxide with very low interface defect density, CVC was developed without the capability of extracting a detailed distribution as a function of energy.