Advanced Calibration for Device

Simulation User Guide

Version V-2024.03, March 2024
Copyright and Proprietary Information Notice
© 2024 Synopsys, Inc. This Synopsys software and all associated documentation are proprietary to Synopsys, Inc. and
may only be used pursuant to the terms and conditions of a written license agreement with Synopsys, Inc. All other
use, reproduction, modification, or distribution of the Synopsys software or the associated documentation is strictly
prohibited.
Destination Control Statement
All technical data contained in this publication is subject to the export control laws of the United States of America.
Disclosure to nationals of other countries contrary to United States law is prohibited. It is the reader’s responsibility to
determine the applicable regulations and to comply with them.
Disclaimer
SYNOPSYS, INC., AND ITS LICENSORS MAKE NO WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, WITH
REGARD TO THIS MATERIAL, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
Trademarks
Synopsys and certain Synopsys product names are trademarks of Synopsys, as set forth at
https://www.synopsys.com/company/legal/trademarks-brands.html.
All other product or company names may be trademarks of their respective owners.
Free and Open-Source Licensing Notices
If applicable, Free and Open-Source Software (FOSS) licensing notices are available in the product installation.
Third-Party Links
Any links to third-party websites included in this document are for your convenience only. Synopsys does not endorse
and is not responsible for such websites and their practices, including privacy practices, availability, and content.

www.synopsys.com

Advanced Calibration for Device Simulation User Guide 2

V-2024.03
Contents

Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Customer Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Statement on Inclusivity and Diversity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1. Introduction to Advanced Calibration Device . . . . . . . . . . . . . . . . . . . . . . . 10

Overview of Advanced Calibration Device . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Location of Advanced Calibration Device Files . . . . . . . . . . . . . . . . . . . . 10
Target Devices and Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Content of Parameter Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Special Parameter Files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Using Advanced Calibration Device. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2. Guide to Device Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Specifying Simulation Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Accessing Parameter Files and Model Frameworks . . . . . . . . . . . . . . . . . . . . 24
Path A: Basic CMOS Simulation Command File Sections . . . . . . . . . . . . 24
Path B: Add Influence of Mechanical Stress to Physics Section Shown in Path A
25
Path C: Add High-k Mobility Degradation to Physics Section Shown in Path B 27
Path D: Thin-Layer MOS Simulations Including Thin-Film Effects . . . . . . 29
Path E: FinFET Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Path F: Smart-Power HVMOS and LDMOS Devices. . . . . . . . . . . . . . . . 32
Path G: GaN HEMT Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Path H: SiC Diode Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Path I: InGaAs FinFET Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Path J: SiC HVMOS Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

3. Calibration Methodology for Device Simulations . . . . . . . . . . . . . . . . . . . . 38

Calibration of CMOS Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

3
Contents

C–V Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Id–Vg Mobility Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Calibration of Parasitic Series Resistance . . . . . . . . . . . . . . . . . . . . . . . . 42
Id–Vg Mobility Calibration at Low-Drain Bias Voltage. . . . . . . . . . . . . . . . 43
Id–Vg Mobility Calibration at High-Drain Bias Voltage . . . . . . . . . . . . . . . 44

References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

4. Descriptions of Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Overview of Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
CMOS Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Unstrained Band Structure and Electrostatics . . . . . . . . . . . . . . . . . . . . . 46
Strained Band Structure and Electrostatics . . . . . . . . . . . . . . . . . . . . . . . 46
Quantization and Band Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Unstrained Long-Channel Low-Field Mobility . . . . . . . . . . . . . . . . . . . . . 47
Strained Low-Field Mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Influence of Mechanical Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
First-Order (Linear) Piezoresistance Mobility Model . . . . . . . . . . . . 52
Second-Order Piezoresistance Mobility Model . . . . . . . . . . . . . . . . 53
Nonlinear Piezoresistance Models for Electrons and Holes (MCmob and
SBmob). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Intel Stress-Induced Hole Mobility Model (hSixBand) . . . . . . . . . . . 54
eSubband Model for Electrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
hSubband Model for Holes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Unstrained High-Field Mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Strained High-Field Mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Ballistic Mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Gallium Nitride HEMTs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Silicon Carbide Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Silicon Smart-Power Devices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

5. Quality of Fitting and Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

Low-Field Mobility of Planar CMOS Devices With Silicon Channel . . . . . . . . . 59
Oxide–Silicon Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Influence of Metal Gate on Inversion Layer Mobility . . . . . . . . . . . . . . . . 66
Influence of High-k Gate Stack on Inversion Layer Mobility. . . . . . . . . . . 67
Extraction of Contact Resistance Parameters . . . . . . . . . . . . . . . . . . . . . 68
Extraction of Parameters for Mobility Models in Accumulation Regime . . 69

4
Contents

Basic Properties of Silicon Carbide Devices . . . . . . . . . . . . . . . . . . . . . . . . . . 70

Basic Properties of Gallium Nitride Devices . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Basic Properties of Indium Gallium Arsenide Devices . . . . . . . . . . . . . . . . . . . 79
Permittivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Lattice Heat Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Thermal Conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Band Gap and Electron Affinity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Density-of-States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Bandgap Narrowing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Quantization Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Low-Field Bulk Mobility and Its Doping Dependency . . . . . . . . . . . . . . . . 87
Philips Unified Mobility Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Inversion Layer Mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
Inversion and Accumulation Layer Mobility Model . . . . . . . . . . . . . . 90
Interface Charge Mobility Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Strained Low-Field Mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Multivalley Electron Mobility Model . . . . . . . . . . . . . . . . . . . . . . . . . 93
Piezoresistance Mobility Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Basic Properties of Silicon Germanium Devices . . . . . . . . . . . . . . . . . . . . . . . 94
Permittivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Thermal Conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Band Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Bulk Low-Field Mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
Ballistic Mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

A. Physical Model Interface Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

Features of the MCmob and SBmob Models. . . . . . . . . . . . . . . . . . . . . . . . . . 106
Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Specifying Device Coordinate Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Specifying Local Coordinate Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Using Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Parameter Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Parameter Interface and Equations of SBmob Model . . . . . . . . . . . . . . . 115

PMI for Calculating the Sheet Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

5
About This Guide
The Synopsys® Sentaurus™ Device tool is a quantum drift-diffusion device simulator that
solves semiconductor equations in one, two, or three spatial dimensions.
Extensions for hydrodynamic transport, solving the heat conduction equation, and many
other features are also available. Because of the many models and corresponding model
parameters and their dependency on fabrication processes and device types, it is difficult to
derive automatically a well-adjusted device simulation environment for specific simulation
tasks. Therefore, preselection of models and precalibration of parameter sets are
necessary. Advanced Calibration Device refers to the preselection of models and the
precalibration of parameter sets.
This user guide describes the contents and use of the Advanced Calibration Device files and
is designed to give users fast access to parameter sets and model selections needed for
device simulation. This user guide is intended for users who are familiar with Sentaurus
Device and want to achieve a higher accuracy in device simulation. See the Sentaurus™
Device User Guide.
For additional information, see:
• The TCAD Sentaurus release notes, available on the Synopsys SolvNetPlus site (see
Accessing SolvNetPlus on page 8)
• Documentation available on the SolvNetPlus site

Conventions
The following conventions are used in Synopsys documentation.

Convention Description

Courier font Identifies text that is displayed on the screen or that the user must enter. It
identifies the names of files, directories, paths, parameters, keywords, and
variables.

Italicized text Used for emphasis, the titles of books and journals, and non-English words. It
also identifies components of an equation or a formula, a placeholder, or an
identifier.

Advanced Calibration for Device Simulation User Guide 7

V-2024.03
About This Guide
Customer Support

Customer Support
Customer support is available through the SolvNetPlus site.

Accessing SolvNetPlus
The SolvNetPlus site includes a knowledge base of technical articles and answers to
frequently asked questions about Synopsys tools. The site also gives you access to a wide
range of Synopsys online services including software downloads, documentation, and
technical support.
To access the SolvNetPlus site:
1. Go to https://solvnetplus.synopsys.com.
2. Enter your user name and password. (If you do not have a Synopsys user name and
password, follow the instructions to register.)

Contacting Synopsys Support

If you have problems, questions, or suggestions, you can contact Synopsys support in the
following ways:
• Go to the Synopsys Global Support Centers site on www.synopsys.com to find email
addresses and telephone numbers for Synopsys support centers throughout the world.
• Go to the SolvNetPlus site and open a case (Synopsys user name and password
required).

Contacting Your Local TCAD Support Team Directly

Send an email message to:
• [email protected] from within North America and South America
• [email protected] from within Europe
• [email protected] from within Asia Pacific (China, Taiwan, Singapore,
Malaysia, India, Australia)
• [email protected] from Korea

• [email protected] from Japan

Advanced Calibration for Device Simulation User Guide 8

V-2024.03
About This Guide
Statement on Inclusivity and Diversity

Statement on Inclusivity and Diversity

Synopsys is committed to creating an inclusive environment where every employee,
customer, and partner feels welcomed. We are reviewing and removing exclusionary
language from our products and supporting customer-facing collateral. Our effort also
includes internal initiatives to remove biased language from our engineering and working
environment, including terms that are embedded in our software and IPs. At the same time,
we are working to ensure that our web content and software applications are usable to
people of varying abilities. You may still find examples of non-inclusive language in our
software or documentation as our IPs implement industry-standard specifications that are
currently under review to remove exclusionary language.

Advanced Calibration for Device Simulation User Guide 9

V-2024.03
1
Introduction to Advanced Calibration Device
1

This chapter introduces the use of Advanced Calibration Device in Sentaurus Device
simulations.

Overview of Advanced Calibration Device

Advanced Calibration Device is a set of models and parameters that is recommended for
device simulations of certain device types and technologies. This set of models and
parameters is contained in text files, which standard text editors can open.
You select the standard Synopsys calibration by starting Sentaurus Device with the
corresponding Advanced Calibration Device command file section and by loading the
Advanced Calibration Device parameter file from that command file.
For more information about how to use the command file that defines models and the
parameter file that defines the parameters of the models selected in the command file, see
the Sentaurus™ Device User Guide.
This user guide describes the parameter file and command file sections for Advanced
Calibration Device.

Location of Advanced Calibration Device Files

Each release of TCAD Sentaurus provides new Advanced Calibration Device files that
include the latest set of models and parameters. The MaterialDB folder of the standard
installation of Sentaurus Device stores the parameter files and is located at:
$STROOT/tcad/$STRELEASE/lib/sdevice/MaterialDB

The STROOT environment variable indicates where the Synopsys TCAD distribution has
been installed.
See Table 1 on page 13 for a list of the available material-related parameter files.

Advanced Calibration for Device Simulation User Guide 10

V-2024.03
Chapter 1: Introduction to Advanced Calibration Device
Content of Parameter Files

Target Devices and Technologies

Advanced Calibration Device provides a set of parameters for CMOS technologies from
micrometer dimensions down to the 5 nm node, including planar gate-first and gate-last
technologies, nonplanar technologies, and technologies based on novel channel materials.
In addition, Advanced Calibration Device provides parameters and models for silicon
smart-power and power devices (IGBT, LDMOS, NMOS, and PMOS), GaN HEMTs, InGaAs
channel devices, SiC diodes, and SiC FETs.
The main process features of planar and nonplanar bulk and SOI technology nodes covered
by the model and parameter sets are:
• Stress and crystallographic orientation engineering
• Poly/SiON or SiO2 gate stacks and metal/HfO2/SiON or SiO2 gate stacks
• Gate-first and gate-last technologies
• Silicon, germanium, SiGe, GaN, SiC, or InGaAs channels
For these devices and technologies, Advanced Calibration Device provides model
selections and parameter sets for the simulation of the following device properties:
• Band structure, electrostatic properties, and quantization
• Low-field, high-field, and ballistic transport
• Stress and mole fraction dependency of mobility
Note:
The parameter sets must be considered as initial parameter sets, that is, starting
points for further calibration.

Content of Parameter Files

The information presented here relates to the models presented in Chapter 2 on page 23
and to the files stored in the MaterialDB folder of Sentaurus Device.
Table 1 on page 13 lists the content and features of the material-related parameter files.
References and additional comments for the parameters can be found in the parameter
files.

Advanced Calibration for Device Simulation User Guide 11

V-2024.03
Chapter 1: Introduction to Advanced Calibration Device
Content of Parameter Files

When simulating silicon or SiGe CMOS, or silicon smart-power or power devices, you must
load one or more of the following files:
• Silicon.par
• Germanium.par
• SiliconGermanium.par
• Siliconc100.par
• Siliconc110.par
• SiliconGermaniumc100.par
• SiliconGermaniumc110.par

When simulating wide-bandgap devices, depending on the material, you must load one or
more of the following files:
• AlGaN.par, AlInGaN.par, AlInN.par, AlN.par
• Diamond.par

• Ga2O3.par
• GaN.par

• InGaN.par, InN.par

When simulating III–V arsenide devices, you must load one or more of the following files:
• AlAs.par
• AlGaAs.par
• GaAs.par
• InAs.par
• InGaAs.par

When simulating SiC devices, you must load one of the following files:
• SiliconCarbide.par
• 4H-SiC.par/4HSiC.par/SiC_4H.par
• 6H-SiC.par

Advanced Calibration for Device Simulation User Guide 12

V-2024.03
Chapter 1: Introduction to Advanced Calibration Device
Content of Parameter Files

Table 1 Parameter files in MaterialDB folder of Sentaurus Device

Parameter file Material Parameters

AlAs.par AlAs • Permittivity

• Heat capacitance
• Thermal conductivity
• Band gap and affinity
• Density-of-states
• Parameters in Multivalley section
• Optical parameters
Other model parameter sections are available in
the parameter file but are not yet updated and
reviewed.

AlGaAs.par AlGaAs • Permittivity

• Heat capacitance
• Thermal conductivity
• Band gap and affinity
• Density-of-states
• Parameters in Multivalley section
• Optical parameters
Other model parameter sections are available in
the parameter file but are not yet updated and
reviewed.

Advanced Calibration for Device Simulation User Guide 13

V-2024.03
Chapter 1: Introduction to Advanced Calibration Device
Content of Parameter Files

Table 1 Parameter files in MaterialDB folder of Sentaurus Device (Continued)

Parameter file Material Parameters

AlGaN.par AlGaN • For most models, mole fraction dependency

is interpolated linearly between the corner
materials GaN and AlN
• U-shape of thermal conductivity is described
by a piecewise linear spline function in the
parameter file
• Mole fraction–dependent band gap
(parabolic interpolation)
• Mole fraction–dependent mass is
interpolated linearly between GaN and AlN
for density-of-states calculation
• Mole fraction–dependent mobility with doping
dependency is described by a piecewise
linear spline function in the parameter file
• Mole fraction–dependent high field
dependency is described by a piecewise
linear spline function in the parameter file
• Mole fraction–dependent impact ionization is
described by a piecewise linear spline
function in the parameter file

AlInGaN.par AlInGaN • For most models, mole fraction dependency

is interpolated linearly between the corner
materials AlN, GaN, and InN

AlInN.par AlInN • For most models, mole fraction dependency

is interpolated linearly between the corner
materials AlN and InN

Advanced Calibration for Device Simulation User Guide 14

V-2024.03
Chapter 1: Introduction to Advanced Calibration Device
Content of Parameter Files

Table 1 Parameter files in MaterialDB folder of Sentaurus Device (Continued)

Parameter file Material Parameters

AlN.par AlN • Isotropic and anisotropic permittivity

• Thermal conductivity
• Lattice heat capacity
• Band gap and affinity
• Density-of-states and masses
• Mobility with doping dependency
• High field dependency
• Shockley–Read–Hall
• Auger
• Impact ionization
• Piezoelectric polarization
• Optical material properties
• Mechanical compliance parameters

CobaltSilicide.par CoSi/CoSi2 • Permittivity

• Thermal conductivity
• Lattice heat capacity
• Resistivity
• Optical material parameters

Diamond.par Diamond • Isotropic and anisotropic permittivity

• Thermal conductivity
• Incomplete ionization
• Band gap and affinity
• Density-of-states
• Constant mobility
• High field dependency
• Impact ionization

Ga2O3.par Ga2O3 • Permittivity

• Heat capacitance
• Thermal conductivity
• Band structure (band gap and affinity)
• Density-of-states for electrons
• High-field saturation
• Impact ionization

Advanced Calibration for Device Simulation User Guide 15

V-2024.03
Chapter 1: Introduction to Advanced Calibration Device
Content of Parameter Files

Table 1 Parameter files in MaterialDB folder of Sentaurus Device (Continued)

Parameter file Material Parameters

GaAs.par GaAs • Permittivity

• Heat capacitance
• Thermal conductivity
• Band structure (band gap, affinity, and
parameters for the Multivalley section)
• Bandgap narrowing
• Density-of-states
• Bulk and inversion mobility for different
surface and channel orientations
• Quantization parameters
• Optical properties
Other model parameter sections are available in
the parameter file but are not yet updated and
reviewed.

GaAsSb.par GaAsSb • Permittivity

• Heat capacitance
• Thermal conductivity
• Band gap and affinity
• Density-of-states
• Parameters in Multivalley section

GaInSb.par GaInSb • Permittivity

• Heat capacitance
• Thermal conductivity
• Band gap and affinity
• Density-of-states
• Parameters in Multivalley section

Advanced Calibration for Device Simulation User Guide 16

V-2024.03
Chapter 1: Introduction to Advanced Calibration Device
Content of Parameter Files

Table 1 Parameter files in MaterialDB folder of Sentaurus Device (Continued)

Parameter file Material Parameters

GaN.par GaN • Isotropic and anisotropic permittivity

• Thermal conductivity
• Lattice heat capacity
• Band gap and affinity
• Density-of-states and masses
• Bulk mobility with doping dependence
• Inversion layer mobility (IALMob)
• High field dependency
• Shockley–Read–Hall
• Auger
• Radiative recombination
• Impact ionization
• Piezoelectric polarization
• Optical material properties
• Mechanical compliance parameters

GaSb.par GaSb • Permittivity

• Heat capacitance
• Thermal conductivity
• Band gap and affinity
• Density-of-states
• Parameters in Multivalley section

Germanium.par Germanium • Permittivity

• Heat capacitance
• Thermal conductivity
• Band gap and bandgap narrowing
• Density-of-states
• Bulk mobility
• High field dependency
• Optical properties
Other model parameter sections are available in
the parameter file but are not yet updated and
reviewed.

Advanced Calibration for Device Simulation User Guide 17

V-2024.03
Chapter 1: Introduction to Advanced Calibration Device
Content of Parameter Files

Table 1 Parameter files in MaterialDB folder of Sentaurus Device (Continued)

Parameter file Material Parameters

InAs.par InAs • Permittivity

• Heat capacitance
• Thermal conductivity
• Band structure (band gap, affinity, and
parameters for the Multivalley section)
• Bandgap narrowing
• Density-of-states
• Bulk and inversion mobility for different
surface and channel orientations
• Quantization parameters
• Optical properties
Other model parameter sections are available in
the parameter file but are not yet updated and
reviewed.

InGaAs.par InGaAs • Permittivity

• Heat capacitance
• Thermal conductivity
• Band structure (band gap, affinity, and
parameters for the Multivalley section)
• Bandgap narrowing
• Density-of-states
• Bulk and inversion mobility for different
surface and channel orientations
• Quantization parameters
• Optical properties
Other model parameter sections are available in
the parameter file but are not yet updated and
reviewed.

InGaN.par InGaN • For most models, mole fraction dependency

is interpolated linearly between the corner
materials GaN and InN

Advanced Calibration for Device Simulation User Guide 18

V-2024.03
Chapter 1: Introduction to Advanced Calibration Device
Content of Parameter Files

Table 1 Parameter files in MaterialDB folder of Sentaurus Device (Continued)

Parameter file Material Parameters

InN.par InN • Isotropic and anisotropic permittivity

• Heat capacitance
• Thermal conductivity
• Band gap and affinity
• Density-of-states
• Mobility with doping dependency
• High field dependency
• Impact ionization
• Piezoelectric polarization
• Mechanical compliance parameters

InSb.par InSb • Permittivity

• Heat capacitance
• Thermal conductivity
• Band gap and affinity
• Density-of-states
• Parameters in Multivalley section

Silicon.par Si • Permittivity
Siliconc100.par • Heat capacitance
Siliconc110.par
• Thermal conductivity
• Band gap and bandgap narrowing
• Density-of-states
• Quantization
• Bulk and inversion mobility for different
surface and channel orientations, and film
thicknesses
• High field dependency
Other model parameter sections are available in
the parameter file but are not yet updated and
reviewed.

Advanced Calibration for Device Simulation User Guide 19

V-2024.03
Chapter 1: Introduction to Advanced Calibration Device
Content of Parameter Files

Table 1 Parameter files in MaterialDB folder of Sentaurus Device (Continued)

Parameter file Material Parameters

SiliconCarbide.par SiC • Isotropic and anisotropic permittivity

4H-SiC.par/4HSiC.par/ • Isotropic and anisotropic thermal conductivity
SiC_4H.par • Heat capacitance
6H-SiC.par • Band gap and bandgap narrowing
• Density-of-states
• Bulk mobility with doping dependence
• Inversion layer mobility (Enormal)
• Inversion layer mobility (IALMob)
• Anisotropic mobility
• Isotropic and anisotropic high field
dependency
• Shockley–Read–Hall
• Auger
• Isotropic and anisotropic impact ionization
• Incomplete ionization

SiliconGermanium.par SiGe (0–100%) • All models are mole fraction dependent

SiliconGermaniumc100.par can be used for • Permittivity
SiliconGermaniumc110.par Ge as well
• Heat capacitance
• Thermal conductivity
• Band gap and bandgap narrowing
• Density-of-states
• Quantization
• Bulk and inversion mobility for different
surface and channel orientations, and film
thicknesses
• High field dependency
• Auger
Other model parameter sections are available in
the parameter file but are not yet updated and
reviewed.

Advanced Calibration for Device Simulation User Guide 20

V-2024.03
Chapter 1: Introduction to Advanced Calibration Device
Special Parameter Files

Special Parameter Files

The MaterialDB/custom directory of Sentaurus Device contains the following special
parameter files that reflect certain application cases:
• GaAs_DEEPEN.par, InAs_DEEPEN.par, and InGaAs_DEEPEN.par contain parameters in
the Bandgap and Multivalley sections extracted from hybrid functional calculations
within the scope of the European Union DEEPEN project.
• For AlN_DEEPEN.par, GaN_DEEPEN.par, and InN_DEEPEN.par, the parameters in the
Piezoelectric_Polarization section are included, based on the results of hybrid
functional calculations. In addition, the bowing parameters are updated in the Bandgap
section for the AlGaN_DEEPEN.par, AlInN_DEEPEN.par, and InGaN_DEEPEN.par files.
• GatePolySilicon.par is for the polysilicon gate of FETs.

• mcmob.par contains the default parameter set for the MCmob model (see Appendix A on
page 106).
• Silicon_kvm.par and SiliconGermanium_kvm.par contain calibrated kinetic velocity
model (KVM) parameters for better simulation of ballistic effects in short-channel devices
(see the Sentaurus™ Device User Guide).
• In52Al48As.par, In53Ga47As.par, and In52Al48As_In53Ga47As.par contain
material parameters specifically for In0.52Al0.48As, In0.53Ga0.47As, and
(In0.52Al0.48As)x(In0.53Ga0.47As)1–x alloys, respectively, which are lattice matched to InP
material. To be recognized in Sentaurus Device, these material names can be added
manually to the datexcodes.txt file:
Materials {
In52Al48As {
label = "In52Al48As"
group = Semiconductor
}
In53Ga47As {
label = "In53Ga47As"
group = Semiconductor
}
In52Al48As_In53Ga47As {
label = "In52Al48As_In53Ga47As"
group = Semiconductor
}
}

The mole fraction–dependent material (In0.52Al0.48As)x(In0.53Ga0.47As)1–x must be

defined as well in the Molefraction.txt file:
#(In52Al48As)(x)(In53Ga47As)(1-x)
In52Al48As_In53Ga47As (x=0) = In53Ga47As
In52Al48As_In53Ga47As (x=1) = In52Al48As

Advanced Calibration for Device Simulation User Guide 21

V-2024.03
Chapter 1: Introduction to Advanced Calibration Device
Using Advanced Calibration Device

Using Advanced Calibration Device

To use Advanced Calibration Device, you can look up the parameter files and the command
files that are best suited for your device and technology requirements in Chapter 2 on
page 23.
The European Union projects DEEPEN and III-V-MOS have generated additional parameter
sets that complement the general parameter files of Advanced Calibration Device. The
parameter files from the DEEPEN project are stored in the MaterialDB/custom folder for
AlGaN, AlInN, AlN, GaAs, GaN, InAs, InGaAs, InGaN, and InN (see Special Parameter Files
on page 21).

Advanced Calibration for Device Simulation User Guide 22

V-2024.03
2
Guide to Device Simulations
2

This chapter provides a guide for the quick selection of parameter sets and device models.

Specifying Simulation Cases

Table 2 and Table 3 present typical simulation cases, where a path is defined that you can
use to select the proper device simulation parameter set and the appropriate Physics
section for the command file. In the tables, the characters A to J indicate different paths that
are described in subsequent sections.
Table 2 Specifying the simulation case: CMOS

Node

> 130 nm 32–130 nm 14–32 nm < 14 nm

Selection path A B C C B C C C E D E E I I E E E

Planar bulk and SOI X X X X X X X X

FinFET X X X X

Nanowire X X X X

Thin-layer SOI X

Si X X X X X X X X X

SiGe X X X X

Ge X X

InGaAs X X

Gate-first X X X

Advanced Calibration for Device Simulation User Guide 23

V-2024.03
Chapter 2: Guide to Device Simulations
Accessing Parameter Files and Model Frameworks

Table 2 Specifying the simulation case: CMOS (Continued)

Node

> 130 nm 32–130 nm 14–32 nm < 14 nm

Selection path A B C C B C C C E D E E I I E E E

Gate-first with metal gate X X X X

Gate-last with metal gate X X X X X X X X X X

Without stress X X X X

With stress X X X X X X X X X X X X X

Table 3 Specifying the simulation case: Smart power and power

Device

HVMOS HVMOS LDMOS HEMT Diode IGBT

Selection path F J F G H F

Si X X X

SiC X X

GaN X

Accessing Parameter Files and Model Frameworks

The following sections describe the parameter files and the model frameworks. For
information about handling and loading parameter files, see the Sentaurus™ Device User
Guide.

Path A: Basic CMOS Simulation Command File Sections

Quantization uses the density gradient model of Sentaurus Device (eQuantumPotential
and hQuantumPotential). However, for technologies with gate oxide thicknesses greater
than 3 nm, you can neglect quantum correction without greatly affecting the simulation of the
gate capacitance.

Advanced Calibration for Device Simulation User Guide 24

V-2024.03
Chapter 2: Guide to Device Simulations
Accessing Parameter Files and Model Frameworks

The orientation of the conducting interface is detected automatically. However, you must
define the channel or the current flow direction manually using the correct mobility
parameter file. The corresponding parameter files are located in the MaterialDB folder.
For the <110> channel direction, you must load the Siliconc110.par file. For the <100>
channel direction, you must load the Siliconc100.par file.
Physics {Fermi}
Physics (Material="Silicon") {
eQuantumPotential(AutoOrientation density)
hQuantumPotential(AutoOrientation density)
Mobility (Enormal (IALMob(AutoOrientation)) HighFieldSaturation)
# For PMOS, it is useful to use PhononCombination=2 in the IALMob.
EffectiveIntrinsicDensity(OldSlotboom)
Recombination(SRH)
# Recombination(SRH Band2Band(Model=NonlocalPath))
# For off-current calculation, band-to-band
# tunneling models must be switched on.
}

Path B: Add Influence of Mechanical Stress to Physics Section

Shown in Path A
Taking into account mechanical stress, you must combine the low-field mobility model with
a model describing the stress influence. You can model the mobility change by mechanical
stress in one of the following ways:
• Use the eSubband and hSubband models for a more physics-based calculation of the
mobility enhancement by mechanical stress in silicon devices that is directly based on
band structure and subbands. For the subband models, you must set the channel
direction in the command file when it is not the <110> channel.
• Use the MCmob physical model interface (PMI) model (parameters are extracted from
Sentaurus Device Monte Carlo) or the SBmob PMI model (parameters are extracted from
Sentaurus Band Structure). For the MCmob and SBmob models, you must set the channel
direction in the parameter file (see Appendix A on page 106).
The orientation of the conducting interface is detected automatically. However, you must
define the channel or the current flow direction manually using the correct mobility
parameter file. The corresponding parameter files are located in the MaterialDB folder.
For the <110> channel direction, you must load the Siliconc110.par file. For the <100>
channel direction, you must load the Siliconc100.par file.

Advanced Calibration for Device Simulation User Guide 25

V-2024.03
Chapter 2: Guide to Device Simulations
Accessing Parameter Files and Model Frameworks

With subband models, you must use the following syntax:

Physics {Fermi}
Physics (Material="Silicon") {
eQuantumPotential(AutoOrientation density)
hQuantumPotential(AutoOrientation density)
Mobility (Enormal (IALMob(AutoOrientation)) HighFieldSaturation)
# For the inclusion of ballistic effects, you can use the command
# BalMob(KVM(Full) Frensley Fermi) in the Mobility statement.
# For PMOS, it is useful to use PhononCombination=2 in the IALMob.
EffectiveIntrinsicDensity(OldSlotboom)
Recombination(SRH)
# Recombination(SRH Band2Band(Model=NonlocalPath))
# For off-current calculation, band-to-band
# tunneling models must be switched on.

eMultivalley(MLDA kpDOS -density) # hMultivalley for holes.

# For holes, defining kpDOS and parfile
# together results in double-counting
# for holes when gamma hole valleys are
# defined in the parameter file.

Piezo (
Model (
Mobility (
saturationfactor=0.2
eSubband(Fermi EffectiveMass Scattering(MLDA))
# eSubband(Fermi EffectiveMass Scattering(MLDA) -RelChDir110)
# for <100> channel.
# hSubband(Fermi EffectiveMass Scattering(MLDA)) for holes
# For holes, the replacement of the option Fermi by the
# option Doping can result in speedup of the simulation.
)
DOS(eMass hMass)
DeformationPotential(Minimum ekp hkp)
)
)
}

With the MCmob model or SBmob model, you must use the following syntax:
Physics {Fermi}

Physics (Material="Silicon") {
eQuantumPotential(AutoOrientation density)
hQuantumPotential(AutoOrientation density)
Mobility (Enormal (IALMob(AutoOrientation)) HighFieldSaturation)
# For PMOS, it is useful to use PhononCombination=2 in the IALMob.
EffectiveIntrinsicDensity(OldSlotboom)
Recombination(SRH)
# Recombination(SRH Band2Band(Model=NonlocalPath))
# For off-current calculation, band-to-band
# tunneling models must be switched on.

Advanced Calibration for Device Simulation User Guide 26

V-2024.03
Chapter 2: Guide to Device Simulations
Accessing Parameter Files and Model Frameworks

Piezo (
Model (
Mobility (
saturationfactor=0.2
efactor(Kanda sfactor=SBmob(Type=0)) # Type=1 for holes
#efactor(Kanda sfactor=MCmob(Type=0)) # when using MCmob.
)
DOS(eMass hMass)
DeformationPotential(Minimum ekp hkp)
)
)
}

These command file sections are designed for electron transport (NMOS). For hole
transport (PMOS), replace eSubband with hSubband, eMultivalley with hMultivalley,
and specify Type=1 for MCmob and SBmob.
The mobility of unstrained long-channel devices at low-drain bias voltage does not usually
depend on the channel direction for the (100) surface. Even so, the warping of selected
valence bands results in an orientation-dependent mass; the measurements do not show a
dependence on the channel direction for very long channels. This might be a result of the
many scattering processes and randomizations of the k-direction in long-channel devices.
For short-channel devices, the situation is completely different. First, the saturation velocity
depends on the channel direction for higher fields. Second, the ballistic transport depends
on the channel direction. Third, the warping of the valence band structure becomes
important because there are fewer scattering effects.

Path C: Add High-k Mobility Degradation to Physics Section Shown

in Path B
The orientation of the conducting interface is detected automatically. However, you must
define the channel or the current flow direction manually using the correct mobility
parameter file. The corresponding parameter files are located in the MaterialDB folder.
You must load the following parameter files for the corresponding channel direction:
• For the silicon <100> channel direction, load the Siliconc100.par file.
• For the silicon <110> channel direction, load the Siliconc110.par file.
• For the SiGe <100> channel direction, load the SiliconGermaniumc100.par file.
• For the SiGe <110> channel direction, load the SiliconGermaniumc110.par file.

Advanced Calibration for Device Simulation User Guide 27

V-2024.03
Chapter 2: Guide to Device Simulations
Accessing Parameter Files and Model Frameworks

Physics {Fermi}

Physics (Material="Silicon") {
#or Physics (Material="SiliconGermanium") {
eQuantumPotential(AutoOrientation density)
hQuantumPotential(AutoOrientation density)
Mobility (
Enormal (
IALMob(AutoOrientation)
# For PMOS, it is useful to use PhononCombination=2 in IALMob.
RPS # Used for remote phonon scattering (RPS).
NegInterfaceCharge (SurfaceName="s1")
# Used for remote Coulomb scattering (RCS)
# and remote dipole scattering (RDS).
PosInterfaceCharge (SurfaceName="s1")
# Used for RCS and RDS.
)
HighFieldSaturation) # To include ballistic effects, you can
# use the BalMob(KVM(Full) Frensley Fermi)
# command in the Mobility statement.
EffectiveIntrinsicDensity(OldSlotboom)
# For SiGe with high Ge mole fraction or for pure
# Ge, it is recommended to switch off the Fermi
# correction and to use
# EffectiveIntrinsicDensity(OldSlotboom NoFermi)
Recombination(SRH)
# Recombination(SRH Band2Band) # For off-current calculation,
# band-to-band tunneling models
# must be switched on.

eMultivalley(MLDA kpDOS -density) # hMultivalley for holes.

# For SiGe, use eMultivalley(MLDA kpDOS parfile -density)

Piezo (
Model (
Mobility (
saturationfactor=0.2
eSubband(Fermi EffectiveMass Scattering(MLDA))
# eSubband(Fermi EffectiveMass Scattering(MLDA) -RelChDir110)
# for <100> channel.
# hSubband(Fermi EffectiveMass Scattering(MLDA)) for holes
# For holes, the replacement of the option Fermi by the
# option Doping can result in speedup of the simulation.
)
DOS(eMass hMass)
DeformationPotential(Minimum ekp hkp)
)
)
}

Advanced Calibration for Device Simulation User Guide 28

V-2024.03
Chapter 2: Guide to Device Simulations
Accessing Parameter Files and Model Frameworks

The surface name specifies the surface or interface that causes the mobility degradation, for
example, the HfO2–oxide interface in the high-k gate stack of a MOS transistor. The surface
must be defined in the Math section of the command file. For more information about high-k
mobility degradation models, see Unstrained Long-Channel Low-Field Mobility on page 47.
The mobility of unstrained long-channel devices at low-drain bias voltage does not usually
depend on the channel direction for the (100) surface. Even so, the warping of selected
valence bands results in an orientation-dependent mass; the measurements do not show a
dependence on the channel direction for very long channels. This might be a result of the
many scattering processes and randomizations of the k-direction in long-channel devices.
For short-channel devices, the situation is completely different. First, the saturation velocity
depends on the channel direction for higher fields. Second, the ballistic transport depends
on the channel direction. Third, the warping of the valence band structure becomes
important because there are fewer scattering effects.

Path D: Thin-Layer MOS Simulations Including Thin-Film Effects

The Mobility command in the Physics section of Path C must be replaced by:
Mobility (
ThinLayer (IALMob(AutoOrientation))
# For PMOS, it is useful to use PhononCombination=2 in IALMob.
Enormal (
RPS # Used for remote phonon scattering (RPS).
NegInterfaceCharge (SurfaceName="s1")
# Used for remote Coulomb scattering (RCS)
# and remote dipole scattering (RDS).
PosInterfaceCharge (SurfaceName="s1") # Used for RCS and RDS.
)
HighFieldSaturation)

All other options and statements remain the same.

The mobility of unstrained long-channel devices at low-drain bias voltage does not usually
depend on the channel direction for the (100) surface. Even so, the warping of selected
valence bands results in an orientation-dependent mass; the measurements do not show a
dependence on the channel direction for very long channels. This might be a result of the
many scattering processes and randomizations of the k-direction in long-channel devices.
For short-channel devices, the situation is completely different. First, the saturation velocity
depends on the channel direction for higher fields. Second, the ballistic transport depends
on the channel direction. Third, the warping of the valence band structure becomes
important because there are fewer scattering effects.

Advanced Calibration for Device Simulation User Guide 29

V-2024.03
Chapter 2: Guide to Device Simulations
Accessing Parameter Files and Model Frameworks

Path E: FinFET Simulations

You can use the same Physics section as shown in Path C. The AutoOrientation option
for the density gradient model must be switched on. As a starting point for calibration, the
same parameter files Siliconc100.par, Siliconc110.par,
SiliconGermaniumc100.par, and SiliconGermaniumc110.par can also be used.
However, because the interface quality of the fin sidewalls can be very different from the
interface quality of a planar transistor, parameter adjustments are often necessary.
In general, the electron inversion mobility of FinFETs is higher at the (110) sidewalls than in
planar (110) transistors [1].
Table 4 and Table 5 show the approach for modeling planar and nonplanar FinFET devices
regarding orientation-related parameters.
Table 4 Parameter files for planar FETs and FinFETs with silicon channel

Device Number of Conducting Channel Parameter file to load

conducting planes plane orientation orientation

Planar MOS 1 (100) <100> Siliconc100.par

Planar MOS 1 (110) <100> Siliconc100.par

Planar MOS 1 (100) <110> Siliconc110.par

Planar MOS 1 (110) <110> Siliconc110.par

FinFET 3 (100) <100> Siliconc100.par

FinFET 3 (100) <110> Siliconc110.par

(110) <110>

Table 5 Parameter files for planar FETs and FinFETs with SiGe channel

Device Number of Conducting Channel Parameter file to load

conducting plane orientation orientation
planes

Planar MOS 1 (100) <100> SiliconGermaniumc100.par

Planar MOS 1 (110) <100> SiliconGermaniumc100.par

Planar MOS 1 (100) <110> SiliconGermaniumc110.par

Planar MOS 1 (110) <110> SiliconGermaniumc110.par

Advanced Calibration for Device Simulation User Guide 30

V-2024.03
Chapter 2: Guide to Device Simulations
Accessing Parameter Files and Model Frameworks

Table 5 Parameter files for planar FETs and FinFETs with SiGe channel (Continued)

Device Number of Conducting Channel Parameter file to load

conducting plane orientation orientation
planes

FinFET 3 (100) <100> SiliconGermaniumc100.par

FinFET 3 (100) <110> SiliconGermaniumc110.par

(110) <110>

For very short gate lengths (less than 15 nm), the isotropic density gradient model can
overestimate the leakage current in the transport direction and quantization effects at doping
and mole fraction gradients. In those cases, the alpha parameter in the density gradient
model can help to suppress artificial source/drain leakage and quantization effects in the
transport direction. For FinFET simulations with a gate length below 15 nm, using an alpha
in the transport direction smaller than 0.1 is a good starting point. For more information
about handling and using the anisotropic density gradient model, see the Sentaurus™
Device User Guide.
When using EparallelToInterface as a driving force in the high-field saturation model,
the electric field components that are normal to the channel direction and in-plane with the
fin cross section can lower the mobility. This is an artifact and makes the use of this driving
force very questionable.
Note:
Using the DOS(eMass hMass) model in the Piezo section and the temperature
equation at the same time might result in convergence problems. Using the
NumericalIntegration option, for example,
DOS(hMass(NumericalIntegration)), can improve the numeric behavior.

The mobility of unstrained long-channel devices at low-drain bias voltage does not usually
depend on the channel direction for the (100) surface. Even so, the warping of selected
valence bands results in an orientation-dependent mass; the measurements do not show a
dependence on the channel direction for very long channels. This might be a result of the
many scattering processes and randomizations of the k-direction in long-channel devices.
For short-channel devices, the situation is completely different. First, the saturation velocity
depends on the channel direction for higher fields. Second, the ballistic transport depends
on the channel direction. Third, the warping of the valence band structure becomes
important because there are fewer scattering effects.

Advanced Calibration for Device Simulation User Guide 31

V-2024.03
Chapter 2: Guide to Device Simulations
Accessing Parameter Files and Model Frameworks

Path F: Smart-Power HVMOS and LDMOS Devices

In principle, you should use Path A for silicon-based smart-power devices exhibiting a
traditional polysilicon/oxide gate stack when no mechanical stress occurs in the structure.
When mechanical stress occurs in the structure, you must use Path B. However, in many
cases, the linear piezoresistance model might be sufficient because the mechanical stress
is relatively low. For complex devices or 3D simulation structures, you might need to simplify
the model complexity to reduce simulation time or to achieve convergence. You can use the
following simplified Physics section in those cases:
Physics {Fermi}

Physics (Material="Silicon") {
Mobility (Enormal (IALMob(AutoOrientation)) HighFieldSaturation)
# For PMOS, it is useful to use PhononCombination=2 in IALMob.
EffectiveIntrinsicDensity(OldSlotboom)
Recombination (Auger SRH(DopingDep TemDep))
# Recombination(Auger SRH(DopingDep TemDep) Band2Band)
# For off-current calculation, band-to-band tunneling models
# must be switched on.
}

Quantum correction is neglected here, resulting in better simulation performance. The loss
of accuracy can mostly be compensated by a corresponding workfunction shift and a
reduction of the gate isolation permittivity. For devices with a gate isolation oxide thickness
greater than 3 nm, the influence of quantum correction decreases. The Physics section
above is designed for Id–Vd and Id–Vg simulations, but it does not include the models for
breakdown, substrate current, and complex electrostatic discharge simulation. In addition,
you must switch on the commands needed to account for self-heating.
Parameter files and the model section have been tested for LDMOS devices. No tests or
investigations have been conducted for IGBTs, thyristors, or bipolar devices.

Path G: GaN HEMT Simulations

Typically, you should use the density gradient model for quantum correction. However, no
model for quantum correction is included because of missing parameters.
The values for the trap concentration and the energy level given here are typical for
nitride-passivated GaN HEMTs. However, the actual values depend on the device and might
differ from the values presented here. In addition to the interface traps shown in the
command file Physics section here, bulk traps are important and must be implemented in
the simulation, for example, when investigating current collapse and gate leakage effects.
For III–V nitride wide-bandgap materials, you must load the parameter files GaN.par,
AlGaN.par, AlN.par, InGaN.par, and InN.par, which contain the material parameters.

Advanced Calibration for Device Simulation User Guide 32

V-2024.03
Chapter 2: Guide to Device Simulations
Accessing Parameter Files and Model Frameworks

The following Physics section should be used together with the parameter files:
Physics {
Mobility (
Enormal(IALMob) # Arora and Masetti models can also be
# used but cannot describe the mobility
# in the channel inversion layer.
# Model parameters are in the parameter
# files GaN.par, AlGaN.par, and AlN.par.
HighFieldSaturation
)
EffectiveIntrinsicDensity(noBandGapNarrowing)
Piezoelectric_Polarization(strain)
Recombination(Radiative)
Fermi
Thermodynamic
Thermionic
HeteroInterface
eBarrierTunneling "STUN" # For tunneling at the source contact,
# you must define a nonlocal mesh in
# the Math section.
eBarrierTunneling "DTUN" # For tunneling at the drain contact,
# you must define a nonlocal mesh in
# the Math section.
}

Physics (MaterialInterface="AlGaN/Nitride") {
Traps(Donor Level Conc=1E13 EnergyMid=0.4 FromMidBandGap)
# The values for the trap concentration and the energy
# are calibration parameters as well as the trap
# location and the trap type. Several combinations
# of different trap distributions are usually
# necessary to reproduce measured results.
Piezoelectric_Polarization(strain activation=0)
}

In addition to tunneling models at the source and drain contacts, tunneling models below the
gate contact, finally covering the entire gate stack, down to the GaN inversion channel, are
necessary to simulate gate leakage. These tunneling models must consider both direct
tunneling and trap-assisted tunneling.
Note:
The choice of the transition between the 2D electron gas and the 3D bulk situation
is critical. This transition is controlled by the parameters lcrit, lcrit_c, and S
in the parameter file.

Path H: SiC Diode Simulations

The parameter files 4H-SiC.par and 6H-SiC.par contain the material parameters for
SiC-based wide-bandgap materials.

Advanced Calibration for Device Simulation User Guide 33

V-2024.03
Chapter 2: Guide to Device Simulations
Accessing Parameter Files and Model Frameworks

The following Physics section should be used together with the parameter files:
Physics {
eMobility (
DopingDependence(Arora) # For 6H-SiC, Arora parameters are not
# available in the parameter file.
# Therefore, it is recommended to
# switch to the Masetti model.
HighFieldSaturation
)
hMobility (
DopingDependence(Arora) # For 6H-SiC, Arora parameters are not
# available in the parameter file.
# Therefore, it is recommended to
# switch to the Masetti model.
HighFieldSaturation
)
EffectiveIntrinsicDensity(Slotboom NoFermi)
IncompleteIonization
Recombination(SRH(DopingDep TempDep Tunneling) Band2Band Avalanche
(Okuto))
# For forward characteristics, you can
# neglect the commands Band2Band and
# Tunneling. This can improve
# convergence behavior.
Fermi
#Thermodynamic # Add for thermodynamic simulations.
#AnalyticTEP # Add for thermodynamic simulations.
Temperature=300.0
eBarrierTunneling "NLM" (Twoband Transmission)
# Nonlocal mesh definition in Math section necessary.
Traps (
Donor Level EnergyMid=1.0 FromConductionband Conc=1e14
eXSection=1.0e-12 hXSection=1.0e-12 PooleFrenkel
TrapVolume=1.0e-9
HuangRhys=0.1 PhononEnergy=0.05 eBarrierTunneling (Twoband)
) # Bulk traps are often necessary for the physically
# correct calculation of the thermal behavior of the
# reverse leakage current.
}

Physics (Electrode="top") {
#MSPeltierHeat # Add for thermodynamic simulations.
Schottky BarrierLowering
}

Depending on how the mobility model was calibrated, that is, whether or not the total
chemical concentration or the active concentration was used, you can add
IncompleteIonization to the Mobility statement. For holes, the suggested parameters
for the Arora model in the 4H-SiC.par file have been calibrated to the total acceptor
concentration. In addition, scattering on the neutral impurities is often comparable to
scattering on the ionized impurities for 4H-SiC p-type layers.

Advanced Calibration for Device Simulation User Guide 34

V-2024.03
Chapter 2: Guide to Device Simulations
Accessing Parameter Files and Model Frameworks

Note:
Using the IncompleteIonization model might cause convergence problems.

To distinguish between cubic and hexagonal 4H-SiC lattice sites, you can activate the
following option in the IncompleteIonization statement of the Physics section of the
command file:
IncompleteIonization(Split (Doping="NitrogenActive" Weights=(0.5 0.5)))
The corresponding parameters must be defined in the parameter file. Precalibrated values
are in the 4H-SiC.par and 6H-SiC.par files for phosphorus and nitrogen. For phosphorus,
the parameters corresponding to hexagonal sites are active, and modification of the
datexcodes.txt file is necessary to treat both sites.

The values for the trap concentration, the energy levels, and other parameters in the Traps
command are typical. However, the actual values depend on the device and might differ
from the values presented here. Especially for the leakage current in reverse mode,
tunneling by defect traps can play an important role and should be investigated.
The use of the NoFermi option depends on the extraction of the parameters for the bandgap
narrowing model. This option is applied if you use parameters for bandgap narrowing that
have been extracted assuming Fermi statistics.
When including heat generation and conduction, a lumped model (thermal network) using
the thermal resistance of layers and material regions that are not included in the simulation
can improve the simulation results.
You can use an alternative model, the JainRoulston model, which is physically more suited
for bandgap narrowing.
For 2D and 3D structures, anisotropic material properties for mobility, permittivity, thermal
conductivity, and impact ionization must be taken into account in the simulation. This can be
performed by the following statement in the Physics section:
Aniso (Mobility Poisson Temperature Avalanche)

The corresponding parameters must be defined in the parameter file. Precalibrated values
are in the 4H-SiC.par (same as SiliconCarbide.par) and 6H-SiC.par files for mobility,
permittivity, thermal conductivity, impact ionization, incomplete ionization, high-field
saturation, and other models.

Path I: InGaAs FinFET Simulations

For InGaAs FinFET devices with a gate width less than 10 nm, the following quantization
models of Sentaurus Device are available to model quantized distributions of carriers:
• Use the modified local-density approximation (MLDA) model with the ThinLayer option
specified in the command file to activate size quantization with the multivalley model. For

Advanced Calibration for Device Simulation User Guide 35

V-2024.03
Chapter 2: Guide to Device Simulations
Accessing Parameter Files and Model Frameworks

a 2D structure, the layer thickness can be extracted automatically, and the depletion at
corner regions corresponds better to the results of the Schrödinger equation using the
MaxFit option. Setting MaxFitWeight=0.35 is advisable to start with, although an
adjustment might be required depending on fin geometries.
• Use the density gradient model with the multivalley model. The quantization parameter
(-parameter) of the density gradient model depends on the layer thickness and the fin
geometry. Therefore, it is advisable to calibrate the solutions from the Schrödinger
equation in test structures with a similar channel geometry.
In addition, the inversion and accumulation layer mobility (IALMob) model is available to
model inversion-type n-channel InGaAs FinFETs. The parameters have been calibrated to
available experimental data under the condition of PhononCombination=1.
The parameter files GaAs.par, InGaAs.par, and InAs.par contain the material parameters
for III–V arsenide materials.
For the MLDA model, you must use the following Physics section with the parameter files:
Physics {Fermi}

Physics (Material="InGaAs") {
Mobility (
eHighFieldSaturation # In case of convergence issues, you can use
# eHighFieldSaturation(EparallelToInterface).
Enormal(IALMob(PhononCombination=1))
)
EffectiveIntrinsicDensity(BandGapNarrowing(JainRoulston) NoFermi)
Recombination(SRH Auger)
eMultiValley(MLDA Nonparabolicity ThinLayer)
LayerThickness(MaxFitWeight=0.35)
}

For the density gradient model, you must use the following Physics section with the
parameter files:
Physics {Fermi}

Physics (Material="InGaAs") {
Mobility (
eHighFieldSaturation # In case of convergence issues, you can use
# eHighFieldSaturation(EparallelToInterface).
Enormal(IALMob(PhononCombination=1))
)
EffectiveIntrinsicDensity(BandGapNarrowing(JainRoulston) NoFermi)
Recombination(SRH Auger)
eMultiValley(Nonparabolicity)
eQuantumPotential(Density) # -parameter requires calibration.
}

Advanced Calibration for Device Simulation User Guide 36

V-2024.03
Chapter 2: Guide to Device Simulations
References

Path J: SiC HVMOS Simulations

Typically, you should use the density gradient model for quantum correction. However, no
model for quantum correction is included because of missing model parameters.
For SiC wide-bandgap materials, you must load the parameter file 4H-SiC.par or
SiliconCarbide.par, which contains the material parameters.
Note:
The choice of the transition between the 2D electron gas and the 3D bulk situation
is critical. This transition is controlled by the parameters lcrit, lcrit_c, and S
in the parameter file.

The following Physics sections should be used together with the parameter file:
Physics {
Fermi
DefaultParametersFromFile
}

Physics (Material = "SiliconCarbide") {

IncompleteIonization
Mobility (
IncompleteIonization
Enormal(IALMob(PhononCombination = 2))
HighFieldSaturation
)
IncompleteIonization(Split (Doping = "NitrogenConcentration"
Weights = (0.5 0.5)))
EffectiveIntrinsicDensity(BandGapNarrowing(Slotboom) NoFermi)
Recombination (SRH(DopingDependence TempDependence) Auger)
}

References
[1] C. D. Young et al., “(110) and (100) Sidewall-oriented FinFETs: A performance and
reliability investigation,” Solid-State Electronics, vol. 78, pp. 2–10, December 2012.

Advanced Calibration for Device Simulation User Guide 37

V-2024.03
3
Calibration Methodology for Device Simulations
3

This chapter describes the basic calibration methodology used to improve the accuracy of
CMOS device simulations.

Calibration of CMOS Devices

Figure 1 shows a general flow of the calibration methodology used for CMOS devices.
Different measurements and layout configurations allow you to separate device simulation
model parameters. However, in many cases, iterations are necessary.

Figure 1 Calibration methodology used for CMOS devices (Id–Vg calibration can be
complemented by involving mobility measurements)

C–V Calibration to Extract Permittivities of Gate Stack and Workfunction, and to Validate Channel Doping Profile

Id–Vg Mobility Calibration (Low-Drain Bias Voltage, Long-Channel Device)

Calibration of Parasitic Series Resistance Using Ron of Different Gate Length

and a Corresponding Ballistic Mobility Model

Id–Vg Mobility Calibration (Low-Drain Id–Vg Mobility Calibration (High-Drain

Bias Voltage, Different Gate Length) Bias Voltage, Different Gate Length)

The next sections describe this calibration methodology in detail. It is assumed that all
process calibration issues are solved beforehand and the doping profile is correct.

C–V Calibration
Capacitance–voltage (C–V) calibration is the first step in the device calibration and is
strongly connected to the doping profile and to process simulation or calibration. Because it

Advanced Calibration for Device Simulation User Guide 38

V-2024.03
Chapter 3: Calibration Methodology for Device Simulations
Calibration of CMOS Devices

is assumed that the doping profile is correct, the bottom of the C–V curve (marked in
Figure 2) fits the experiment already.
C–V calibration is performed on large-area MOS transistors or MOS capacitor test
structures at low frequency. Therefore, in some cases, you can have slightly different
workfunctions compared to smaller transistor devices.

Figure 2 C–V calibration methodology (NMOS)

B: Extract average gate-stack permittivity

C: Check poly depletion

and high-frequency effects;
iterate with B regarding
average gate-stack permittivity

A: Extract workfunction

Table 6 Main steps and extracted parameters of C–V calibration

Step Target Parameter to Validity range of Comments

extract parameter

A Region of weak Workfunction Metal gate Workfunction or barrier voltages

or moderate workfunction: extracted from C–V and used for
inversion NMOS (4.0–4.7 eV) long-channel I–V simulation can
require corrections, particularly
PMOS (4.5–5.2 eV)
when special C–V test
Polygate barrier: structures are used.
–0.2 V – 0.2 V

B Accumulation Average Permittivity of high-k For high-k gate stacks with

part of C–V permittivity of material: 14–25 several material layers, first the
gate isolation Permittivity of permittivity of the interfacial
oxide layer is adjusted. Second,
interfacial oxide:
the permittivity of the high-k
3.9–6
material is calibrated.

C Inversion part Poly doping in Depends on –

of C–V case of process conditions
polygate

Advanced Calibration for Device Simulation User Guide 39

V-2024.03
Chapter 3: Calibration Methodology for Device Simulations
Calibration of CMOS Devices

Id–Vg Mobility Calibration

The calibration of long-channel devices is necessary to extract low-field mobility
parameters. Typically, for this, you use the Id–Vg characteristics of devices with a gate length
greater than 500 nm, at a low-drain bias voltage smaller than 100 mV. An iterative process
extracts the parameters of the mobility model in different voltage regions (see Figure 3,
Figure 4, and Table 7).

Figure 3 Long-channel Id–Vg mobility calibration methodology in linear scale

B: Extract delta parameter

of surface roughness model

C: Extract C parameter of surface

phonon mobility degradation term

A: Extract workfunction

Figure 4 Long-channel Id–Vg mobility calibration methodology in logarithmic scale

D: Extract D1_inv, D2_inv, and

nu0_inv of 2D Coulomb scattering
term of IALMob model

D and E: Extract Conc parameter

in the trap model

Advanced Calibration for Device Simulation User Guide 40

V-2024.03
Chapter 3: Calibration Methodology for Device Simulations
Calibration of CMOS Devices

Table 7 Main steps and extracted parameters of long-channel Id–Vg mobility calibration

Step Target Parameter to Validity range of Comments

extract parameter

A Threshold Workfunction Depends on gate Take initial value for

voltage material. Stress in workfunction from C–V
the metal gate, grain calibration. Note there can
sizes, and be an influence of remote
orientations influence Coulomb or dipole
the workfunction and scattering (RCS or RDS),
can result in a or 2D impurity Coulomb
dependency on gate scattering on the
length and geometric threshold voltage, which
parameters. couples workfunction and
mobility extraction.

B Strong delta parameter of Depends on process Surface roughness can

inversion – high surface roughness conditions and other change when changing
gate bias model parameters. No process conditions and
voltage validity range is surface orientations.
known. Fine-tuning is always
necessary to fit Id–Vg
characteristics.

C Linear region of C parameter of Depends on process Extraction requires mostly

Id–Vg curve surface phonon conditions and other an iteration between steps
mobility degradation parameters. No B and C and, sometimes,
term validity range is even between steps B, C,
known. and D. Interface
properties and material
composition at the
interface can change
when changing process
conditions. Fine-tuning is
always necessary to fit Id–
Vg characteristics.

Advanced Calibration for Device Simulation User Guide 41

V-2024.03
Chapter 3: Calibration Methodology for Device Simulations
Calibration of CMOS Devices

Table 7 Main steps and extracted parameters of long-channel Id–Vg mobility calibration

Step Target Parameter to Validity range of Comments

extract parameter

D Transition 1. Parameters of 2D 1. No validity range Calibration focuses on the

between Coulomb mobility is known Coulomb roll-off visible in
logarithmic and degradation term the µ(Eeff) curve, but it
linear slope – also can be done to the
2. Parameters of 2. Parameters of
threshold bending of the Id–Vg curve
InterfaceCharge InterfaceCharge
voltage region
model mode between subthreshold
and linear region. These
mobility degradation
terms also affect the
threshold voltage and the
drain-induced barrier
lowering (DIBL) of the
device.

E Subthreshold Parameters of No validity range is Calibration focuses on the

region InterfaceCharge known Coulomb roll-off visible in
model the µ(Eeff) curve, but it
also can be done to the
subthreshold region. This
mobility degradation term
also affects the threshold
voltage and the DIBL of
the device.

Calibration of Parasitic Series Resistance

With the inclusion of a ballistic mobility model that gives a finite resistance for an infinitely
small channel length, the calibration of the channel mobility is coupled to the calibration of
the source/drain channel series resistance. Therefore, the extraction of this series
resistance is most important for the calibration of the parameters of the ballistic mobility
model. Iterations might be needed to update the extracted series resistance when changing
parameters of the ballistic mobility model in the calibration of the channel mobility.

Advanced Calibration for Device Simulation User Guide 42

V-2024.03
Chapter 3: Calibration Methodology for Device Simulations
Calibration of CMOS Devices

Figure 5 Extraction methodology of source/drain series resistance

Ron

R0

F: Extract source/drain series resistance

Gate Length

Equation 1 shows the parameters R0 and Rs forming the on-resistance:

R on = R0 + Lg  Rs (1)

Table 8 Main step and extracted parameter of source/drain series resistance

Step Target Parameter to Validity range of Comments

extract parameter

F Measured On-resistance at R0 is determined by R0 extraction is

dependency of zero gate length R0 contact and source/ especially important
on-resistance at (in Equation 1) drain series resistance. for step G
low-drain bias The value depends on (extraction of
voltage on gate the technology node “ballistic”
length and device type, and resistance).
can vary greatly. No
validity range is known.

Id–Vg Mobility Calibration at Low-Drain Bias Voltage

Reference [1] discusses a possible overestimation of the current response close to
equilibrium when having strong electrical built-in fields, for example, caused by p-n
junctions. This effect might be connected to deviations from the Einstein relation in certain
situations, but it is not considered in the calibration methodology because of convergence
issues. This means, to some extent, that the calibration of the parameters of the ballistic
mobility model includes this physically different effect.

Advanced Calibration for Device Simulation User Guide 43

V-2024.03
Chapter 3: Calibration Methodology for Device Simulations
Calibration of CMOS Devices

Figure 6 Short-channel Id–Vg mobility calibration methodology in linear scale, with low-drain
bias voltage

G: Extract “ballistic resistance”

mobility contribution

Table 9 Main step and extracted parameter of Id–Vg mobility calibration at low-drain bias
voltage

Step Target Parameter to extract Validity range of Comments

parameter

G Moderate and Adjust “ballistic resistance” No validity range This involves

strong inversion mobility contribution using iteration with step F.
parameters of the ballistic
mobility models in Sentaurus
Device

Id–Vg Mobility Calibration at High-Drain Bias Voltage

This step extracts the saturation velocity. The extraction of vsat0 from Idsat roll-off data or
Id–Vg curves at large drain bias voltage, using experiments or the Boltzmann transport
equation solver, is straightforward. However, the calibration of both Idsat and Idlin at the same
time using vsat0, source/drain resistance, and ballistic mobility parameters is more
challenging.
Nevertheless, it is mostly possible to achieve a good agreement during the device
calibration because of the inclusion of the ballistic mobility model [2].

Advanced Calibration for Device Simulation User Guide 44

V-2024.03
Chapter 3: Calibration Methodology for Device Simulations
References

Figure 7 Short-channel Id–Vg mobility calibration methodology in linear scale, with high-drain
bias voltage

H: Extract vsat0 parameter

of high-field saturation model

Table 10 Main step and extracted parameter of Id–Vg mobility calibration at high-drain bias
voltage

Step Target Parameter to Validity range of Comments

extract parameter

H Whole Id–Vg curve Adjust vsat0 For silicon: < 5e7 cm/s This might involve iteration
and Idsat with steps F and G.

References
[1] A. T. Pham, C. Jungemann, and B. Meinerzhagen, “Modeling and validation of
piezoresistive coefficients in Si hole inversion layers,” Solid-State Electronics, vol. 53,
no. 12, pp. 1325–1333, 2009.
[2] A. Erlebach, K. H. Lee, and F. M. Bufler, “Empirical Ballistic Mobility Model for
Drift-Diffusion Simulation,” in Proceedings of the 46th European Solid-State Device
Research Conference (ESSDERC), Lausanne, Switzerland, pp. 420–423,
September 2016.

Advanced Calibration for Device Simulation User Guide 45

V-2024.03
4
Descriptions of Models
4

This chapter presents a review of selected models of Sentaurus Device.

Overview of Models
Given the many models implemented in Sentaurus Device for the simulation of device
characteristics, it is often difficult to quickly design appropriate command files for device
simulation. Therefore, this chapter presents a review of selected models, which are
evaluated with respect to their capabilities in simulating certain device aspects.
In the quantum drift-diffusion (QDD) approximation, electron and hole mobilities determine
charge carrier transport. Advanced Calibration Device focuses on low-field mobility and
high-field mobility, with and without mechanical stress, for planar and nonplanar silicon and
SiGe FETs. In addition, there is a discussion of models and parameters for InGaAs, AlGaN,
and SiC devices.

CMOS Devices
This section discusses models relevant to CMOS devices.

Unstrained Band Structure and Electrostatics

In Sentaurus Device, the band gap, the electron affinity, the bandgap narrowing, the
density-of-states, and the dimensions of the device define the band structure. The
temperature dependency of these quantities and the doping dependency of the bandgap
narrowing are represented by empirical functions.

Strained Band Structure and Electrostatics

The band structure changes with the mechanical stress in the device. The dependency of
the conduction and valence bands and, therefore, of the band gap and electron affinity, on

Advanced Calibration for Device Simulation User Guide 46

V-2024.03
Chapter 4: Descriptions of Models
CMOS Devices

the mechanical stress or strain is described by the deformation potential model where the
band structure information comes from the k·p calculation. You can also extract the stress
dependency of the density-of-states from the k·p calculation.

Quantization and Band Structure

The bulk band structure changes in thin potential wells, such as inversion layers and thin
films. These changes due to the quantization interact, compensate, or strengthen the stress
influence on the band structure. In Sentaurus Device, the quantization is typically calculated
by the density gradient model or the modified local-density approximation (MLDA) model.

Unstrained Long-Channel Low-Field Mobility

In the limit of low-driving fields, several mechanisms can degrade mobility. Typically,
MISFET simulation takes into account mobility degradation by ionized dopants, surface
roughness scattering, surface phonon scattering, and bulk phonon scattering. For high-k
gate stacks, you must consider additional remote scattering mechanisms such as remote
Coulomb scattering (RCS), remote phonon scattering (RPS), and remote dipole scattering
(RDS). A crucial point is the transition between bulk mobility and inversion or accumulation
layer mobility. Because of the transition from 3D to 2D charge carrier gas, the scattering
behavior of electrons and holes changes and screening becomes different. Furthermore, the
mobility model must include dependency on the channel material (for example, the
germanium mole fraction), the channel direction, and the surface orientation. For thin silicon
films in the range of a few nanometers, additional contributions come into effect that cause
a dependency of the mobility on the film thickness.
In summary, a drift-diffusion or QDD low-field mobility model for the simulation of SiON/SiO2
or high-k gate stack FETs without mechanical stress should have the following features:
• Mobility degradation by ionized impurities – impurity Coulomb scattering (ICS)
• Mobility degradation by the roughness of the surface or interface – surface roughness
scattering (SRS)
• Mobility degradation by interaction with bulk phonons – bulk phonon scattering (BPS)
• Mobility degradation by interaction with surface phonons – surface phonon scattering
(SPS)
• Mobility degradation by charges in the gate isolator (high-k gate stack) (RCS)
• Mobility degradation by different permittivity in the gate isolator (high-k gate stack) (RPS)
• Mobility degradation by dipole configurations in the gate isolator (high-k gate stack)
(RDS)

Advanced Calibration for Device Simulation User Guide 47

V-2024.03
Chapter 4: Descriptions of Models
CMOS Devices

• Mole fraction dependency of the low-field mobility model parameters (for example, alloy
scattering)
• Transition between bulk scattering and scattering in 2D charge carrier gas for ICS
• Parameters for different channel and surface or interface orientations; channel: <100>
and <110>; surface: (100) and (110)
• Mobility degradation for very thin silicon films
• Mobility in thin films, including influence of quantization
• Influence of mobility anisotropy as in SiC and GaN
Sentaurus Device offers the following low-field mobility degradation models for MISFET
simulation:
• Bulk phonon-limited mobility model (BPS):
Physics { ConstantMobility }

• Mobility degradation by ionized impurities (in some cases, it contains bulk

phonon-limited mobility, in the low-doping limit) (ICS):
◦ Masetti model provides mole fraction–dependent model parameters:
Physics { Mobility ( DopingDependence (Masetti) ) }

◦ Arora model provides mole fraction–dependent model parameters:

Physics { Mobility ( DopingDependence (Arora) ) }

◦ Philips unified mobility model includes screening and carrier–carrier scattering, and
provides mole fraction–dependent model parameters:
Physics { Mobility (PhuMob) }

◦ University of Bologna bulk mobility model is specially calibrated for an extended

temperature range and provides mole fraction–dependent model parameters:
Physics { Mobility ( DopingDependence (UniBo) ) }

◦ Carrier–carrier scattering models (Conwell–Weisskopf and Brooks–Herring):

Physics { Mobility (CarrierCarrierScattering) }

• Mobility degradation at interfaces:

◦ Enhanced Lombardi model includes automatic detection of surface orientation and
provides mole fraction–dependent model parameters:
Physics { Mobility ( Enormal (Lombardi) ) }

Advanced Calibration for Device Simulation User Guide 48

V-2024.03
Chapter 4: Descriptions of Models
CMOS Devices

◦ Inversion and accumulation layer mobility model includes automatic detection of

surface orientation, provides mole fraction–dependent model parameters, and
contains the Philips unified mobility model, the enhanced Lombardi model, and
additional terms for Coulomb scattering in 2D electron gas:
Physics { Mobility ( Enormal (IALMob) ) }

◦ University of Bologna surface mobility model is specially calibrated for an extended

temperature range, includes screening of bulk ICS, no special model for ICS in 2D
charge carrier gas, and simplified Lombardi-style model for acoustic phonon and
surface roughness scattering:
Physics { Mobility ( Enormal (UniBo) ) }

◦ 2D Coulomb scattering model for ionized impurities, which must not be used with the
PhuMob model or any other DopingDependence model because mobility degradation
is counted twice, is designed for Coulomb scattering (ICS) in 2D charge carrier gas:
Physics { Mobility ( Enormal (Lombardi Coulomb2D) ) }

◦ 2D Coulomb scattering models for positive and negative charges at remote interfaces
allow you to distinguish between RCS and RDS:
Physics { Mobility ( Enormal (Lombardi NegInterfaceCharge
PosInterfaceCharge) ) }

◦ Thin-layer mobility model describes mobility degradation due to finite silicon film
thickness, includes dependency of phonon scattering on quantization as well as
additional empirical mobility degradation terms, and includes automatic detection of
surface orientation. The model can be combined with Enormal-like models:
Physics { Mobility (ThinLayer) }

Because of the use of Mathiessen’s rule, several combinations of these mobility models are
possible. For MISFETs, you usually need to combine bulk phonon and Coulomb scattering
with scattering at interfaces and surfaces. Historically, the following combinations have been
used:
1. DopingDependence(Masetti) + Lombardi
2. PhuMob + Lombardi
3. IALMob
4. IALMob + RPS + NegInterfaceCharge + PosInterfaceCharge
5. RPS + NegInterfaceCharge + PosInterfaceCharge + ThinLayer(IALMob)
Combination 2 has been the standard for a long time. However, several problems require a
change to another low-field mobility model. First, the transition between 3D bulk Coulomb
scattering and Coulomb scattering in the inversion layer is not described correctly because

Advanced Calibration for Device Simulation User Guide 49

V-2024.03
Chapter 4: Descriptions of Models
CMOS Devices

the PhuMob model, which is a bulk model, is used for the mobility degradation by scattering
at ionized impurities from accumulation to strong inversion. This causes problems in
describing the onset of moderate or strong inversion around the threshold voltage in the Id–
Vg curve. Second, in thin silicon layers, it is mandatory to include the dependency on the film
thickness.
Note:
For these reasons, you must change the low-field mobility framework and switch
to combination 3, 4, or 5. For MISFET simulation, the recommended models are
combination 3 (devices without high-k gate stacks), combination 4 (devices with
high-k gate stacks), and combination 5 (thin films).

Strained Low-Field Mobility

This section discusses strained low-field mobility.

Influence of Mechanical Stress

Modeling the influence of stress on the channel mobility must deal with several typical
cases. These cases must be simulated accurately to minimize additional device calibration.
The typical cases are as follows:
• Planar silicon channel PMOS with compressive stress up to 3 GPa in the channel
direction and up to 2 GPa tensile stress normal to the silicon surface (SiGe pockets and
dual stress liner (DSL))
• Planar silicon channel NMOS with tensile stress up to 2 GPa in the channel direction and
up to 2 GPa normal to the silicon surface (SiC pockets, DSL, and stress memorization
technique)
• Planar PMOS channel with biaxial strained SiGe channel and stress between 1 MPa and
4 GPa in the channel direction and up to 2 GPa tensile stress in the normal direction
(SiGe pockets and DSL)
• Nonplanar silicon channel NMOS (FinFET) with mixed compressive or tensile stress
components of up to 2 GPa depending on the configuration (for example, with SiC
pockets wrapped around the fin and metal gate as a liner replacement)
• Nonplanar silicon channel PMOS (FinFET) with mixed compressive or tensile stress
components of up to 2 GPa depending on the configuration (for example, with SiGe
pockets wrapped around the fin and metal gate as a liner replacement)

Advanced Calibration for Device Simulation User Guide 50

V-2024.03
Chapter 4: Descriptions of Models
CMOS Devices

• Nonplanar SiGe or Ge channel NMOS (FinFET) with mixed compressive or tensile

stress components of up to 4 GPa depending on the configuration
• Nonplanar SiGe or Ge channel PMOS (FinFET) with mixed compressive or tensile stress
components of up to 4 GPa depending on the configuration
You must simulate stress configurations for several wafer orientations (such as (001) and
(110)), channel directions (such as <100> and <110>), and channel materials (such as
silicon and SiGe). Typically, all three stress components (sxx, syy, and szz) are important. A
model validity check, which is limited to uniaxial or biaxial stress configurations, is not
sufficient. In addition, the response of mobility degradation components (such as phonon
scattering, Coulomb scattering, and remote scattering) to mechanical stress differs and
requires the separate treatment of each component.
Typically, in Sentaurus Device, the total mobility is scaled with a stress enhancement factor
calculated with the models discussed here. The following models in Sentaurus Device
describe the influence of stress on mobility:
• Piezoresistance mobility models:
◦ First-Order (Linear) Piezoresistance Mobility Model
◦ Second-Order Piezoresistance Mobility Model
◦ Nonlinear Piezoresistance Models for Electrons and Holes (MCmob and SBmob)
• Occupation-based and band structure–based models:
◦ Intel Stress-Induced Hole Mobility Model (hSixBand)
◦ eSubband Model for Electrons
◦ hSubband Model for Holes
You can choose several additional options. For the piezoresistance mobility models, the
Kanda option includes the dependency of the stress-induced mobility change on doping
concentration. Differences in the influence of mechanical stress on minority and majority
carriers can be considered by an additional fitting parameter. You can also calibrate the
influence of the normal electric field in a metal–insulator–semiconductor (MIS) structure and
of the germanium mole fraction on the first-order piezoresistive coefficients. Because the
first-order and second-order piezoresistance mobility models are bulk models, the default
piezoresistive coefficients are not accurate for inversion layers and must be adjusted. The
first-order and second-order piezoresistance mobility models can be used as factor models
or tensor models, thereby introducing anisotropic properties.

Advanced Calibration for Device Simulation User Guide 51

V-2024.03
Chapter 4: Descriptions of Models
CMOS Devices

You can fine-tune the influence of the saturation velocity on stress-dependent mobility by
setting the SaturationFactor parameter that controls how strongly the saturation velocity
depends on stress as follows:
• Setting SaturationFactor=0 means the mobility change is applied to the low-field
mobility only.
• Setting SaturationFactor=1 means the mobility change applies to the total mobility.
This is the default in Sentaurus Device.
Because it is known from Monte Carlo simulation and from experiments that short-channel
transistors show a much lower mobility enhancement due to mechanical stress, it is better
to apply the mobility change to the low-field mobility only. Further adjustment of this
parameter should be performed with good references only.
The occupation-based and band structure–based models characterize the dependency of
occupation and band structure on mechanical stress for the calculation of the mobility.
These models have their own dependency on the doping concentration, and a Kanda-like
option is not necessary. In addition, the electric field dependency should be a result of the
model itself.

First-Order (Linear) Piezoresistance Mobility Model

This model uses the piezoresistance tensor with constant coefficients to calculate the
stress-induced mobility change.

Options
The model can be used as a factor (isotropic) model or a tensor (anisotropic) model. The
piezoresistive coefficients can depend on the normal electric field in an inversion layer.
Minority and majority charge carrier transport can be distinguished. You can include
dependency on the doping concentration. Calibration of dependency of the influence of
mechanical stress on the saturation velocity is possible.

Advantages
The model is well adjusted to measurements for the bulk case. It is fully transformable, so it
can consider all stress configurations and current directions in a bulk situation. It is easy to
use and to calibrate.

Disadvantages
It is accurate for low stress only (< 200 MPa). Tensor symmetry is broken for nonbulk
conditions as in an inversion layer, which is not reflected by the model. The default
piezoresistive coefficients are not accurate for inversion layers.

Advanced Calibration for Device Simulation User Guide 52

V-2024.03
Chapter 4: Descriptions of Models
CMOS Devices

Second-Order Piezoresistance Mobility Model

This model uses the 6th-grade piezoresistance tensor with constant coefficients to calculate
the stress-induced mobility change.

Options
The model can be used as a factor (isotropic) model or a tensor (anisotropic) model. Minority
and majority charge carrier transport can be distinguished. You can include dependency on
the doping concentration. Calibration of dependency of the influence of mechanical stress
on the saturation velocity is possible.

Advantages
Parameters are available for the bulk case. The model is fully transformable, so it can
consider all stress configurations and current directions in a bulk situation.

Disadvantages
It is accurate for low and moderate stress values only (< 500 MPa). Tensor symmetry is
broken for nonbulk conditions as in an inversion layer, which is not reflected by the model.
The default piezoresistive coefficients are not accurate for inversion layers.

Nonlinear Piezoresistance Models for Electrons and Holes (MCmob

and SBmob)
These models use nonlinear piezoresistive coefficients and higher-order cross-term
corrections that are extracted from Monte Carlo simulation, Sentaurus Band Structure
simulation, or other sources (experiments and simulations available from the literature).

Options
Minority and majority charge carrier transport can be distinguished. You can include
dependency on the doping concentration. Calibration of dependency of the influence of
mechanical stress on the saturation velocity is possible. Germanium mole fraction
dependency of the model coefficients can be switched on. Parameters for different
orientations can be used and an auto-orientation option is available. You can load complete
parameter sets using the Sentaurus Device parameter file.

Advantages
Parameters are available for strong inversion and for different channel materials (silicon and
SiGe). The model can be extended to other conditions when there are reference tools such
as Monte Carlo simulations or experiments. The model is available for different channel and
wafer orientations. It is fast and can be calibrated easily.

Advanced Calibration for Device Simulation User Guide 53

V-2024.03
Chapter 4: Descriptions of Models
CMOS Devices

Disadvantages
Interpolation especially for untypical stress conditions must always be validated. The
physics is in the reference tool, for example, the Monte Carlo simulator. Improving the model
without improving the reference tool or without obtaining better experiments is not possible,
except with respect to the interpolation. The settings for the local coordinate system must be
specified in the Sentaurus Device parameter file.

Intel Stress-Induced Hole Mobility Model (hSixBand)

This model uses the change of band structure and occupation to calculate stress-dependent
mobility.

Options
You can switch on dependency on the doping or carrier concentration.

Advantages
The model is derived from physical assumptions.

Disadvantages
Calibration is not straightforward. Only silicon as a channel material is available. The
influence of surface orientation cannot be described. The model has been developed and
calibrated for uniaxial stress in the <110> direction. The combination of the component
normal to the MIS plane with the in-plane stress components is not validated.

eSubband Model for Electrons

This model uses the change of band structure and occupation to calculate the mobility
change.

Options
You can switch on dependency on the doping or carrier concentration. Stress-related
change in the scattering can be included. Change of effective mass with stress can be
added to the model. For the inversion layer, the MLDA option can be used.

Advantages
The MLDA option allows you to move from bulk to inversion layer conditions, and to take
surface or channel orientation dependency into account. The model is derived from physical
assumptions. The model is validated and calibrated for silicon and SiGe for different channel
and surface orientations. Mole fraction dependency for SiGe is available.

Advanced Calibration for Device Simulation User Guide 54

V-2024.03
Chapter 4: Descriptions of Models
CMOS Devices

Disadvantages
Calibration is not straightforward.

hSubband Model for Holes

This model uses the change of band structure and occupation to calculate stress-dependent
mobility.

Options
You can switch on dependency on the doping or carrier concentration. Stress-related
change in the scattering can be included. Change of effective mass with stress can be
added to the model. For the inversion layer, the MLDA option can be used.

Advantages
The MLDA option allows you to move from bulk to inversion layer conditions, and to take
surface or channel orientation dependency into account. The model is derived from physical
assumptions. The model is validated and calibrated for silicon and SiGe for different channel
and surface orientations. Mole fraction dependency for SiGe is available.

Disadvantages
Calibration is not straightforward.

Unstrained High-Field Mobility

The low-field mobility models are valid for small driving forces only. To take into account
high-field effects, the additional high-field saturation model is introduced in the drift-diffusion
model framework.
Note:
Calibrated mole fraction–dependent parameters are available and extensive
experience exists. Numeric problems are known or are already solved.

The high-field saturation model should limit the velocity-field relation to the saturation
velocity of the material when there is sufficient scattering, and it must provide a way to
calibrate the influence of ballistic and quasiballistic transport on the charge carrier transport.
In addition, it should allow you to calibrate the germanium mole fraction dependency of the
model parameters. Furthermore, because of deficiencies in the drift-diffusion model to
describe the current response close to equilibrium, a model is needed that allows you to
compensate or, at least, to calibrate for this deficiency.

Advanced Calibration for Device Simulation User Guide 55

V-2024.03
Chapter 4: Descriptions of Models
CMOS Devices

In summary, a drift-diffusion or QDD high-field mobility model for the simulation of SiON/
SiO2 or high-k gate stack FETs without mechanical stress should have the following
features:
• Limiting the drift velocity to the saturation velocity for cases with sufficient scattering
• Possibility to calibrate the model with respect to ballistic and quasiballistic transport
• Germanium mole fraction–dependent model parameters
• Possibility to calibrate the model with respect to current response close to equilibrium
• Orientation-dependent model parameters
Sentaurus Device offers the following high-field mobility models for MISFET simulations:
• Extended Canali model is the main model for drift-diffusion simulation and can also be
used with the hydrodynamic model. It has calibrated parameters for silicon and SiGe for
situations where ballistic or quasiballistic transport is not present. Calibration to
short-channel devices where quasiballistic transport occurs can be done using model
parameters:
Physics { Mobility (HighfieldSaturation) }

• Transferred electron model is specially designed for III–V arsenides and III–V nitrides
(GaAs, GaN, and so on):
Physics { Mobility (HighfieldSaturation
(TransferredElectronEffect)) }

• Different driving forces can be used to improve convergence or to include nonlocal

properties using the hydrodynamic model:
Physics { Mobility (HighfieldSaturation (CarrierTempDriveBasic)) }

Physics { Mobility (HighfieldSaturation (CarrierTempDriveME)) }

The common model for drift-diffusion or QDD simulations is the extended Canali model.

Strained High-Field Mobility

In the QDD model, the dependency of high-field mobility on mechanical stress is determined
by the strain dependency of the saturation velocity. The saturation velocity covers both the
strain dependency of the physical saturation velocity and the strain dependency of the
quasiballistic transport. The model is again the Canali model where the parameter
SaturationFactor controls the stress dependency.

Advanced Calibration for Device Simulation User Guide 56

V-2024.03
Chapter 4: Descriptions of Models
Gallium Nitride HEMTs

Ballistic Mobility
For short-channel transistors with a few scattering events, ballistic transport is important.
The following ballistic mobility models in Sentaurus Device map ballistic transport to a
mobility and, finally, to the channel resistance:
• Simple channel length–dependent ballistic mobility model
This model depends on the gate length. Therefore, you must specify the gate length. It
adds an additional term to the low-field mobility by using Matthiessen’s rule, which is
essentially constant and contains only one fitting parameter that represents the
source-side injection velocity. Injection from the drain side is neglected.
• Injection velocity–dependent ballistic mobility model
This model depends on the gate length. Therefore, you must specify the gate length. It
adds an additional term to the low-field mobility by using Matthiessen’s rule, which is
essentially constant and contains only one fitting parameter that represents the
source-side injection velocity. Injection from the drain side is considered empirically and
is especially important at small drain bias voltage.
• Channel length–dependent model
This model also depends on the gate length, but it is used in the framework of the kinetic
velocity model as an option.
• Kinetic velocity model
The kinetic velocity model (KVM) is a local model that does not need the explicit
specification of a gate length. It includes the thermionic emission of charge carriers from
the source side and their acceleration by the electric field. This is the best and most
appropriate ballistic mobility model in Sentaurus Device.
• Multivalley ballistic mobility model
This model in the framework of the KVM includes the influence of stress on ballistic
mobility.

Gallium Nitride HEMTs

GaN HEMTs have unique qualities that require special attention. First, there is the large
band gap that results in very low charge-carrier densities and numeric problems. Second, in
the on-state, the current conduction is performed using a 2D electron gas channel in the
GaN layer. This has implications for the modeling of mobility. Third, most of these structures
use material combinations in addition to pure GaN and AlN, such as AlGaN. The mole
fraction of these alloys determines the thermal conductivity, the mobility, and the band gap.
For that reason, you need mole fraction–dependent parameters that make the transition

Advanced Calibration for Device Simulation User Guide 57

V-2024.03
Chapter 4: Descriptions of Models
Silicon Carbide Devices

from GaN to AlN. In general, linear or parabolic interpolation between GaN and AlN is not
sufficient because of the alloy influence. Furthermore, there is the influence of strain causing
polarization in the device. In addition, traps strongly influence the device operation, and the
saturation velocity behavior differs from silicon.

Silicon Carbide Devices

As another wide-bandgap material, SiC shows some differences to silicon as well. There are
numeric problems connected with the small densities caused by the large band gap. The
anisotropy of the mobility and of the avalanche coefficients is difficult to handle numerically.
Traps, contact behavior, incomplete ionization, and tunneling play more critical roles in SiC
devices than in silicon devices. You must include the incomplete ionization
(IncompleteIonization) model in the simulation because of the wide band gap of SiC.
Anisotropic behavior can be considered for thermal conductivity, mobility, and impact
ionization.

Silicon Smart-Power Devices

Model requirements are similar to those for silicon deep-submicron devices (see Path A:
Basic CMOS Simulation Command File Sections on page 24) fabricated by gate-first
processes, but with a greater variety of process and doping conditions and layout
specifications. In addition, there is a need for:
• High accuracy over a wide temperature range up to very high temperatures
• Good modeling capabilities for avalanche and breakdown
• Good modeling capabilities for electrostatic discharge (ESD)
For avalanche generation, the UniBo2 model is calibrated for a temperature range from
300 K to 773 K and is best suited for ESD and high-temperature simulations. It also is
recommended for room-temperature breakdown simulations.

Advanced Calibration for Device Simulation User Guide 58

V-2024.03
5
Quality of Fitting and Extraction
5

This chapter discusses fitting quality and extraction issues.

Low-Field Mobility of Planar CMOS Devices With Silicon Channel

This section discusses the requirements and the quality of the extraction of the parameters
for the low-field mobility of planar CMOS devices.

Oxide–Silicon Interface
This section briefly explains how the low-field mobility parameter sets have been calibrated.
Data from Takagi et al. [1][2], Nakamura et al. [3], and Nayfeh et al. [4] has been used to
calibrate mobility in the transistor channel. All data is from pure polysilicon/SiO2 gate stacks
with oxide thicknesses of 25 nm for Takagi, 2 nm for Nakamura, and 5 nm for Nayfeh.
This data does not have high-k gate stacks and, therefore, allows you to extract the model
parameters of surface phonon, surface roughness, and Coulomb scattering at ionized
impurities. On the other hand, process-induced variations in surface roughness and the
material composition of the first atomic layers at the isolator–silicon interface can influence
mobility and result in different mobilities for different gate-formation process steps. These
differences cannot be reflected by a single parameter set. Therefore, the parameter set
presented here is a starting point for calibration.
Usually, the extraction of low-field mobility as a function of the effective electric field Eeff
begins with measuring the drain current in the linear regime and, from that, the mobility is
calculated for each gate voltage. At the same time, the inversion charge per area Ninv is
calculated from C–V characteristics. Because Eeff is not measured, it must be calculated
from Ninv using Eeff = Eeff (Ndepl, Ninv, ). Mostly, the depletion approximation with Ndepl as
the depletion charge is used for this, where the doping is extracted from other experiments
and is assumed to be constant. The result is the dependency µ = µ(Eeff). To obtain the
universal behavior, that is, the doping-independent behavior of the mobility for strong
inversion, the fitting parameter  is introduced in the equation used to determine the
effective field. You must use different values for  to obtain the universal behavior for

Advanced Calibration for Device Simulation User Guide 59

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Low-Field Mobility of Planar CMOS Devices With Silicon Channel

different charge carriers and surface orientations (electrons: 1/2 for (100) and 1/3 for (110),
holes: 1/3).
These standard parameters provide mostly good universal behavior for a (100) surface
orientation and low-to-moderate doping. For a (110) surface orientation, high doping, or
nonplanar and double-gate devices, the universal behavior is mostly not fully achieved. In
general, it remains a question of whether the use of this extraction method, and especially
the use of the same  in the extraction from measurement and simulation, is correct.
If  = 1, then Eeff corresponds exactly to the electric field at the interface in the depletion
approximation. However,  < 1 lowers the field strength, thereby taking into account that the
current flow not only is at the surface but also is distributed a few nanometers into the depth
of the silicon. Because it is unclear whether the field and current density distribution and the
local dependency of mobility on the electric field are described sufficiently in the simulation,
it might be difficult to achieve universal behavior in simulations with the same  parameter
as in measurement.
Often, this universality is achieved only by introducing an artificial doping dependency for
surface phonon scattering (SPS) and surface roughness scattering (SRS) in the Lombardi
terms for the mobility models. However, it is unclear whether this doping dependency exists
at all, or whether it is only a result of this forced parameter fitting. It can result in an unusual
situation where the mobility degradation in intrinsic or very low–doped regions is actually
stronger than in high-doped regions and drops to completely unrealistic values.
Because of this, another extraction methodology is used for the mobility parameters. The
mobility model is calibrated to the measured µ = µ(Ninv) curves. For this, no calibration of 
is necessary, and universal behavior is not requested. After this, a check is performed with
µ = µ(Eeff) but additional calibration is performed only if there are strong disagreements.
Figure 8 to Figure 16 show the agreement between measurement and simulation.
When simulating CMOS devices with precalibrated parameter sets, you must always take
into account that these parameter sets are initial parameter sets that can undergo additional
fine-tuning and major changes. The reasons are mainly connected to the dependency of
mobility degradation on the processing.
For example, the interface or surface roughness depends on the process conditions and the
materials involved. Applying the parameters already extracted to modern FinFET devices
with a <110> channel direction often underestimates the inversion mobility, because the
surface roughness–related mobility degradation is smaller on the (110) fin sidewalls than in
the planar (110) case [5].
Similarly, for surface phonon–related interface mobility, the process conditions and the
materials involved introduce some uncertainties that do not allow you to derive a parameter
set that is valid for several device types or technology generations. Recalibration is usually
required.
The strength of remote Coulomb scattering (RCS) or remote dipole scattering (RDS)
depends on the number of charges in the high-k gate stack and their position. The model

Advanced Calibration for Device Simulation User Guide 60

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Low-Field Mobility of Planar CMOS Devices With Silicon Channel

must be calibrated to measurements when switching it on. RCS and RDS have the same
origin: the mobility degradation by charges in the high-k gate stack. The difference is in the
presence of both negative and positive charges in the case of RDS. For remote phonon
scattering (RPS), the parameters are extracted from Monte Carlo simulation. Depending on
the properties of the high-k gate stack, fine-tuning might be necessary. However, because it
is difficult to extract the correct parameters without having mobility measurements for
different interfacial layer thicknesses and different temperatures, you should keep the
parameters for RPS and focus on the calibration of RCS and RDS.
The Philips unified mobility (PhuMob) model is calibrated to bulk mobility measurements.
Usually, fine-tuning is necessary only when performing a custom calibration with
high-accuracy requirements. In addition, mobility fine-tuning cannot be separated from the
calibration of the doping concentration or dopant activation. Therefore, you should keep the
default parameters.
In the framework of the inversion and accumulation layer mobility (IALMob) model, when
going to inversion layer conditions, 3D Coulomb scattering turns to 2D Coulomb scattering,
the PhuMob model is switched off, and the 2D Coulomb scattering model takes over. The 2D
Coulomb scattering model is a very simple description of reality with many fitting
parameters. This model mainly influences the onset of the I–V curves. The default
parameters are not very good, and additional calibration is necessary.
The stress dependency of mobility resulting from hSubband (holes), eSubband (electrons),
MCmob, and SBmob simulations has been compared to the literature and in-house Monte
Carlo (Sentaurus Device Monte Carlo) and Kubo–Greenwood (Sentaurus Band Structure)
simulations [6][7][8].

Advanced Calibration for Device Simulation User Guide 61

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Low-Field Mobility of Planar CMOS Devices With Silicon Channel

Figure 8 Electron mobility versus areal inversion density for (100) surface in linear scale, with
(top to bottom) channel doping of 3.9e15, 2e16, 7.2e16, 3e17, 7.7e17, and 2.4e18

800

600
Mobility [cm2/V/s]

400

200

1.0e+13 2.0e+13 3.0e+13

Areal Inversion Density [cm−2]

Figure 9 Electron mobility versus areal inversion density for (100) surface in logarithmic
scale with (top to bottom) channel doping of 3.9e15, 2e16, 7.2e16, 3e17, 7.7e17,
and 2.4e18
Mobility [cm2/V/s]

102

101
1.0e+11 1.0e+12 1.0e+13
Areal Inversion Density [cm−2]

Advanced Calibration for Device Simulation User Guide 62

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Low-Field Mobility of Planar CMOS Devices With Silicon Channel

Figure 10 Hole mobility versus areal inversion density for (100) surface in linear scale with (top
to bottom) channel doping of 7.8e15, 1.6e16, 5.1e16, 2.7e17, and 6.6e17

200
Mobility [cm2/V/s]

100

1.0e+13 2.0e+13 3.0e+13

Areal Inversion Density [cm−2]

Figure 11 Hole mobility versus areal inversion density for (100) surface in logarithmic scale
with (top to bottom) channel doping of 7.8e15, 1.6e16, 5.1e16, 2.7e17, and 6.6e17
Mobility [cm2/V/s]

102

1.0e+11 1.0e+12 1.0e+13

Areal Inversion Density [cm−2]

Advanced Calibration for Device Simulation User Guide 63

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Low-Field Mobility of Planar CMOS Devices With Silicon Channel

Figure 12 Electron mobility versus effective field for (100) surface in linear scale with (top to
bottom) channel doping of 3.9e15, 2e16, 7.2e16, 3e17, 7.7e17, and 2.4e18

800

600
Mobility [cm2/V/s]

400

200

1.0e+06 2.0e+06
Effective Field [V/cm]

Figure 13 Electron mobility versus effective field for (100) surface in logarithmic scale with (top
to bottom) channel doping of 3.9e15, 2e16, 7.2e16, 3e17, 7.7e17, and 2.4e18
Mobility [cm2/V/s]

102
1.0e+05 1.0e+06
Effective Field [V/cm]

Advanced Calibration for Device Simulation User Guide 64

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Low-Field Mobility of Planar CMOS Devices With Silicon Channel

Figure 14 Electron mobility versus areal inversion density for (100) surface as a reference and
for (110)/<100> and (110)/<110> in linear scale, with channel doping between 5e17
and 9e17

600
Mobility [cm2/V/s]

400

200

1.0e+13 2.0e+13 3.0e+13

Areal Inversion Density [cm−2]

Figure 15 Hole mobility versus areal inversion density for (100) surface as a reference and for
(110)/<100> and (110)/<110> in linear scale, with channel doping between 4.5e17
and 6.6e17
200
Mobility [cm2/V/s]

150

100

50

1.0e+13 2.0e+13
Areal Inversion Density [cm−2]

Advanced Calibration for Device Simulation User Guide 65

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Low-Field Mobility of Planar CMOS Devices With Silicon Channel

Figure 16 Hole mobility versus effective field and areal inversion density for (100) surface as
a reference and for (110)/<100> and (110)/<110> in double logarithmic scale, with
channel doping between 4.5e17 and 6.6e17

Mobility [cm2/V/s] 103

102

101
105 106 107 108 109 1010 1011 1012 1013 1014
−2
Effective Field and Areal Inversion Density [cm ]

Influence of Metal Gate on Inversion Layer Mobility

The parameter set used to produce the results shown in Figure 8 to Figure 16 is derived
from measurements on devices with a polysilicon/oxide gate stack. In devices with a metal/
oxide gate stack, the channel mobility can differ because of the influence of the metal gate.
The same applies to the comparison between polysilicon/HfO2 and metal/HfO2 gate stacks.
Usually, the transition from the polysilicon gate to the metal gate results in improved mobility.
One reason for this could be the better screening of surface phonons and charges in the
gate stack by the charge carriers in the metal gate [9].
Table 11 Influence of different gate materials on inversion layer mobility for electrons [9]

Gate stack Maximum mobility [cm2/V/s] Mobility at Eeff = 1 MV/cm [cm2/V/s]

HfO2/Polysilicon 175 145

HfO2/TiN 245 200

Advanced Calibration for Device Simulation User Guide 66

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Low-Field Mobility of Planar CMOS Devices With Silicon Channel

The increasing mobility caused by the metal gate can be reflected by an increase of the
IALMob parameters B, C, and delta by a factor of 1.4.

Influence of High-k Gate Stack on Inversion Layer Mobility

The parameter set used to produce the results shown in Figure 8 to Figure 16 is derived
from measurements on devices with a polysilicon/oxide gate stack. In devices with a high-k
isolation (for example, with HfO2), the channel mobility can degrade because of the
influence of charges in the gate isolator (RCS or RDS) and RPS.
Often, the charges are arranged in dipole-like configurations. The RCS-related mobility
degradation depends on the number and the arrangement of charges in the gate isolator,
especially on their distance to the inversion layer. RCS depends greatly on the process
conditions and the materials used in the gate stack, and so the variation between
technologies and process splits and even over the wafer and between wafers can be large.
Therefore, it is not possible to provide well-calibrated default parameters, and custom
calibration to the process and device under consideration is usually required. However, the
model should describe the major tendencies such as dependency on the thickness of the
interfacial oxide or on the concentration of charges in the gate stack.
Table 12 and Table 13 list model parameters for the RPS model and for the
InterfaceCharge models that fulfill these requirements.
Table 12 Recommended parameters and values for RPS model

Parameter Value for electrons Value for holes Comments

d_crit 1.0e-7 cm 1.0e-7 cm Distance between high-k interface and

silicon interface

l_crit 7.7e-8 cm 7.7e-8 cm [10]

murps0 124 cm2/(Vs) 124 cm2/(Vs) Extracted from Sentaurus Device Monte
Carlo simulation

Table 13 Recommended parameters and values for InterfaceCharge models (RCS and
RDS, NegInterfaceCharge and PosInterfaceCharge models use the same
parameters)

Parameter Value for electrons Value for holes Comments

c_exp 0.5 0.5

E0 1.0e-6 1.0e-6

Advanced Calibration for Device Simulation User Guide 67

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Low-Field Mobility of Planar CMOS Devices With Silicon Channel

Table 13 Recommended parameters and values for InterfaceCharge models (RCS and
RDS, NegInterfaceCharge and PosInterfaceCharge models use the same
parameters) (Continued)

Parameter Value for electrons Value for holes Comments

l_crit 7.7e-8 7.7e-8 [10]

mu1 250 250

Extraction of Contact Resistance Parameters

Stavitski et al. [11] measured the contact resistance between PtSi and NiSi, and silicon.
Based on this data, the parameters for the Schottky resistance (SchottkyResist) model in
Sentaurus Device have been extracted (see Table 14).
Note:
No additional tests or validations have been performed with this parameter set. In
addition, the parameters are not included in the parameter files in the MaterialDB
folder.

Table 14 Parameters for the SchottkyResist model for PtSi and NiSi

Parameter Value for electrons Value for holes

PtSi

barrier 0.8 eV 0.4 eV

mt 0.08 0.08

Rinf 3.0e-10 cm2 3.0e-10 cm2

NiSi

barrier 0.6 eV 0.6 eV

mt 0.08 0.08

Rinf 1.0e-9 cm2 5.0e-10 cm2

Advanced Calibration for Device Simulation User Guide 68

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Low-Field Mobility of Planar CMOS Devices With Silicon Channel

Extraction of Parameters for Mobility Models in Accumulation

Regime
In the accumulation regime, mobility is influenced by the same mobility degradation
mechanisms as in the inversion regime. However, model parameters and the importance of
single mobility degradation mechanisms can vary. Especially at higher background doping
concentrations, the screening of Coulomb scattering differs between accumulation and
inversion.
Because of the importance of the on-state and the measurement data available, the focus is
on high accumulation areal densities. Reference [12] provides some specific and well
explained measurements of mobilities in accumulation layers in planar bulk devices for
different background doping concentrations at room temperature. These measurements
have been used as a reference for the simulations with Sentaurus Device using the
inversion and accumulation layer mobility (IALMob) model.
Despite small differences between accumulation and inversion layer mobility in the on-state
claimed in [12], a calibrated parameter set is presented that improves the agreement for
electrons. The parameter set also shows how to adjust the accumulation layer model
parameters without changing much in the inversion regime. Table 15 shows the parameters
used for calibration.
Table 15 Parameters of IALMob model used for calibration (electrons)

Parameter Recommended value Parameter Recommended value

lambda 0.02 alpha_sr_D 2.0e-14

alpha_ph2d_A 0.0 lambda_sr 0.02

alpha_ph2d_D 2.0e-14 lambda_sr_A 0.0

lambda_ph2d_A 0.0 lambda_sr_D 1.0

lambda_ph2d_D 1.0 D1_acc 1200

alpha_sr_A 0.0 D2_acc 1200

Essentially with these parameter settings, the doping dependence for the inversion case
remains switched off and, when there is no accumulation doping, there is no change in the
results. However, with alpha_ph2d_D or alpha_sr_D, and lambda or lambda_sr larger than
zero, the doping dependence for the accumulation regime is switched on. In this way, you
can separate the accumulation and inversion regimes.
The parameter set in Table 15 gives reasonable agreement to the measurements in [12]
(see Figure 17).

Advanced Calibration for Device Simulation User Guide 69

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Carbide Devices

Figure 17 Mobility versus effective electric field for electrons and different donor doping
concentrations; red lines: simulation and black lines: measurements from [12]
2000
1800 measurement, doping=1e16
simulation, doping=1e16
1600 measurement, doping=4.8e16
Mobility [cm2/V/s]

1400 simulation, doping=4.8e16

1200 measurement, doping=7.5e16
simulation, doping=7.5e16
1000 measurement, doping=1.5e17
800 simulation, doping=1.5e17
600
400
200
0
0 500000 1000000 1500000
Effective Electric Field [V/cm]

Basic Properties of Silicon Carbide Devices

Parameter values and corresponding references can be found in the 4H-SiC.par and
6H-SiC.par files and in [13].

All silicon carbide (SiC) polytypes are indirect semiconductors, and the maximum of the
upper valence bands is located at the center (-point) of the Brillouin zone. Conduction band
minima of 3C-SiC, 6H-SiC, and 4H-SiC are located at point X, point U on the (L-M) line, and
point M, respectively.
Photoluminescence or absorption measurements give an exciton band gap Egx. There are
no reliable measured values for binding energies Ex of free excitons, which are required to
obtain the indirect band gap Eg = Egx + Ex. Since measured values of Ex range from 10 mV
to 80 mV, a mean direct band gap can be assumed by shifting Egx by 40 mV. In the
parameter file, a value of Eg0=3.285 eV is used.
Temperature dependency of the band gap is measured for 6H-SiC in the temperature
ranges from 6 K to 200 K and from 300 K to 700 K. The dependency of 3C-SiC is measured
in the temperature range from 295 K to 700 K. The dependency of 4H-SiC is assumed to be
similar because no experimental data is available.
The effective mass of electrons and holes has been measured by optically detected by
either cyclotron resonance or Raman scattering and has been compared to theoretical
calculation. The theoretical calculation reproduced well the experimental results.
There is no published measured data of bandgap narrowing. The theoretical calculation was
first performed by Lindefelt [14] and followed by Persson et al. [15]. The former model

Advanced Calibration for Device Simulation User Guide 70

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Carbide Devices

constitutes an extension of a semiempirical method originally developed by Jain and

Roulston [16]. The latter is based on random phase approximation. Both calculations give
the analytic form of band-edge displacement depending on the doping density. The analytic
form corresponds to the JainRoulston model implemented in Sentaurus Device.

Figure 18 Results of bandgap narrowing of 4H-SiC from [14] and [15] with the Slotboom model
implemented in Sentaurus Device with calibrated coefficients
700
Lindefelt n-type
600 Lindefelt p-type
Bandgap Narrowing [meV]

Slotboom n-type
500 Slotboom p-type
Persson n-type
400
Persson p-type
300

200

100

0
1017 1018 1019 1020 1021
Ionized Impurity Concentration [cm−3]

Electron affinity is calculated assuming the linear dependency of hexagonality together with
the measured band offset of the conduction band for 3C-SiC or 6H-SiC, and the calculated
result of the band offset of the valence band.
Parameter sets for the Hatakeyama model [17] and the Okuto–Crowell impact ionization
model are implemented in the parameter files (see Content of Parameter Files on page 11).
The parameter set for the Okuto–Crowell impact ionization model along the c-axis
corresponds to the one reported by Niwa et al. [18].
Figure 19 and Figure 20 show the comparison of the impact ionization coefficients of
different references.

Advanced Calibration for Device Simulation User Guide 71

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Carbide Devices

Figure 19 Hole impact ionization coefficient along the c-axis of 4H-SiC [17][18][19]
1.00e+05

Impact Ionization Coefficient [cm−1]

1.00e+04

Konstantinov
Hatakeyama
1.00e+03 Niwa

1.00e+02

1.00e+01
3.0e-07 5.0e-07 7.0e-07 9.0e-07
Inverse of Electric Field [cm/V]

Figure 20 Electron impact ionization coefficient along the c-axis of 4H-SiC [17][18][19]
1.00e+05
Impact Ionization Coefficient [cm−1]

1.00e+04
Konstantinov
Hatakeyama
Niwa
1.00e+03

1.00e+02

1.00e+01
3.0e-07 4.0e-07 5.0e-07 6.0e-07 7.0e-07 8.0e-07 9.0e-07
Inverse of Electric Field [cm/V]

Temperature dependency of thermal conductivity of 4H-SiC and 6H-SiC perpendicular to the

c-axis has been reported. At low temperatures, T3 dependency and T2 dependency are
observed for 4H-SiC and 6H-SiC, respectively. T2 dependency is attributed to the scattering
of phonons by electrons in the impurity band. Thermal conductivity in 6H-SiC single crystals
decreases with increasing electron concentration, and the slope at low temperature can be
represented by the T2 law, which confirms the dominance of electron scattering.
Temperature dependency of thermal conductivity using data from [20], [21], [22], and [23] as

Advanced Calibration for Device Simulation User Guide 72

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Carbide Devices

cited in [24] is implemented in the parameter files (see Content of Parameter Files on
page 11). Experimental data is that in the direction perpendicular to the c-axis.
Parameter sets for the Masetti and Arora doping-dependent mobility models are
implemented in the parameter files (see Content of Parameter Files on page 11).
Figure 21 and Figure 22 show a comparison with different literature data.

Figure 21 Doping-dependent electron mobility at 500 K for n-doped 4H-SiC [25][26][27]

300

250 Schaffer [25]

Low-Field Mobility [cm2/Vs]

Masetti
200 Arora
Arora2
150 Roschke [26]
Kagamihara [27]
100

50

0
1015 1016 1017 1018 1019 1020 1021
−3
Concentration [cm ]

Figure 22 Doping-dependent electron mobility at 300 K for n-doped 4H-SiC [25][26][27]

1000
900 Schaffer [25]
Low-Field Mobility [cm2/Vs]

800 Masetti
700 Arora
600 Arora2
500 Roschke [26]
400 Kagamihara [27]
300
200
100
0
1015 1016 1017 1018 1019 1020 1021
−3
Concentration [cm ]

Advanced Calibration for Device Simulation User Guide 73

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Carbide Devices

For high-field saturation, all measured data refers to the current flow perpendicular to the
c-axis. Parameters for the Canali model are extracted and implemented in the parameter
file.

Figure 23 Drift velocity of electrons in 4H-SiC, parallel to basal plane (0001) [28][29]
108
Electron Velocity [cm/s]

107

106 Khan 296 K [28]

Khan 593 K [28]
Lades 296 K [29]
Lades 573 K [29]
5
10
103 104 105 106
Electric Field [V/cm]

A nitrogen atom residing on the C lattice site becomes the shallow donor in SiC. A
phosphorus atom also is used as a shallow donor (like nitrogen) and has approximately the
same ionization energy Ec – 93 meV at the cubic site and Ec – 53 meV at the hexagonal site.
However, unlike nitrogen, P mainly substitutes at a silicon site, and less than 10% of
substitutional P is located at C sites.
Phosphorus has a slightly larger ionization energy at C sites than at Si sites. The ionization
energy of N and P donors decreases with increasing donor concentration. Experimental
data for cubic and hexagonal N donors in 6H-SiC is available. For N donors in 4H-SiC,
experimental data of donor concentration dependency and fitting with empirically formula
has been reported.
Aluminum and boron residing on Si lattice sites become acceptors in SiC. For Al acceptors
in 4H-SiC, experimental data of acceptor doping concentration dependency and fitting with
empirical formulas have been reported [30][31][32].
Figure 24 to Figure 28 show the experimental data and the fit implemented in the SiC.par
parameter file.

Advanced Calibration for Device Simulation User Guide 74

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Carbide Devices

Figure 24 Ionization energy for aluminum-doped 4H-SiC with extracted fit function [30][31][32]
300

Ionization Energy [meV] 250

200

150

100 09Koizumi [30]

05Pernot [31]
50
97Troffer [32]

0
1015 1016 1017 1018 1019 1020 1021
NAl [cm−3]

Figure 25 Ionization energy for nitrogen-doped 4H-SiC (hexagonal site) with extracted fit
function [27]
100
90 04Kagamihara h-site [27]
80
Ionization Energy [meV]

70
60
50
40
30
20
10
0
1015 1016 1017 1018 1019 1020
−3
NN [cm ]

Advanced Calibration for Device Simulation User Guide 75

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Carbide Devices

Figure 26 Ionization energy for nitrogen-doped 4H-SiC (cubic site) with extracted fit
function [27]
140
04Kagamihara k-site [27]
120
Ionization Energy [meV]

100

80

60

40

20

0
1015 1016 1017 1018 1019 1020
NN [cm−3]

Figure 27 Ionization energy for phosphorus-doped 4H-SiC with extracted fit function
120

100
Donor Ionization Energy [meV]

80 hexagonal
hexagonal
cubic
60
cubic

40

20

0
1.e+15 1.e+16 1.e+17 1.e+18 1.e+19 1.e+20 1.e+21
−3
Donor Concentration [cm ]

Advanced Calibration for Device Simulation User Guide 76

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Carbide Devices

Figure 28 Ionization energy for nitrogen-doped 6H-SiC with extracted fit function

hexagonal
Ionization Energy [meV] 150 hexagonal
cubic
cubic

100

50

0
1016 1017 1018 1019 1020 1021
−3
NN [cm ]

The inversion layer mobility parameters have been calibrated for 4H-SiC interfaces to oxide
for different doping and trap concentrations as well as gate oxide formation processes.
Experimental data is taken from [33], [34], and [35]. Surface phonon, surface roughness,
and 2D Coulomb scattering parameters have been adjusted. The parameter files
4H-SiC.par, 4HSiC.par, SiC_4H.par, and SiliconCarbide.par include the calibrated
parameters. Figure 29 shows the comparison between experimental data and simulation
results.

Figure 29 Simulated and measured [33] inversion layer mobility

3.00e+02
Measured
2.50e+02 Simulated

2.00e+02
Mobility [cm2/V/s]

1.50e+02

1.00e+02

5.00e+01

0.00e+00
0.0e+00 2.0e-01 4.0e-01 6.0e-01 8.0e-01 1.0e+00
Effective Field [MV/cm]

Advanced Calibration for Device Simulation User Guide 77

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Gallium Nitride Devices

In the calibration of the inversion layer mobility, the incomplete ionization model [36] has
been included to achieve better results, especially at lower temperatures, where a large part
of the dopants is not activated. To describe the temperature dependence of the mobility, the
k parameter in the phonon scattering term, in the IALMob model, has been adjusted.

Figure 30 shows the comparison between experimental data [37] and simulation results.
The measured mobility values over temperature have been normalized to the measured
values at 300 K in the set of parameter files used in Figure 29 to compensate for fluctuations
between different sets of measurements due to different trap distributions.

Figure 30 Simulated and measured [37] inversion layer mobility at different temperatures and
low doping
300

280

260 Measurement, normalized

Measurement
Mobility [cm2/V/s]

240
Simulation
220

200

180

160

140

120

100
200 220 240 260 280 300 320 340 360
Temperature [K]

Basic Properties of Gallium Nitride Devices

The inversion layer mobility parameters have been calibrated for GaN to AlGaN or AlN
interfaces.
Experimental data is taken from [38] and [39]. Surface phonon, surface roughness, and 2D
Coulomb scattering parameters have been adjusted. The parameter file GaN.par includes
the calibrated parameters. Figure 31 shows the comparison between experimental data and
simulation results.
In the IALMob parameter set, the s parameter that controls the quantization is set to a higher
value, assuming 2D Coulomb scattering dominates in the channel. Therefore, the transition
to bulk Coulomb scattering is controlled by lcrit_c only.

Advanced Calibration for Device Simulation User Guide 78

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

Figure 31 Simulated and measured [38][39] inversion layer mobility

Measurement: A
2000
Measurement: B
1800
Measurement: C
1600 Waller [38]
Mobility [cm2/V/s]

1400 Chang [39]

1200

1000

800

600

400

200

0
0.00E+00 5.00E+12 1.00E+13 1.50E+13
−2
Inversion Layer Density [cm ]

Basic Properties of Indium Gallium Arsenide Devices

Parameter values and references can be found in the InGaAs.par file. The following
sections briefly show the results of parameter calibration and review various aspects of the
calibration.

Permittivity
The permittivity values for GaAs and InAs materials have been reviewed following the data
reported by Adachi [40]. Since there is no experimental data for semiconductor alloys, a
linear interpolation scheme has been adopted for indium gallium arsenide (In1–xGaxAs)
alloys.

Lattice Heat Capacity

Similarly, a linear interpolation scheme for heat capacity is used for In1–xGaxAs alloys, in
which the heat capacity values for GaAs and InAs at 300 K are updated with the ones
reported by Adachi [41].

Advanced Calibration for Device Simulation User Guide 79

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

Thermal Conductivity
The mole fraction dependency of thermal conductivity for In1–xGaxAs alloys has been
measured by Abrahams et al. [42] and Arasly et al. [43] at 300 K. Due to a random
distribution of Ga and In atoms in the sub-lattice sites, the thermal conductivity values for
In1–xGaxAs alloys tend to be smaller than for the corresponding binary materials. Figure 32
shows the thermal conductivity model and comparison to experimental data.

Figure 32 Thermal conductivity versus Ga mole fraction for In1–xGaxAs alloys at 300 K;
experimental data is taken from [42] and [43], and binary data is taken from [44]
0.5
Experimental data
Thermal Conductivity [W/cmK]

Interpolation in MaterialDB
0.4

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1
x-Ga

Band Gap and Electron Affinity

InGaAs ternary materials are direct bandgap materials over the entire composition range,
and both the maxima of the upper valence bands and the minima of the conduction band are
located at the -point of the Brillouin zone. Photoluminescence and photoreflectance
measurements are often used to determine the bandgap energies of In1–xGaxAs [45][46]
[47]. In the case of the lattice-matched alloy to InP, In0.53Ga0.47As, which has been
investigated extensively, the bandgap energy is generally recognized to be 0.816 eV at 0 K.
Therefore, a temperature-independent bowing parameter of 0.477 eV is suggested [48]. As
for the Varshni parameters, the values for In1–xGaxAs alloys are determined through a linear
interpolation based on the ones from GaAs and InAs. Figure 33 shows the composition
dependency of bandgap energy at 300 K, plotted with the new set of the bowing parameter
and Varshni parameters. The revised bandgap model fits well to the experimental data.
The electron affinity values for the GaAs and InAs binary materials are taken from Adachi
[40] and are further derived to obtain the values at 0 K using the revised Varshni parameters.
Since little experimental data is available, linearly interpolated electron affinity values are
assumed for In1–xGaxAs alloys in the parameter file.

Advanced Calibration for Device Simulation User Guide 80

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

Figure 33 Band gap versus Ga mole fraction for In1–xGaxAs alloys at 300 K; experimental data
is taken from [45], [46], and [47]; binary data is taken from [48]
1.5
Band Gap [eV]

1.0

0.5
Experimental data
Vurgaftman
Interpolation in MaterialDB
0.0
0 0.2 0.4 0.6 0.8 1
x-Ga

Density-of-States
The electron effective masses of the -valley of the In1–xGaxAs alloys have been measured
extensively [41], and the results are shown in Figure 34.

Figure 34 Electron effective mass of the -valley versus Ga mole fraction for In1–xGaxAs
alloys at 300 K; experimental data is reprinted from Adachi [41]
0.08
0.07
0.06
me (Γ) / m0

0.05
0.04
0.03
0.02
Experimental data
0.01
Interpolation in MaterialDB
0
0 0.2 0.4 0.6 0.8 1
x-Ga

A simple linear interpolation scheme using the corresponding GaAs and InAs binary data
yields a good fit to the experimental data. In the InGaAs.par file, the composition

Advanced Calibration for Device Simulation User Guide 81

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

dependency of density-of-states for electrons is a set of tabulated values that are computed
from the electron effective masses at 300 K using:
3---
19 m e 2
N C  300 K  = 2.5094 10  ------ (2)
m0
Regarding the hole effective masses for In1–xGaxAs alloys, Vurgaftman et al. proposed a
bowing parameter for each Luttinger valence band parameter:  1 ,  2 , and  3 [48]. In
measurement,   x  , defined as  = – 1  3  1 + 2  3 2 +  3 – 2  3 , has been reported for
x = 0.47 and x = 0.82–0.92 [49][50][51].
Figure 35 shows that the interpolation scheme from Vurgaftman et al. agrees well with the
experimental data to some extent and is used to calculate the composition dependency of
the overall hole effective masses, mp.

Figure 35 -values versus Ga mole fraction for In1–xGaxAs alloys; experimental data is taken
from [40], [49], [50], and [51]; red line represents the interpolation function
suggested by [48]
8
7
6
5
4
κ

3
Alavi
2 Warburton
Traynor
1 Adachi
Interpolation in MaterialDB
0
0 0.2 0.4 0.6 0.8 1
x-Ga
Similarly, a set of tabulated values of density-of-states for holes at 300 K is computed using:
3---
19 m p 2
N V  300 K  = 2.5094 10  ------ (3)
 m 0
Advanced Calibration Device includes these tabulated values (see Figure 36).

Advanced Calibration for Device Simulation User Guide 82

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

Figure 36 Valence band density-of-states at 300 K versus Ga mole fraction for In1–xGaxAs
alloys; experimental data is computed based on Luttinger parameters taken from
[49]; red line is computed following the interpolation scheme suggested by [48]
1.4e+19

1.2e+19

1.0e+19
NV300 [cm−3]

8.0e+18

6.0e+18

4.0e+18

2.0e+18 Alavi
Interpolation in MaterialDB
0.0e+00
0 0.2 0.4 0.6 0.8 1
x-Ga
-

Bandgap Narrowing
Doping-dependent bandgap narrowing for GaAs, InAs, and In1–xGaxAs materials is
accessed using the Jain–Roulston model [52]. The coefficients for p-type doping ( A p , B p ,
C p , and D p ) in GaAs material are derived from [52], in which the quantities are refined with
available up-to-date basic material parameters of GaAs.
Figure 37 shows the extent of bandgap narrowing for p-GaAs obtained with the Jain–
Roulston model, and it agrees well with the experimental data measured using
photoluminescence in the carrier density range from 2.7e17 to 3.9e20 cm–3 [53]–[57]. For
p-InAs, the coefficients are retrieved in the same way, but the model validation cannot be
qualified due to a lack of experimental data (see Figure 38).
Regarding n-type GaAs and InAs, the formulas in the Jain–Roulston model cannot describe
the doping dependencies of bandgap narrowing. Therefore, the corresponding coefficients
( A n , B n , C n , and D n ) are determined preferably through curve fitting based on
experimental data. As shown in Figure 37, the bandgap narrowing for n-GaAs with fitted
Jain–Roulston coefficients agrees well with experimental data [53][58]. It should be noted
that, for n-InAs, the bandgap narrowing value reaches nearly 0.2 eV at the carrier density of
3e18 cm–3 [59], and a higher density is probable, which would alter the band structure
drastically. Therefore, a minimum band gap (EgMin) value of 0.15 eV is set in the Bandgap
model to avoid the occurrence of an unusually low (or negative) band gap.
Since there is little experimental data reported in the literature to resolve the mole fraction
dependency of bandgap narrowing for In1–xGaxAs, a simple linear interpolation scheme is
adopted for each coefficient in the Jain–Roulston model.

Advanced Calibration for Device Simulation User Guide 83

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

Figure 37 Doping-dependent bandgap narrowing for p-GaAs and n-GaAs at 300 K;

experiment data is taken from [53]–[58]; black line is computed following Jain–
Roulston model with up-to-date GaAs basic material parameters, and red line is
fitted curve based on experimental data from [53] and [58]
0.25
Experimental data (p)
Jain−Roulston (p)
Bandgap Narrowing [eV]

0.20 Experimental data (n)

Jain−Roulston (n)

0.15 GaAs

0.10

0.05

0.00
1e+17 1e+18 1e+19 1e+20 1e+21
Carrier Density [cm−3]

Figure 38 Doping-dependent bandgap narrowing for p-InAs and n-InAs at 300 K; black line is
computed following Jain–Roulston model with up-to-date InAs basic material
parameters, and red line is fitted curve based on experimental data from [59]
0.20
Bandgap Narrowing [eV]

0.15

0.10

InAs
0.05
Jain−Roulston (p)
Experimental data (n)
Jain−Roulston (n)
0.00
1e+17 1e+18 1e+19 1e+20 1e+21
Carrier Density [cm−3]

Quantization Effects
The quantization model is considered in MOS devices to obtain the correct densities and
field distributions in channels. The density gradient model in Sentaurus Device would be the
first choice for planar and nonplanar In1–xGaxAs device simulations, and then a calibration

Advanced Calibration for Device Simulation User Guide 84

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

to the solutions from the Schrödinger equations is required to extract the corresponding
quantization parameters.
The 1D Schrödinger solver in Sentaurus Band Structure provides a reference in terms of
carrier density distribution in the inversion layer, and the fitting factor () in the density
gradient model then is fitted by minimizing the root-mean-square error of the charge integral
over a range of applied gate voltages between these two approaches. The quantization
parameter depends on the carrier polarity, the lattice orientation, the material composition,
the layer thickness, the channel geometry, and the strain. The
QuantumPotentialParameters parameter set provides the fitted  values for simulations of
unstrained In1–xGaxAs alloys, for both the bulk configuration and the double-gate
configuration with channel thicknesses of 10 nm and 20 nm.
For In1–xGaxAs materials, you cannot neglect the electron occupancy in the L-valley at a
higher gate voltage. Therefore, including the multivalley band structure model in the
simulation allows you to determine the electron density-of-states more accurately.
The parameters for the MultiValley model in the InGaAs.par file with mole fraction
dependency have been reviewed following the literature [41][48] and the band-structure
calculation using the empirical pseudopotential method.

Figure 39 Inversion electron densities for a 10 nm thick double-gate (110) MOS structure with
In0.53Ga0.47As channel: red line is simulated with the density gradient model and
blue line is simulated with the multivalley and density gradient models; in both
cases, e_gamma parameters have been optimized
1.0e+13
Inversion Electron Density [cm−2]

Sentaurus Band Structure-Schrödinger

DG (e_gamma=1.67)
8.0e+12 DG/MV (e_gamma=1.78)

6.0e+12

4.0e+12

2.0e+12

0.0e+00
0.0 0.5 1.0
Gate Voltage [V]

As shown in Figure 39, the electron charge integral in a 10 nm thick In0.53Ga0.47As

double-gate structure simulated with the multivalley and density gradient models fits better
to the result obtained with the Schrödinger solver of Sentaurus Band Structure, compared to
the one with the density gradient model only. A similar conclusion can be drawn when
observing the carrier distributions at a 1 V gate voltage along the 10 nm channel, as shown
in Figure 40.

Advanced Calibration for Device Simulation User Guide 85

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

Figure 40 Carrier distributions at gate voltage of 1 V for a 10 nm thick double-gate (110) MOS
structure with In0.53Ga0.47As channel: red line is simulated with the density gradient
model and blue line is simulated with the multivalley and density gradient models;
in both cases, e_gamma parameters have been optimized
2.5e+19
Sentaurus Band Structure-Schrödinger
DG (e_gamma=1.67)
Electron Density [cm−3]

2.0e+19 DG/MV (e_gamma=1.78)

1.5e+19

1.0e+19

5.0e+18

0.0e+00
0 0.005 0.01
Distance [µm]

Figure 41 shows the tabulated values of fitted e_gamma and h_gamma parameters using the
density gradient and multivalley models, which are provided in the InGaAs.par file for bulk
(100) and (110) orientations. It is evident that there is no orientation dependency for the
electron quantization parameter, and the compositional variation is also less distinct
compared to the hole quantization parameter. The quantization parameter is also a function
of layer thickness, as shown in Figure 42.
It is important to implement the corresponding quantization parameters designated for a
specific layer thickness during simulations, since the discrepancy on the charge distribution
is very susceptible to thin layers. For example, the e_gamma value optimized for a 10 nm
channel width could still be used for quantum correction on a 20 nm thick channel, but with
a slightly larger error. However, the same value might be only appropriate to use for thin
layers down to 8 nm. If the layer thickness is less than 8 nm, you should perform a
quantization calibration again to extract the correct quantization parameter.

Advanced Calibration for Device Simulation User Guide 86

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

Figure 41 Fitted gamma values versus Ga mole fraction for bulk MOS structures with
In1–xGaxAs alloys using density gradient and multivalley models
7.0
n-Bulk100
n-Bulk110
6.0 p-Bulk100
p-Bulk110
5.0
gamma

4.0

3.0

2.0

1.0

0.0
0.0 0.2 0.4 0.6 0.8 1.0
x-Ga

Figure 42 Fitted gamma values versus Ga mole fraction for 10 nm and 20 nm double-gate
MOS structures with In1–xGaxAs alloys using density gradient and multivalley
models
7.0
n-DG110-10nm
p-DG110-10nm
6.0 n-DG110-20nm
p-DG110-20nm
5.0
gamma

4.0

3.0

2.0

1.0

0.0
0.0 0.2 0.4 0.6 0.8 1.0
x-Ga

Low-Field Bulk Mobility and Its Doping Dependency

Low-field bulk mobility is a critical parameter to determine the current–voltage
characteristics of InGaAs metal–insulator–semiconductor (MIS) devices, and it is dependent
on mole fraction, temperature, doping, inversion charge, and channel direction for
unstrained InGaAs materials.

Advanced Calibration for Device Simulation User Guide 87

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

Accurate theoretical models accounting for diverse scattering mechanisms can be complex
and might apply only to limited conditions. Therefore, empirical fitting based on available
experimental data can be an alternative.
The coefficients in the ConstantMobility and DopingDependence parameter sets of the
InGaAs.par file have been reviewed and implemented following the empirical mobility
model suggested in [60].
Reference [60] provides calibrated parameters in the empirical Caughey–Thomas mobility
model for temperature and doping dependency, for unstrained GaAs, InAs, and
In0.53Ga0.47As materials. These parameters are transformed to the parameters of the Arora
model for use in Sentaurus Device. Figure 43 demonstrates the doping dependency of
electron mobility for unstrained GaAs and In0.53Ga0.47As using the Arora model.

Figure 43 Doping-dependent low-field electron mobility for (left) GaAs and (right)
In0.53Ga0.47As at 300 K; red lines are computed using the Arora model with fitted
parameters from [60]
10000 16000
Experimental data Experimental data
Electron Mobility [cm2/Vs] 14000
Electron Mobility [cm2/Vs]

Arora model Arora model

8000
12000

6000 10000
8000
4000 6000
4000
2000
2000
0 0
1.e+13 1.e+16 1.e+19 1.e+13 1.e+16 1.e+19
Doping Concentration [cm−3] Doping Concentration [cm−3]

The doping dependency of low-field mobility for In1–xGaxAs alloys, apart from
In0.53Ga0.47As material, is usually less investigated. Therefore, a two-sectional linear
interpolation scheme is used, based on InAs, In0.53Ga0.47As, and GaAs for each parameter
in the Arora model.
Figure 44 shows the compositional dependency of the low-field electron mobility for
In1–xGaxAs alloys suggested in the InGaAs.par file with some experimental data.

Advanced Calibration for Device Simulation User Guide 88

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

Figure 44 Low-field electron mobility at 300 K versus Ga mole fraction for In1–xGaxAs alloys;
experimental data is reprinted from [61]; red line is the interpolation curve
suggested in the InGaAs.par file
40000
Experimental data
35000
Electron Mobility [cm2/Vs]
Interpolation in MaterialDB

30000
25000
20000
15000
10000
5000
0
0 0.2 0.4 0.6 0.8 1
x-Ga
-

Philips Unified Mobility Model

This model describes both majority and minority carrier mobility in bulk, taking into account
phonon scattering, impurity scattering, carrier–carrier scattering, and screening of scattering
processes. In the MaterialDB folder, the parameters for majority carrier mobility in InAs,
In0.53Ga0.47As, and GaAs materials have been calibrated without considering the minority
impurity scattering where other coefficients, such as the clustering functions and the
screening parameters, are preserved as the default ones from silicon material.
The maximum carrier mobility (µmax), the minimum carrier mobility (µmin), and the exponent
() for In1–xGaxAs alloys are transformed from the corresponding ConstantMobility and
Arora models. Then, the reference density (Nref) and the exponent () coefficients can be
fitted to the doping-dependent mobility given by the Arora model within a doping range
between 1.0e13 cm–3 and 1.0e20 cm–3 for GaAs and In0.53Ga0.47As. Since the intrinsic
density of InAs is relatively larger compared to GaAs and In0.53Ga0.47As, the fitting range for
InAs is limited to the range from 1e15 cm–3 to 1e20 cm–3.
Figure 45 presents the electron and hole mobility in In0.53Ga0.47As material at various
doping levels for the Philips unified mobility model with fitted coefficients.

Advanced Calibration for Device Simulation User Guide 89

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

Figure 45 Doping-dependent low-field mobility for (left) electron and (right) hole majority
carriers in In0.53Ga0.47As alloy at 300 K; red lines are computed with the Philips
unified mobility model with fitted parameters suggested in the InGaAs.par file
16000 400
14000
12000 300
Mobility [cm2/Vs]

Mobility [cm2/Vs]
10000
8000 200
6000
4000 100
Arora model Arora model
2000
Philips unified mobility model Philips unified mobility model
0 0
1e+13 1e+16 1e+19 1e+13 1e+16 1e+19
ND [cm−3] NA [cm−3]

Inversion Layer Mobility

This section discusses inversion layer mobility.

Inversion and Accumulation Layer Mobility Model

In a MOSFET device, the surface roughness and the surface phonon scattering can
degrade carrier mobility at the proximity of a semiconductor–insulator interface due to a high
electric field, normal to the channel interface imposed across the channel region. Since
In1–xGaxAs alloys have the potential to realize n-MOSFET devices, parameter calibration
with respect to available experimental data is required to establish a more accurate mobility
degradation model.
This section describes how the parameters in the MaterialDB folder have been calibrated
for the inversion and accumulation layer mobility (IALMob) model.
The calibration methodology involves two main steps and concerns mainly the electron
mobility in the inversion layer. First, with some initial calibrations with respect to the
experimental data to account for the process variations, the 1D mobility calculator of
Sentaurus Band Structure provides a reference in terms of mobility in the inversion layer for
the entire mole fraction range. The experimental data used in this step was reported by
Xuan et al. [62] on an inversion-type n-channel In0.53Ga0.47As MOSFET with 8 nm thick
atomic layer deposition (ALD) Al2O3 as the gate dielectric. Second, the coefficients of each
scattering mechanism of the IALMob model are extracted by comparing the mobility values
between these two simulation approaches.

Advanced Calibration for Device Simulation User Guide 90

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

Figure 46 shows the dispersion curves of the mobility field in the In0.53Ga0.47As inversion
layer simulated using the 1D mobility calculator of Sentaurus Band Structure with various
combinations of the alloy, surface roughness (SR), phonon (PH), and Coulomb (CO)
scattering models, in which parameters  and  of the surface-roughness power-spectrum
model with exponential density function have been optimized for the effective electric field in
the range 0.6–1.0 MV/cm. Afterwards, the mobility offset with respect to experimental data,
after considering all the available scattering models such as alloy, surface roughness,
phonon, and Coulomb scattering, as well as the fixed charge trap model, can be reduced
further in the range 0.1–0.4 MV/cm by adjusting the interface charge density.

Figure 46 Effective electron mobility versus effective electric field for In0.53Ga0.47As
(p-doping=1e17 cm–3 in channel) MOSFET with ALD Al2O3; simulation results are
obtained using 1D mobility calculator of Sentaurus Band Structure with various
combinations of scattering models, in which the surface roughness parameters
(=2.2e-7 cm; =4.6e-8 cm) and the interface charge density (Nit=4.3e12 cm–2)
have been calibrated; experimental data is taken from [62]
100000 Xuan et al.
Sband (Alloy+SR)
Sband (Alloy+SR+PH)
Sband (Alloy+SR+PH+CO)
Sband (Alloy+SR+PH+CO+Nit)
10000
µeff [cm2/Vs]

1000

100
0.0 0.5 1.0
Eeff [MV/cm]

For the IALMob model, the calibration has been performed with the parameters of 3D
phonon scattering and 3D Coulomb scattering taken from the constant mobility model and
the Philips unified mobility model, respectively, and the 2D and 3D parts of phonon
scattering are combined with PhononCombination=1. Since it is necessary to distinguish
the contribution of various scattering mechanisms in the IALMob model, first 2D Coulomb
scattering is disregarded by designating a large value in factors such as D1,inv and D2,inv.
Then, the resulting effective mobility values are compared to Sentaurus Band Structure
results, for which only surface roughness scattering and phonon scattering are activated, for
p-doping between 1e15 and 1e18 cm–3.
At this step, the reviewed parameters principally include coefficients accounting for 2D
phonon scattering (B, C, and ) and those representing surface roughness scattering (, A*,
and sr), where the  and sr coefficients describe the doping dependencies of phonon and
surface roughness scattering, respectively. Other parameters are kept as the ones for
silicon material.

Advanced Calibration for Device Simulation User Guide 91

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

After the coefficients for surface roughness scattering and 2D phonon scattering are
determined, the parameters for 2D Coulomb scattering (D1,inv, D2,inv, and 1,inv) can be
fitted with respect to Sentaurus Band Structure results when Coulomb scattering is
implemented.
Figure 47 compares the effective electron mobility versus sheet density at various p-doping
levels in the channel region between the IALMob model with calibrated parameters and
Sentaurus Band Structure results for In0.53Ga0.47As, InAs, and GaAs. Regarding the
interpolation scheme for the IALMob model in In1–xGaxAs alloys, a two-sectional linear curve
is used, based on In0.53Ga0.47As, InAs, and GaAs for each parameter previously
mentioned.

Figure 47 Effective electron mobility versus sheet density for (left) In0.53Ga0.47As, (middle)
InAs, and (right) GaAs MOSFET with ALD Al2O3 at various p-doping levels in the
channel region; dotted lines are obtained using 1D mobility calculator of Sentaurus
Band Structure in combination with surface roughness, phonon, and Coulomb
scattering models, and the alloy scattering model is further considered for
In0.53Ga0.47As; solid lines are obtained using the IALMob model with calibrated
parameters suggested in the InGaAs.par file
In0.53Ga0.47As InAs GaAs
15000 40000 10000 Sband (1e15)
Sband (1e15) Sband (1e15)
Sband (1e16) Sband (1e16) Sband (1e16)
12500 Sband (1e17) Sband (1e17) 8000 Sband (1e17)
Sband (1e18) 30000 Sband (1e18) Sband (1e18)

µeff [cm2/Vs]
µeff [cm2/Vs]

IALMob (1e15)
µeff [cm2/Vs]

10000 IALMob (1e15) IALMob (1e15)

IALMob (1e16)
IALMob (1e17)
IALMob (1e16) 6000 IALMob (1e16)
IALMob (1e17) IALMob (1e17)
7500 IALMob (1e18) 20000
4000
5000
10000
2500 2000

0 0 0
0 5e+12 1e+13 0 5e+12 1e+13 0 1e+13
−2 −2 −2
Sheet Density [cm ] Sheet Density [cm ] Sheet Density [cm ]

Interface Charge Mobility Model

Following the results of the mobility calibration for the IALMob model, the mobility
degradation model due to interface charges has been fine-tuned, including the
NegInterfaceCharge and PosInterfaceCharge models in Sentaurus Device. As shown in
Figure 46, the Sentaurus Band Structure mobility calculator can fit well to the experimental
data from Xuan et al. [62] when interface charge density (Nit) is fitted to be 4.3e12 cm–2.
To calibrate parameters such as  l , C trans , l crit , and E 0 for the NegInterfaceCharge and
PosInterfaceCharge models, the mobility values for two more Nit levels, 0 and 2.0e12, are
also computed using Sentaurus Band Structure.
Figure 48 presents the curves of mobility versus field for In0.53Ga0.47As material when
implementing the IALMob model and the negative and positive interface charge mobility
models with calibrated parameters with respect to Sentaurus Band Structure results. Similar

Advanced Calibration for Device Simulation User Guide 92

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Indium Gallium Arsenide Devices

to other mobility models, the interpolation scheme for the NegInterfaceCharge and
PosInterfaceCharge models in In1–xGaxAs alloys uses a two-sectional linear curve based
on InAs, In0.53Ga0.47As, and GaAs.

Figure 48 Effective electron mobility versus effective electric field for In0.53Ga0.47As
(p-doping=1e17 cm–3 in channel) MOSFET with ALD Al2O3; dotted lines are
obtained using 1D mobility calculator of Sentaurus Band Structure with alloy,
surface roughness, phonon, and Coulomb scattering models, as well as interface
trap model with charge density of 0, 2.0e12, and 4.3e12 cm–2; solid lines are
obtained using the NegInterfaceCharge and PosInterfaceCharge models with
calibrated parameters suggested in the InGaAs.par file

Xuan et al.
Sband (Nit=0)
100000 Sband (Nit=2.0e12)
Sband (Nit=4.3e12)
Neg|PosInterfaceCharge (Nc=0)
Neg|PosInterfaceCharge (Nc=2.0e12)
Neg|PosInterfaceCharge (Nc=4.3e12)
10000
µeff [cm2/Vs]

1000

100
0.0 0.2 0.4 0.6 0.8
Eeff [MV/cm]

Strained Low-Field Mobility

This section discusses strained low-field mobility.

Multivalley Electron Mobility Model

Mechanical stress in devices can change the band structure and carrier mobility, and is
considered an approach to enhance mobility. The multivalley electron mobility (eSubband)
model with default parameters has been tested on several In1–xGaxAs-based FETs on which
either uniaxial or biaxial stress is applied [63][64][65][66]. The simulations reproduce the
trend of electron mobility enhancement for tensile stress. However, the enhancement
generally tends to be lower than for experimental data.
Since the investigated device structures and conditions are very diverse and incomplete, a
more unified and better set of model parameters has not been established (for example, to

Advanced Calibration for Device Simulation User Guide 93

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Germanium Devices

consider the dependencies on mole fraction and geometry for In1–xGaxAs materials) and
default parameters have been kept.

Piezoresistance Mobility Model

Due to a lack of published experimental results for In1–xGaxAs alloys, the related
piezoresistive coefficients are difficult to determine. Nainani et al. [67] demonstrated the
conductivity extraction for n-In0.2Ga0.8As material using the transfer length method and
suggested a similar value for the longitudinal piezoresistive coefficient for n-In0.2Ga0.8As as
for n-Si.

Basic Properties of Silicon Germanium Devices

The SiliconGermaniumc100.par and SiliconGermaniumc110.par files contain the
parameters for silicon germanium (SiGe) MOS devices. See Content of Parameter Files on
page 11.

Permittivity
The low-frequency relative dielectric constant values are interpolated linearly between the
corner values of 11.7 (silicon) and 16.2 (germanium) for each Ge mole fraction (see [68]).

Thermal Conductivity
The mole fraction dependency of the thermal conductivity is extracted from data from
Schaffler [68] (see Figure 49). Data is implemented piecewise linearly in the parameter file.

Advanced Calibration for Device Simulation User Guide 94

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Germanium Devices

Figure 49 Thermal conductivity versus mole fraction [68][69][70][71][72][73]; solid line is the
implementation in the parameter file
1.8
Thermal Conductivity [W/cm/K] 1.6 Wagner
Schaffler
1.4
Maycock
1.2 Cheaito
1 Yonenaga
0.8 Lee
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1 1.2
Mole Fraction

Band Structure
Band structure data (band gap and affinity) is extracted from [74] (see Figure 50 and
Figure 51). The bandgap narrowing is extracted from Jain and Roulston [16] (see Figure 52)
where the OldSlotboom model is used for fitting (see the Sentaurus™ Device User Guide).

Figure 50 Band gap as a function of mole fraction

1.5

1.4
Sentaurus Device parameter set
1.3 Experimental data

1.2
Band Gap [eV]

1.1

0.9

0.8

0.7

0.6

0.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Mole Fraction

Advanced Calibration for Device Simulation User Guide 95

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Germanium Devices

Linear interpolation between silicon and germanium values is performed for the prefactor of
the OldSlotboom model to describe the mole fraction dependency. All data is implemented
piecewise linearly in the parameter file.

Figure 51 Conduction and valence bands as a function of mole fraction

-3.5

-3.7
Conduction Band
-3.9

-4.1
Energy [eV]

-4.3

-4.5

-4.7
Valence Band
-4.9

-5.1

-5.3

-5.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Mole Fraction

Figure 52 Bandgap narrowing for silicon and germanium; a prefactor of 5 meV for the
OldSlotboom model is used
140
Bandgap Narrowing [meV]

120
Silicon
100 Germanium, exp.
80 Germanium

60

40

20

0 17
10 1018 1019 1020
Dopant Concentration [cm–3]

Bulk Low-Field Mobility

The bulk low-field mobility PhuMob model parameters for silicon, SiGe, and germanium are
extracted from different measurements with the focus on the data from Golikova et al. [75]

Advanced Calibration for Device Simulation User Guide 96

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Germanium Devices

and Fistul' et al. [76]. Figure 53 shows the results of this calibration for germanium. The
model parameters are implemented piecewise linearly.

Figure 53 Measurements from different sources [75][76][77][78], with [79] used as the
reference measurement, for the doping dependency of electron and hole mobility in
Ge and the implementation in the Sentaurus Device parameter files (lines)
104 n, Model
n, Fistul'
p, Model
p, Mirabella
p, Sze
p, Golikova
Mobility [cm2/(Vs)]

103

102

101
1015 1016 1017 1018 1019 1020 1021 1022
Charge Carrier Concentration [cm−3]

Ballistic Mobility
Quasiballistic transport is important and should be considered when scaling devices
towards short channels. For this, it is required to extend the drift-diffusion (DD) model by
combining an additional ballistic mobility with the standard drift-diffusion mobility. In
Sentaurus Device, the ballistic mobility can be described by either the channel
length-dependent model or the kinetic velocity model (KVM) [80]. The latter has the
advantage of avoiding explicit channel length dependence and is used to extract the
corresponding ballistic mobility parameters.
The ballistic mobility model parameters are calibrated with the solution of the
subband-based Boltzmann transport equation (subband-BTE) as reference. The fitting
parameter (k) in the KVM is mostly determined by fitting the on-state current at low field over
a range of channel lengths between these two approaches. Moreover, the parameters
beta0 and the saturation velocity vsat0 in the high-field saturation model are adjusted as
well by fitting the on-state current at high field. In this way, ballistic mobility can be separated
from high-field drift-diffusion mobility. Note that, the ballistic mobility model parameters and
the high-field saturation model parameters can depend on the carrier polarity, the channel
geometry, the layer thickness, the material composition, and the strain.

Advanced Calibration for Device Simulation User Guide 97

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Germanium Devices

The devices used for parameter extraction are 3D square SiGe n-type and p-type MOSFET
nanowires with a cross-sectional area of 5 nm × 5 nm. The gate length ( L gate ) varies from
200 nm to 7 nm, and the channel direction is along the <110> crystal orientation.
To activate the KVM, specify the following command in the Physics section of the
Sentaurus Device command file:
Mobility (
BalMob ( KVM(Full) Fermi Frensley )
)

Here, the thermionic emission and free carrier acceleration terms are both considered in the
KVM. As the first step, the long-channel mobility is calibrated to the subband-BTE simulation
at on-state using the inversion and accumulation layer mobility (IALMob) model in
drift-diffusion. Second, the channel access resistance (R0) is extracted from the gate length
dependence of the total resistance using the equation R = R0 + R'Lgate. The difference of R0
between the subband-BTE and drift-diffusion simulations can be added in the Sentaurus
Device simulation either as an additional lumped resistance or by modifying the mobility
model parameters in source/drain regions. After that, the parameters of the KVM and the
high-field saturation model are optimized to match the on-state current at low-drain and
high-drain bias voltages. It is likely that you must iterate the calibration procedures
previously mentioned to optimize the parameters. For a detailed description, see Chapter 4
on page 46.
Figure 54 shows the on-state drain current obtained using the drift-diffusion simulation with
the KVM at low-drain and high-drain bias voltages. The parameters of the KVM and the
high-field saturation model have been optimized with respect to the simulation results of the
Subband-BTE solver over a wide range of channel lengths and for germanium mole
fractions from 0 to 1, as shown in Figure 55 on page 99.
Note that the slope of resistance as a function of gate length is found to vary nonlinearly at
short channels when the KVM is implemented, and it then leads to a slightly lowered fitting
accuracy for channels shorter than 10 nm. This behavior is under investigation and might
result in a parameter update.

Advanced Calibration for Device Simulation User Guide 98

V-2024.03
Chapter 5: Quality of Fitting and Extraction
Basic Properties of Silicon Germanium Devices

Figure 54 Drain current versus channel length for 5 nm × 5 nm Si (left) n-type MOSFET
nanowire and (right) p-type MOSFET nanowire, at low-drain and high-drain bias
voltages; squares obtained using Subband-BTE solver in combination with surface
roughness, phonon, and Coulomb scattering models; solid lines obtained using
KVM with calibrated parameters for electrons (k=6.3, Tfactor=1.52, and Bfactor=0.3
in KVM; beta0=0.74 and vsat0=4e7 in high-field saturation model) and for holes
(k=5.4, Tfactor=1.52, and Bfactor=0.3 in KVM; beta0=1.0 and vsat0=2e7 in
high-field saturation model)
700 500
SBTE (Vd=0.05V; Vg=0.5V) SBTE (Vd=-0.05V; Vg=-0.5V)
DD+KVM (Vd=0.05V; Vg=0.5V) DD+KVM (Vd=-0.05V; Vg=-0.5V)
600
Drain Current [µA/µm]

Drain Current [µA/µm]

SBTE (Vd=0.5V; Vg=0.5V) SBTE (Vd=-0.5V; Vg=-0.5V)
DD+KVM (Vd=0.5V; Vg=0.5V)
400
DD+KVM (Vd=-0.5V; Vg=-0.5V)
500
n-type MOSFET nanowire p-type MOSFET nanowire
400 300

300 200
200
100
100
0 0
0 50 100 150 200 0 50 100 150 200
Channel Length [nm] Channel Length [nm]

Figure 55 Calibrated parameters (left) k in KVM and (right) vsat0 in high-field saturation model
versus Ge mole fraction for 5 nm × 5 nm SiGe n-type and p-type MOSFET
nanowires
20 5.0e+7
electron hole electron hole

15 4.0e+7

3.0e+7
vsat0

10
k

2.0e+7
5
1.0e+7

0 0.0e+0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
xGe xGe

Advanced Calibration for Device Simulation User Guide 99

V-2024.03
Chapter 5: Quality of Fitting and Extraction
References

References
[1] S. Takagi et al., “On the Universality of Inversion Layer Mobility in Si MOSFET’s:
Part I—Effects of Substrate Impurity Concentration,” IEEE Transactions on Electron
Devices, vol. 41, no. 12, pp. 2357–2362, 1994.
[2] S. Takagi et al., “On the Universality of Inversion Layer Mobility in Si MOSFET’s:
Part II—Effects of Surface Orientation,” IEEE Transactions on Electron Devices,
vol. 41, no. 12, pp. 2363–2368, 1994.
[3] H. Nakamura et al., “Effects of Selecting Channel Direction in Improving Performance
of Sub-100 nm MOSFETs Fabricated on (110) Surface Si Substrate,” Japanese
Journal of Applied Physics, vol. 43, no. 4B, pp. 1723–1728, 2004.
[4] H. M. Nayfeh et al., “Influence of High Channel Doping on the Inversion Layer
Electron Mobility in Strained Silicon n-MOSFETs,” IEEE Electron Device Letters,
vol. 24, no. 4, pp. 248–250, 2003.
[5] C. D. Young et al., “(110) and (100) Sidewall-oriented FinFETs: A performance and
reliability investigation,” Solid-State Electronics, vol. 78, pp. 2–10, December 2012.
[6] A. T. Pham, C. Jungemann, and B. Meinerzhagen, “Modeling and validation of
piezoresistive coefficients in Si hole inversion layers,” Solid-State Electronics, vol. 53,
no. 12, pp. 1325–1333, 2009.
[7] F. M. Bufler, A. Erlebach, and M. Oulmane, “Hole Mobility Model With Silicon
Inversion Layer Symmetry and Stress-Dependent Piezoconductance Coefficients,”
IEEE Electron Device Letters, vol. 30, no. 9, pp. 996–998, 2009.
[8] P. Packan et al., “High Performance Hi-K + Metal Gate Strain Enhanced Transistors
on (110) Silicon,” in IEDM Technical Digest, San Francisco, CA, USA, pp. 1–4,
December 2008.
[9] R. Chau et al., “High-/Metal–Gate Stack and Its MOSFET Characteristics,” IEEE
Electron Device Letters, vol. 25, no. 6, pp. 408–410, 2004.
[10] M. Cassé et al., “Carrier Transport in HfO2/Metal Gate MOSFETs: Physical Insight
Into Critical Parameters,” IEEE Transactions on Electron Devices, vol. 53, no. 4,
pp. 759–768, 2006.
[11] N. Stavitski et al., “Evaluation of Transmission Line Model Structures for
Silicide-to-Silicon Specific Contact Resistance Extraction,” IEEE Transactions on
Electron Devices, vol. 55, no. 5, pp. 1170–1176, 2008.
[12] J. B. McKeon et al., “Experimental Determination of Electron and Hole Mobilities in
MOS Accumulation Layers,” IEEE Electron Device Letters, vol. 18, no. 5, pp. 200–
202, 1997.
[13] Y. Asahi, Physical parameters for 4H-SiC and 6H-SiC, Internal Report, Mountain
View, California: Synopsys, Inc., November 2012.

Advanced Calibration for Device Simulation User Guide 100

V-2024.03
Chapter 5: Quality of Fitting and Extraction
References

[14] U. Lindefelt, “Doping-induced band edge displacements and band gap narrowing in
3C–, 4H–, 6H–SiC, and Si,” Journal of Applied Physics, vol. 84, no. 5, pp. 2628–2637,
1998.
[15] C. Persson, U. Lindefelt, and B. E. Sernelius, “Bang gap narrowing in n-type and
p-type 3C–, 2H–, 4H–, 6H–SiC, and Si,” Journal of Applied Physics, vol. 86, no. 8,
pp. 4419–4427, 1999.
[16] S. C. Jain and D. J. Roulston, “A Simple Expression for Band Gap Narrowing (BGN)
in Heavily Doped Si, Ge, GaAs and GexSi1–x Strained Layers,” Solid-State
Electronics, vol. 34, no. 5, pp. 453–465, 1991.
[17] T. Hatakeyama et al., “Impact ionization coefficients of 4H silicon carbide,” Applied
Physics Letters, vol. 85, no. 8, pp. 1380–1382, 2004.
[18] H. Niwa, J. Suda, and T. Kimoto, “Temperature Dependence of Impact Ionization
Coefficients in 4H-SiC,” Materials Science Forum, vol. 778-780, pp. 461–466, 2014.
[19] A. O. Konstantinov et al., “Ionization rates and critical fields in 4H silicon carbide,”
Applied Physics Letters, vol. 71, no. 1, pp. 90–92, 1997.
[20] Y. S. Touloukian and E. H. Buyco, Specific Heat: Nonmetallic Solids, Thermophysical
Properties of Matter, TPRC Data Series, vol. 5, New York: IFI/Plenum, 1970.
[21] Y. S. Touloukian and C. Y. Ho (eds.), Thermal Conductivity–Nonmetallic Solids,
Thermophysical Properties of Matter, TPRC Data Series, vol. 2, New York: IFI/
Plenum, 1970.
[22] E. A. Burgemeister, W. von Muench, and E. Pettenpaul, “Thermal conductivity and
electrical properties of 6H silicon carbide,” Journal of Applied Physics, vol. 50, no. 9,
pp. 5790–5794, 1979.
[23] D. Morelli et al., “Carrier concentration dependence of the thermal conductivity of
silicon carbide,” in 5th Silicon Carbide and Related Materials Conference,
Washington, DC, USA, pp. 313–315, November 1993.
[24] D. T. Morelli et al., “Phonon-electron scattering in single crystal silicon carbide,”
Applied Physics Letters, vol. 63, no. 23, pp. 3143–3145, 1993.
[25] W. J. Schaffer et al., “Conductivity Anisotropy in Epitaxial 6H and 4H SiC,” in MRS
Symposium Proceedings, Diamond, SiC and Nitride Wide Bandgap Semiconductors,
vol. 339, San Francisco, CA, USA, pp. 595–600, April 1994.
[26] M. Roschke and F. Schwierz, “Electron Mobility Models for 4H, 6H, and 3C SiC,” IEEE
Transactions on Electron Devices, vol. 48, no. 7, pp. 1442–1447, 2001.
[27] S. Kagamihara et al., “Parameters required to simulate electric characteristics of SiC
devices for n-type 4H–SiC,” Journal of Applied Physics, vol. 96, no. 10, pp. 5601–
5606, 2004.
[28] I. A. Khan and J. A. Cooper, Jr., “Measurement of High Field Electron Transport in
Silicon Carbide,” Materials Science Forum, vol. 264-268, pp. 509–512, February
1998.

Advanced Calibration for Device Simulation User Guide 101

V-2024.03
Chapter 5: Quality of Fitting and Extraction
References

[29] M. Lades, Modeling and Simulation of Wide Bandgap Semiconductor Devices: 4H/
6H-SiC, PhD thesis, Technischen Universität, Munich, Germany, 2000.
[30] A. Koizumi, J. Suda, and T. Kimoto, “Temperature and doping dependencies of
electrical properties in Al-doped 4H-SiC epitaxial layers,” Journal of Applied Physics,
vol. 106, no. 1, p. 013716, 2009.
[31] J. Pernot, S. Contreras and J. Camassel, “Electrical transport properties of
aluminum-implanted 4H–SiC,” Journal of Applied Physics, vol. 98, no. 2, p. 023706,
2005.
[32] T. Troffer et al., “Doping of SiC by Implantation of Boron and Aluminum,” Physica
Status Solidi (a), vol. 162, no. 1, pp. 277–298, 1997.
[33] M. Noguchi et al., “Determination of intrinsic phonon-limited mobility and carrier
transport property extraction of 4H-SiC MOSFETs,” in IEDM Technical Digest, San
Francisco, CA, USA, pp. 219–222, December 2017.
[34] M. Noguchi et al., “Hall effect mobility for SiC MOSFETs with increasing dose of
nitrogen implantation into channel region,” Japanese Journal of Applied Physics,
vol. 57, no. 4S, p. 04FR13, 2018.
[35] T. Hatakeyama et al., “Characterization of traps at nitrided SiO2/SiC interfaces near
the conduction band edge by using Hall effect measurements,” Applied Physics
Express, vol. 10, no. 4, p. 046601, 2017.
[36] Sentaurus™ Device User Guide, Version V-2024.03, Mountain View, California:
Synopsys, Inc., 2024.
[37] M. Noguchi et al., “Carrier transport properties in inversion layer of Si-face 4H–SiC
MOSFET with nitrided oxide,” Japanese Journal of Applied Physics, vol. 58, no. 3,
p. 031004, 2019.
[38] W. M. Waller et al., “Subthreshold Mobility in AlGaN/GaN HEMTs,” IEEE Transactions
on Electron Devices, vol. 63, no. 5, pp. 1861–1865, 2016.
[39] S.-J. Chang et al., “Investigation of channel mobility in AlGaN/GaN
high-electron-mobility transistors,” Japanese Journal of Applied Physics, vol. 55, no.
4, p. 044104, 2016.
[40] S. Adachi, Properties of Group-IV, III–V and II–VI Semiconductors, West Sussex,
England: John Wiley & Sons, 2005.
[41] S. Adachi, Properties of Semiconductor Alloys: Group-IV, III–V and II–VI
Semiconductors, West Sussex, England: John Wiley & Sons, 2009.
[42] M. S. Abrahams, R. Braunstein, and F. D. Rosi, “Thermal, Electrical and Optical
Properties of (In,Ga)As Alloys,” Journal of Physics and Chemistry of Solids, vol. 10,
no. 2–3, pp. 204–210, 1959.
[43] D. G. Arasly, R. N. Ragimov, and M. I. Aliev, “Characteristics of phonon scattering in
GaxIn1–xAs solid solutions,” Soviet Physics Semiconductors, vol. 24, no. 2, pp. 225–
226, 1990.

Advanced Calibration for Device Simulation User Guide 102

V-2024.03
Chapter 5: Quality of Fitting and Extraction
References

[44] S. Adachi, “Lattice thermal conductivity of group-IV and III-V semiconductor alloys,”
Journal of Applied Physics, vol. 102, p. 063502, September 2007.
[45] R. E. Nahory, M. A. Pollack, and J. C. DeWinter, “Growth and characterization of
liquid-phase epitaxial InxGa1–xAs,” Journal of Applied Physics, vol. 46, no. 2,
pp. 775–782, 1975.
[46] T. J. Kim et al., “Dielectric functions of InxGa1–xAs alloys,” Physical Review B, vol. 68,
no. 11, p. 115323, 2003.
[47] D. K. Gaskill et al., “Band-gap determination by photoreflectance of InGaAs and
InAlAs lattice matched to InP,” Applied Physics Letters, vol. 56, no. 13, pp. 1269–
1271, 1990.
[48] I. Vurgaftman, J. R. Meyer, and L. R. Ram-Mohan, “Band parameters for III–V
compound semiconductors and their alloys,” Journal of Applied Physics, vol. 89,
no. 11, pp. 5815–5875, 2001.
[49] K. Alavi, R. L. Aggarwal, and S. H. Groves, “Interband magnetoabsorption of
In0.53Ga0.47As,” Physical Review B, vol. 21, no. 3, pp. 1311–1315, 1980.
[50] R. J. Warburton et al., “Valence band spin splitting in strained In0.18Ga0.82As/GaAs
quantum wells,” Semiconductor Science and Technology, vol. 6, no. 5, pp. 359–364,
1991.
[51] N. J. Traynor et al., “Highly nonlinear Zeeman splitting of excitons in semiconductor
quantum wells,” Physical Review B, vol. 55, no. 23, pp. 15701–15705, 1997.
[52] S. C. Jain, J. M. McGregor, and D. J. Roulston, “Band-gap narrowing in novel III-V
semiconductors,” Journal of Applied Physics, vol. 68, no. 7, pp. 3747–3749, 1990.
[53] G. Borghs et al., “Band-gap narrowing in highly doped n- and p-type GaAs studied by
photoluminescence spectroscopy,” Journal of Applied Physics, vol. 66, no. 9,
pp. 4381–4386, 1989.
[54] B. J. Aitchison et al., “Enhanced hot-electron photoluminescence from heavily
carbon-doped GaAs,” Applied Physics Letters, vol. 56, no. 12, pp. 1154–1156, 1990.
[55] D. Olego and M. Cardona, “Photoluminescence in heavily doped GaAs. I.
Temperature and hole-concentration dependence,” Physical Review B, vol. 22, no. 2,
pp. 886–893, 1980.
[56] Z. H. Lu, M. C. Hanna, and A. Majerfeld, “Determination of band gap narrowing and
hole density for heavily C-doped GaAs by photoluminescence spectroscopy,” Applied
Physics Letters, vol. 64, no. 1, pp. 88–90, 1994.
[57] B. P. Yan, J. S. Luo, and Q. L. Zhang, “Study of band-gap narrowing effect and
nonradiative recombination centers for heavily C-doped GaAs by photoluminescence
spectroscopy,” Journal of Applied Physics, vol. 77, no. 9, pp. 4822–4824, 1995.
[58] H. Yao and A. Compaan, “Plasmons, photoluminescence, and band-gap narrowing in
very heavily doped n-GaAs,” Applied Physics Letters, vol. 57, no. 2, pp. 147–149,
1990.

Advanced Calibration for Device Simulation User Guide 103

V-2024.03
Chapter 5: Quality of Fitting and Extraction
References

[59] N. A. Semikolenova, I. M. Nesmelova, and E. N. Khabarov, “Investigation of the

impurity interaction mechanism in indium arsenide,” Soviet Physics Semiconductors,
vol. 12, no. 10, pp. 1139–1142, 1978.
[60] M. Sotoodeh, A. H. Khalid, A. A. Rezazadeh, “Empirical low-field mobility model for
III–V compounds applicable in device simulation codes,” Journal of Applied Physics,
vol. 87, no. 6, pp. 2890–2900, 2000.
[61] Ch. Köpf, H. Kosina, and S. Selberherr, “Physical Models for Strained and Relaxed
GaInAs Alloys: Band Structure and Low-Field Transport,” Solid-State Electronics,
vol. 41, no. 8, pp. 1139–1152, 1997.
[62] Y. Xuan et al., “Submicrometer Inversion-Type Enhancement-Mode InGaAs MOSFET
With Atomic-Layer-Deposited Al2O3 as Gate Dielectric,” IEEE Electron Device
Letters, vol. 28, no. 11, pp. 935–938, 2007.
[63] S. Suthram et al., “Strain Additivity in III-V Channels for CMOSFETs beyond 22nm
Technology Node,” in Symposium on VLSI Technology, Honolulu, HI, USA, pp. 182–
183, June 2008.
[64] L. Xia and J. A. del Alamo, “Mobility Enhancement in Indium-rich N-channel InxGa1–
xAs HEMTs by Application of <110> Uniaxial Strain,” in International Conference on
Indium Phosphide and Related Materials (IPRM), Takamatsu, Japan, pp. 1–4, May
2010.
[65] S. H. Kim et al., “Enhancement Technologies and Physical Understanding of Electron
Mobility in III-V n-MOSFETs with Strain and MOS Interface Buffer Engineering,” in
IEDM Technical Digest, Washington, DC, USA, pp. 311–314, December 2011.
[66] C. Rossel et al., “Strain effects on n-InGaAs heterostructure-on-insulator made by
direct wafer bonding,” Solid-State Electronics, vol. 98, pp. 88–92, August 2014.
[67] A. Nainani et al., “Study of piezoresistance under uniaxial stress for technologically
relevant III-V semiconductors using wafer bending experiments,” Applied Physics
Letters, vol. 96, no. 24, p. 242110, 2010.
[68] F. Schäffler, “Silicon-Germanium (Si1–xGex),” Properties of Advanced Semiconductor
Materials: GaN, AlN, InN, BN, SiC, SiGe, M. E. Levinshtein, S. L. Rumyantsev, and
M. S. Shur (eds.), New York: John Wiley & Sons, 2001.
[69] M. Wagner, Simulation of Thermoelectric Devices, PhD thesis, Technischen
Universität Wien, Austria, November 2007.
[70] R. Cheaito et al., “Experimental Investigation of Size Effects on the Thermal
Conductivity of Silicon-Germanium Alloy Thin Films,” Physical Review Letters,
vol. 109, no. 19, p. 195901, 2012.
[71] I. Yonenaga et al., “Thermal and Electrical Properties of Czochralski Grown GeSi
Alloys,” in Proceedings of the 17th International Conference on Thermoelectrics
(ICT), Nagoya, Japan, pp. 402–405, May 1998.
[72] P. D. Maycock, “Thermal Conductivity of Silicon, Germanium, III–V Compounds and
III–V Alloys,” Solid-State Electronics, vol. 10, no. 3, pp. 161–168, 1967.

Advanced Calibration for Device Simulation User Guide 104

V-2024.03
Chapter 5: Quality of Fitting and Extraction
References

[73] S.-M. Lee, D. G. Cahill, and R. Venkatasubramanian, “Thermal conductivity of Si–Ge

superlattices,” Applied Physics Letters, vol. 70, no. 22, pp. 2957–2959, 1997.
[74] C. Penn, T. Fromherz, and G. Bauer, “Energy gaps and band structure of SiGe and
their temperature dependence,” Properties of Silicon Germanium and SiGe:Carbon,
EMIS Datareviews Series, no. 24, London: INSPEC, 2000.
[75] O. A. Golikova, B. Ya. Moizhes, and L. S. Stil'bans, “Hole Mobility of Germanium as a
Function of Concentration and Temperature,” Soviet Physics - Solid State, vol. 3,
no. 10, pp, 2259–2265, 1962.
[76] V. I. Fistul', M. I. Iglitsyn, and É. M. Omel'yanovskii, “Mobility of Electrons in
Germanium Strongly Doped With Arsenic,” Soviet Physics - Solid State, vol. 4, no. 4,
pp. 784–785, 1962.
[77] S. M. Sze and J. C. Irvin, “Resistivity, Mobility and Impurity Levels in GaAs, Ge, and
Si at 300° K,” Solid-State Electronics, vol. 11, no. 6, pp. 599–602, 1968.
[78] S. Mirabella et al., “Activation and carrier mobility in high fluence B implanted
germanium,” Applied Physics Letters, vol. 92, p. 251909, June 2008.
[79] L. Hutin et al., “Activation level in boron-doped thin germanium-on-insulator (GeOI):
Extraction method and impact of mobility,” Materials Science in Semiconductor
Processing, vol. 11, no. 5–6, pp. 267–270, 2008.
[80] O. Penzin et al., “Kinetic Velocity Model to Account for Ballistic Effects in the
Drift-Diffusion Transport Approach,” IEEE Transactions on Electron Devices, vol. 64,
no. 11, pp. 4599–4606, 2017.

Advanced Calibration for Device Simulation User Guide 105

V-2024.03
A
A Physical Model Interface Models

This appendix describes the physical model interface (PMI) models of Advanced Calibration
Device.

Features of the MCmob and SBmob Models

The PMI models MCmob and SBmob of Sentaurus Device are based on nonlinear relations
between the mobility enhancement and the stress components where the coefficients of the
equations are extracted from reference tools such as Sentaurus Band Structure (SBmob) or
Sentaurus Device Monte Carlo (MCmob). The corresponding command file sections are
explained in Chapter 2 on page 23.
The features of the MCmob model are as follows:
• Calculates the hole mobility enhancement by stress for silicon and SiGe for mole
fractions from 0% to 100% for the channel direction and wafer orientation combinations
<110>/(001), <110>/(110), and <100>/(001)
• Calculates the electron mobility enhancement by stress for silicon for the channel
direction and wafer orientation combinations <110>/(001), <110>/(110), and <100>/(001)
• Determines the crystallographic surface or interface orientation automatically
The features of the SBmob model are as follows:
• Calculates the electron and hole mobility enhancement by stress for silicon and SiGe for
mole fractions from 0% to 100% for the channel direction and wafer orientation
combinations <110>/(001), <110>/(110), and <100>/(001)
The mole fraction is discretized over 10 bins (intervals). For example, a mole fraction of
0.54 would be placed in the 6th bin.
• Determines the crystallographic surface or interface orientation automatically
These models are designed for semiconductor–isolator interfaces as in a metal–insulator–
semiconductor (MIS) structure where the bulk symmetry typical for the standard

Advanced Calibration for Device Simulation User Guide 106

V-2024.03
Appendix A: Physical Model Interface Models
Parameters

piezoresistance model is broken. This makes it necessary to determine the direction of the
interface normal with respect to the device coordinate system.
Note:
This detection is not performed fully and automatically, and you must specify
parameters in the Sentaurus Device parameter file (see Parameters).

Parameters
You use the same parameters and parameter syntax for both the MCmob and SBmob models.
Table 16 Parameters of the MCmob and SBmob models

Parameter Description

autoorientation Switches on the detection of the crystallographic surface

orientation. Valid options are:
• 0: No detection of crystallographic surface orientation, but
crystallographic surface orientation is defined by waferori.
• 1: Detects crystallographic surface orientation.
• 2: Detects crystallographic surface orientation and
adaptation of local coordinate system (coordinate system
used to match the stress components to the MIS plane
orientation and channel direction at each mesh node) for
FinFET-like structures with <110> channel orientation.
• 3: Detects crystallographic surface orientation; local
coordinate systems are defined using parameters in the
Sentaurus Device parameter file.
Default: 0

channelori Sets the channel direction. Valid options are:

• 100: <100> channel direction.
• 110: <110> channel direction.
Default: 110

ConstantMoleFraction Switches between reading the mole-fraction distribution from a

TDR file and setting the constant mole fraction in the parameter
file. Valid options are:
• 0: Read from TDR file.
• 1: User sets mole fraction in Sentaurus Device parameter file.

Advanced Calibration for Device Simulation User Guide 107

V-2024.03
Appendix A: Physical Model Interface Models
Parameters

Table 16 Parameters of the MCmob and SBmob models (Continued)

Parameter Description

ConstantStress Switches between reading the distribution of stress components

from a TDR file and setting the constant stress components in
the parameter file. Valid options are:
• 0: Read from TDR file.
• 1: User sets constant stress components in Sentaurus
Device parameter file.

coordinatesystem Specifies the global coordinate system in Sentaurus Device.

Default: 0
See Specifying Device Coordinate Systems on page 111.

corrections Controls interpolation in the stress space. Valid options are:

• 0: Results in a linear superposition of the mobility response to
the uniaxial stress components.
• 1: Switches on cross-corrections (terms with mixed stress
components).
You must define the cross-corrections in the parameter file
(default: all cross-corrections are set to zero).
Default: 0
Note:
This parameter is available only for the MCmob model.

DeltaMuPrefactors100c100x Scales the mobility enhancement caused by stress in the

current flow direction for channel direction <100> and surface
orientation (100).

DeltaMuPrefactors100c100y Scales the mobility enhancement caused by stress normal to

the MIS interface for channel direction <100> and surface
orientation (100).

DeltaMuPrefactors100c100z Scales the mobility enhancement caused by stress in-plane

with the MIS interface and normal to the current flow direction
for channel direction <100> and surface orientation (100).

DeltaMuPrefactors100c110x Scales the mobility enhancement caused by stress in the

current flow direction for channel direction <110> and surface
orientation (100).

DeltaMuPrefactors100c110y Scales the mobility enhancement caused by stress normal to

the MIS interface for channel direction <110> and surface
orientation (100).

Advanced Calibration for Device Simulation User Guide 108

V-2024.03
Appendix A: Physical Model Interface Models
Parameters

Table 16 Parameters of the MCmob and SBmob models (Continued)

Parameter Description

DeltaMuPrefactors100c110z Scales the mobility enhancement caused by stress in-plane

with the MIS interface and normal to the current flow direction
for channel direction <110> and surface orientation (100).

DeltaMuPrefactors110c110x Scales the mobility enhancement caused by stress in the

current flow direction for channel direction <110> and surface
orientation (110).

DeltaMuPrefactors110c110y Scales the mobility enhancement caused by stress normal to

the MIS interface for channel direction <110> and surface
orientation (110).

DeltaMuPrefactors110c110z Scales the mobility enhancement caused by stress in-plane

with the MIS interface and normal to the current flow direction
for channel direction <110> and surface orientation (110).

MoleFraction SiGe mole fraction when ConstantMoleFraction=1.

referenceplanecoordinatex1 Coordinate of the first reference plane in the x-direction of the

device coordinate system (yz plane, default: 1e6).

referenceplanecoordinatex2 Coordinate of the second reference plane in the x-direction of

the device coordinate system (yz plane, default: 1e6).

referenceplanecoordinatey1 Coordinate of the first reference plane in the y-direction of the

device coordinate system (xz plane, default: 1e6).

referenceplanecoordinatey2 Coordinate of the second reference plane in the y-direction of

the device coordinate system (xz plane, default: 1e6).

referenceplanecoordinatez1 Coordinate of the first reference plane in the z-direction of the

device coordinate system (xy plane, default: 1e6).

referenceplanecoordinatez2 Coordinate of the second reference plane in the z-direction of

the device coordinate system (xy plane, default: 1e6).

referenceplanesystemx1 Defines the local coordinate system for the reference plane
parallel to the yz plane at the position
x=referenceplanecoordinatex1 (default: –1).

referenceplanesystemx2 Defines the local coordinate system for the reference plane
parallel to the yz plane at the position
x=referenceplanecoordinatex2 (default: –1).

Advanced Calibration for Device Simulation User Guide 109

V-2024.03
Appendix A: Physical Model Interface Models
Parameters

Table 16 Parameters of the MCmob and SBmob models (Continued)

Parameter Description

referenceplanesystemy1 Defines the local coordinate system for the reference plane
parallel to the xz plane at the position
y=referenceplanecoordinatey1 (default: –1).

referenceplanesystemy2 Defines the local coordinate system for the reference plane
parallel to the xz plane at the position
y=referenceplanecoordinatey2 (default: –1).

referenceplanesystemz1 Defines the local coordinate system for the reference plane
parallel to the xy plane at the position
y=referenceplanecoordinatez1 (default: –1).

referenceplanesystemz2 Defines the local coordinate system for the reference plane
parallel to the xy plane at the position
y=referenceplanecoordinatez2 (default: –1).

Stress_xx Sxx stress component in the device coordinate system when

ConstantStress=1.

Stress_yy Syy stress component in the device coordinate system when

ConstantStress=1.

Stress_zz Szz stress component in the device coordinate system when

ConstantStress=1.

waferori Sets the wafer orientation. Valid options are:

• 100: (001) wafer orientation.
• 110: (110) wafer orientation.
Default: 100

Note:
The set of DeltaMuPrefactor* parameters for different surface and channel
orientations (for example, DeltaMuPrefactors100c110y) is available only for
the SBmob model. For the MCmob model, only a reduced set is available (for
example, DeltaMuPrefactory).

Advanced Calibration for Device Simulation User Guide 110

V-2024.03
Appendix A: Physical Model Interface Models
Parameters

Specifying Device Coordinate Systems

For planar devices with one conducting MIS plane, you can use the default parameters and
no additional information is necessary when a device coordinate system has the following
configuration:
• x: Channel direction
• y: Normal to the MIS interface
• z: In-plane with the MIS interface but normal to the channel direction, and the wafer
orientation equals (100)
For other device coordinate systems and one conducting MIS plane, you must set the
parameter coordinatesystem in the Sentaurus Device parameter file for the following
configurations:
• coordinatesystem=1:
x: Channel direction
z: Direction normal to the MIS interface
y: In-plane with the MIS interface but normal to the channel direction
• coordinatesystem=2:
y: Channel direction
x: Direction normal to the MIS interface
z: In-plane with the MIS interface but normal to the channel direction
• coordinatesystem=3:
y: Channel direction
z: Direction normal to the MIS interface
x: In-plane with the MIS interface but normal to the channel direction
• coordinatesystem=4:
z: Channel direction
x: Direction normal to the MIS interface
y: In-plane with the MIS interface but normal to the channel direction
• coordinatesystem=5:
z: Channel direction
y: Direction normal to the MIS interface
x: In-plane with the MIS interface but normal to the channel direction

Specifying Local Coordinate Systems

For general applications and for FinFETs with <100> channel direction and (100) wafer
orientation, you must define the local coordinate system manually in the Sentaurus Device
parameter file using the referenceplanesystem* parameters.

Advanced Calibration for Device Simulation User Guide 111

V-2024.03
Appendix A: Physical Model Interface Models
Parameters

You can specify the following configurations of the local coordinate system:
• For a MIS interface in the xz plane close to referenceplanecoordinatey1 or
referenceplanecoordinatey2:

◦ referenceplanesystemy1=0 or referenceplanesystemy2=0 specifies:

x: Channel direction
y: Direction normal to the MIS interface
z: In-plane with the MIS interface but normal to the channel direction
◦ referenceplanesystemy1=1 or referenceplanesystemy2=1 specifies:
z: Channel direction
y: Direction normal to the MIS interface
x: In-plane with the MIS interface but normal to the channel direction
• For a MIS interface in the xy plane close to referenceplanecoordinatez1 or
referenceplanecoordinatez2:

◦ referenceplanesystemz1=0 or referenceplanesystemz2=0 specifies:

x: Channel direction
z: Direction normal to the MIS interface
y: In-plane with the MIS interface but normal to the channel direction
◦ referenceplanesystemz1=1 or referenceplanesystemz2=1 specifies:
y: Channel direction
z: Direction normal to the MIS interface
x: In-plane with the MIS interface but normal to the channel direction
• For a MIS interface in the yz plane close to referenceplanecoordinatex1 or
referenceplanecoordinatex2:

◦ referenceplanesystemx1=0 or referenceplanesystemx2=0 specifies:

y: Channel direction
x: Direction normal to the MIS interface
z: In-plane with the MIS interface but normal to the channel direction
◦ referenceplanesystemx1=1 or referenceplanesystemx2=1 specifies:
z: Channel direction
x: Direction normal to the MIS interface
y: In-plane with the MIS interface but normal to the channel direction
For the simulation of a planar MIS transistor, using the waferori and coordinatesystem
parameters is sufficient. Instead of using waferori, you can set autoorientation=1 to
detect automatically the crystallographic wafer or surface orientation. In this case, waferori
is overwritten by the automatically detected orientation.
For other applications or for FinFETs with <100> channel orientation, you can use the
referenceplanesystem* parameters to define the local coordinate system where the

Advanced Calibration for Device Simulation User Guide 112

V-2024.03
Appendix A: Physical Model Interface Models
Parameters

crystallographic surface orientation is detected automatically. For that, set

autoorientation=3.

Using Parameters
The following examples illustrate how to use the parameters for different orientations and
coordinate systems.
Example 1
Planar MISFET with (100) wafer orientation and device coordinate system where x=channel
direction, y=wafer normal direction, and z=device width direction:
Use default parameter values.
Example 2
Planar MISFET with (100) wafer orientation and device coordinate system where x=channel
direction, z=wafer normal direction, and y=device width direction, specify:
coordinatesystem=1

Example 3
Planar MISFET with (110) wafer orientation and device coordinate system where x=channel
direction, z=wafer normal direction, and y=device width direction, specify either:
coordinatesystem=1
autoorientation=1

or:
coordinatesystem=1
waferori=110

Example 4
FinFET with (100) wafer orientation, <110> channel orientation, and device coordinate
system where x=channel direction, y=wafer normal direction, and z=fin width direction,
specify:
autoorientation=2

Example 5
FinFET with (100) wafer orientation, <110> channel orientation, and device coordinate
system where x=channel direction, z=wafer normal direction, and y=fin width direction,
specify:
coordinatesystem=1
autoorientation=2

Advanced Calibration for Device Simulation User Guide 113

V-2024.03
Appendix A: Physical Model Interface Models
Parameter Interface

Example 6
FinFET with (100) wafer orientation, <100> channel orientation, and device coordinate
system where x=channel direction, z=wafer normal direction, and y=fin width direction,
specify:
coordinatesystem=1
autoorientation=3
referenceplanesystemy1=0
referenceplanesystemy2=0
referenceplanesystemz1=0
referenceplanecoordinatey1=-@W/2.0@
referenceplanecoordinatey2=@W/2.0@
referenceplanecoordinatez1=@H@

Figure 56 Cross section of a FinFET structure normal to the channel direction with height (H)
and width (W); the origin of the coordinate system is at y=0 and z=0

H
Fin

referenceplanecoordinatey1=-W/2
referenceplanecoordinatey2=W/2
referenceplanecoordinatez1=H
referenceplanesystemy1=0
referenceplanesystemy2=0
Z referenceplanesystemz1=0

–W/2 Y W/2

Parameter Interface
The MCmob and SBmob models have a parameter interface that allows you to change all the
coefficients of the fit and interpolation functions. In this way, you can define the mobility
response to mechanical stress using the Sentaurus Device parameter file. For historical
reasons, the syntax of the interface differs for MCmob and SBmob.
Parameter Interface and Equations of SBmob Model explains the parameter interface of
SBmob and its related implementations and equations.

Some settings must be specified in the Sentaurus Device command file. The Type
parameter defines the device type: NMOS (Type=0) or PMOS (Type=1). The Kanda option
switches on the Kanda model (see the Sentaurus™ Device User Guide).

Advanced Calibration for Device Simulation User Guide 114

V-2024.03
Appendix A: Physical Model Interface Models
Parameter Interface

Additional parameter files and explanations are available on request. Contact TCAD
Support (see Contacting Your Local TCAD Support Team Directly) or Engineering &
Consulting ([email protected]).

Parameter Interface and Equations of SBmob Model

The following polynomial describes the mobility response to the uniaxial stress components
sxx, syy, and szz:

 i
-------- = a 6 i s ii1 + a 5 i s ii2 + a 4 i s ii3 + a 3 i s ii4 + a 2 i s ii5 + a 1 i s ii6 (4)
i
where i is the index for the x-, y-, or z-component of the mobility.
The parameter interface allows you to define the coefficients of the polynomials for the
mobility response to uniaxial stress. Table 17 lists most of these coefficients. Those not
listed can be derived from information in the table.
Table 17 Coefficients of polynomials for mobility response to uniaxial stress

Coefficient Description

aunics100c1106h Coefficient a6 of the hole mobility response to stress in the channel direction
for (100) surface orientation and <110> channel direction. The other five
coefficients are named as follows:
• aunics100c1105h
• aunics100c1104h
• aunics100c1103h
• aunics100c1102h
• aunics100c1101h

aunins100c1101h Coefficient a1 of the hole mobility response to the stress normal to the MIS
interface for (100) surface orientation and <110> channel direction.

auniws100c1001h Coefficient a1 of the hole mobility response to stress in the width direction of
the device for (100) surface orientation and <100> channel direction.

auniws100c1101h Coefficient a1 of the hole mobility response to stress in the width direction of
the device for (100) surface orientation and <110> channel direction.

auniws110c1101h Coefficient a1 of the hole mobility response to stress in the width direction of
the device for (110) surface orientation and <110> channel direction.

Advanced Calibration for Device Simulation User Guide 115

V-2024.03
Appendix A: Physical Model Interface Models
Parameter Interface

All a i coefficients are mole fraction dependent. The coefficients must be defined for each of
the 11 points of the mole fraction discretization, that is, for germanium or silicon content of
0, 0.1, 0.2, ..., 1.0. For example, to set the coefficient a6 for holes with stress applied in the
channel direction with channel orientation 110 and wafer orientation 100, use:
aunics100c1101h=(-0.20156, -0.21273, -0.19696, -0.18257, -0.18225,
-0.17145, -0.16738, -0.16675, -0.16368, -0.15079, -0.15600)

To improve interpolation for high stress and for cases where the mobility response becomes
independent of the stress or behaves linearly, additional coefficients for the mobility
response to uniaxial stress are introduced. In the parameter file, these coefficients have the
same name and syntax as the parameters for the a i coefficients but have higher numbers
(see Table 18).
Table 18 Additional coefficients of polynomials for mobility response to uniaxial stress

Coefficient Description

auniws100c1107h Sets the negative stress value in GPa for the switch between the 6th-order
polynomial and linear slope for the hole mobility response to stress in the
width direction of the device for (100) surface orientation and <110> channel
direction (see Figure 57 on page 117).

auniws100c1108h Sets the slope in 1/Pa for the linear slope for stress values less than
auniws100c1107h for the hole mobility response to stress in the width
direction of the device for (100) surface orientation and <110> channel
direction (see Figure 57).

auniws100c1109h Sets the positive stress value in GPa for the switch between the 6th-order
polynomial and linear slope for the hole mobility response to stress in the
width direction of the device for (100) surface orientation and <110> channel
direction (see Figure 57).

auniws100c11010h Sets the slope in 1/Pa for the linear slope for stress values greater than
auniws100c1109h for the hole mobility response to stress in the width
direction of the device for (100) surface orientation and <110> channel
direction (see Figure 57).

Advanced Calibration for Device Simulation User Guide 116

V-2024.03
Appendix A: Physical Model Interface Models
PMI for Calculating the Sheet Resistance

Figure 57 Use of the ai parameters in fitting uniaxial stress curves

Mobility Enhancement

auniws100c110h9
Linear Slope

6th-order auniws100c110h9
polynomial
szz

PMI for Calculating the Sheet Resistance

For the calculation of sheet resistance, you can use the Sentaurus Device PMI
sheetresistance. In this approach, Sentaurus Device solves the Poisson equation or the
complete set of drift-diffusion equations using an arbitrary doping profile as input. Then, the
PMI calculates the sheet resistance using Equation 5:

1
R s = -----------------------------------
- in /square (5)
e    n  n  dx
t
0
All mobility models that are available in Sentaurus Device can be used together with this
PMI. R s is written as a Sentaurus Workbench design-of-experiments parameter into the log
file of Sentaurus Device and can be used by Sentaurus Workbench for further analysis. The
input syntax for the Sentaurus Device command file is:
CurrentPlot {
PMIModel (
Name = "sheetresistance"
rs_name = "sheetres"
# BoxMin = (-1.0) BoxMax = (0.05) # With this command, you can
# change the integration interval,
# for example, in cases where an
# adjustment of the integration depth
# is necessary.
)
}

where sheetres is the name of the design-of-experiments parameter (default: Rs_sim).

Advanced Calibration for Device Simulation User Guide 117

V-2024.03