HEMT Model

Model’s Documentation

Model Version: 3.0

Technical Report

April, 2020

1
Note to the reader
Dear reader

This document provides a description of the equations and parameters used for
the computer simulation version of the EPFL HEMT model V3.0, developed by
developed by ED-LAB, École Polytechnique Fédérale de Lausanne (EPFL) in
cooperation with Amirkabir University of Technology, Technical University of
Crete (TUC) and IMEC. The EPFL-HEMT model is a scalable and compact
simulation model built on fundamental physical properties of the HEMT
structure. This model is dedicated to the design and simulation of low-voltage,
low-current analog, and mixed analog-digital circuits using heterostructure
technologies. The description concentrates on the intrinsic part of the HEMT,
and is intended to give the model user information on parameter handling and
the actual equations used in the computer simulation.
Any comments and suggestions will be appreciated. Therefore, please do not
hesitate to contact us if you have any queries.

EPFL HEMT Team

2
Introduction:
The introduction of High Electron Mobility field effect Transistors (HEMTs) was
an important milestone in semi-conductor industry mainly because of their
outstanding electrical properties, i.e. high voltage capability, low on-resistance,
high breakdown due to high bandgap voltage, and low capacitance because of
no junction to deplete which make these devices suitable to use in high-power
and high-frequency applications. Given that introducing compact models are an
inseparable part of the circuit simulations, several compact models of this device
have already been developed, Advances Spice Model HEMT (ASM-HEMT), MIT
Virtual Source GaN HEMT (MVSG-HV), empirical models, and table-based
models. besides, to the best of our knowledge, none of these models was
validated with respect to TCAD simulations.

The EPFL model is a design-oriented charge-based model for dc operation of

AlGaAs/GaAs and AlGaN/GaN-based high-mobility field-effect transistors. The
intrinsic model is physics-based and the central concept is based on charge
linear approximation.

In this context, a physics-based compact model for HEMT developed in EPFL is

proposed in chapter 1. In addition, the Verilog-A equations and parameter
extractions are further clarified in chapter 2 and 3.

3
Chapter 1
Introduction to core model:

In this model HEMT is considered as a generalized MOSFET which means all the
techniques have been used in MOSFET compact models can be reclaimed for
GaN/GaAs HEMTs as well. The core model is developed for an ideal long device
and other effects will be included afterwards to correct the main core model.

The typical device structure of GaN/GaAs HEMT device is shown in Fig. 1 It

consists of a large (AlGaAs/AlGaN) and a small (GaAs/GaN) band gap
semiconductor. By bonding these two substances together, a triangular
quantum well will be created at the interface. Both Fig. 2a and Fig. 2b for
conduction and valance band are sketched based on this claim. In this case, the
AlGaAs/AlGaN and GaAs/GaN regions are respectively n-doped with 𝑁𝐷 and p-
doped with NA . Note that parameters and symbols which have been used are
defined in Table 1.1 and Table 2.1.

First and foremost, the electrostatic potential must satisfy the Poisson equation.

𝑑 2 𝜓(𝑥) 𝜌(𝑥)
= −
𝑑𝑥 2 𝜖1,2

𝜖1 and 𝜖2 are respectively the permittivity of AlGaAs (AlGaN) and GaAs (GaN),
𝜓(𝑥) is the electrostatic potential and 𝜌(𝑥) is the local charge density. By
solving the above differential equation the general

4
 Figure 1- 3D schematic view of HEMT[7]

Figure 2 - Sketch of the energy band diagram for a GaN HEMT (a,b)[7].

expression of the potential profile across the heterostructure 𝜓𝑛 (𝑥),

𝜓𝑝 (𝑥) along with the electrical fields 𝜀𝑛 (𝑥) , 𝜀𝑝 (𝑥) will be obtained.

𝑞𝑁𝐷 2 𝑞𝑁𝐷
𝜓𝑛 (𝑥) = − 𝑥 + 𝑎𝑥 + 𝑏, 𝜀𝑛 (𝑥) = 𝑥−𝑎
2𝜖1 𝜖1
𝑞𝑁𝐴 2 𝑞𝑁𝐴
𝜓𝑝 (𝑥) = − 𝑥 + 𝑎′ 𝑥 + 𝑏 ′ , 𝜀𝑝 (𝑥) = 𝑥 − 𝑎′
2𝜖2 𝜖2

5
 a and 𝑎′ are coefficients which will be calculated by imposing the boundary
conditions at 𝑥1 and 𝑥2 . Firstly, As the electrical field vanishes beyond 𝑥2 , giving
𝑞𝑁
𝑎′ = ( 𝐴⁄𝜖2 ) 𝑥2 . The difference 𝜓𝑝 between 0 and 𝑥2 gives the potential drop
𝜓2 depends on the depletion depth in the substrate which is identified as 𝜓𝑠 .
𝑄
Secondly, by considering the electrical field of 𝜀𝐺 at 𝑥1 (𝜀𝐺 = 𝜀(𝑥1 ) = 𝐺⁄𝜖1 =
𝑞𝑛𝐺
⁄𝜖1 )with definition of gauss’s surface between the gate Schottky contact
and the bulk 𝜓𝑛 (𝑥) can be written as:

𝑞𝑁𝐷 2 𝑞
𝜓𝑛 (𝑥) = − 𝑥 + (𝑥1 𝑁𝐷 − 𝑛𝐺 )𝑥 + 𝑏
2𝜖1 𝜖1

As the same way, the difference 𝜓𝑛 between 0 and 𝑥1 gives the potential drop
in the AlGaN barrier 𝜓1 .

Table 1.1 Physical parameters of GaN/GaAs HEMT device

Parameter Default Value Unit Short Description

𝑵𝑫 𝟓 × 𝟏𝟎𝟏𝟔 𝒄𝒎−𝟑 Doping of AlGaN layer

𝑵𝐴 𝟏 × 𝟏𝟎𝟏𝟔 𝒄𝒎−𝟑 Doping of GaN layer

𝒕𝒃 = −𝑥1 10 to 200 nm GaN thickness

W 𝟏. 𝟐𝟓 𝝁𝒎 Channel Width

𝑳𝑮 𝟎. 𝟗, 𝟏, 𝟏𝟎 𝝁𝒎 Channel Length

x 0.3, 0.25 - Al mol fraction

Hence, the depletion approximations are as given below:

𝑞𝑁𝐷 2 𝑞𝑛𝐺
𝜓1 = 𝑥 − 𝑥
2𝜖1 1 𝜖1 1
𝑞𝑁𝐴 2
𝜓2 = 𝑥 = 𝜓𝑠
2𝜖2 2

6
Given that Kirchhoff’s voltage law can be simply understood from Fig. 2:
∆𝐸𝐶
𝜓1 + 𝜓2 = 𝑉𝐺𝐵 − 𝜙𝐵 + , ∆𝐸𝐶 = 𝐸𝐶𝐴𝑙𝐺𝑎𝑁 (0) − 𝐸𝐶𝐺𝑎𝑁 (0)
𝑞

Along with the charge neutrality expression:

𝜎𝑝𝑜𝑙
𝑛𝐺 − 𝑥1 𝑁𝐷 − 𝑥2 𝑁𝐴 − 𝑛𝑐ℎ + =0
𝑞
The channel charge can be obtained:

𝑄𝑐ℎ = 𝑞𝑛𝑐ℎ = 𝐶𝑏 (−𝑉𝐺𝐵 + 𝑉𝐴 + 𝜓𝑠 + 𝛾√𝜓𝑠 )

∆𝐸𝐶 𝑞𝑁𝐷 2
𝑉𝐴 = − + 𝜙𝐵,𝑒𝑓𝑓 − 𝑥
𝑞 2𝜖1 1
𝑥1
𝛾 = √2𝜖2 𝑁𝐴 𝑞
𝜖1
𝜖1
𝐶𝑏 = −
𝑥1
And a linear approximation of the channel charge density gives:

𝑄𝑐ℎ = 𝐶𝑏 𝑛𝑞 (𝜓𝑠 − 𝜓𝑝 )
Where:

𝑉𝐺𝐵 − 𝑉𝐴 1 1 𝛾
𝜓𝑝 = 𝑉𝐺𝐵 − 𝑉𝐴 − 𝛾 2 (√ + − ), 𝑛𝑞 = 1 +
𝛾2 4 2 2√𝜓𝑝

Here the assumption is that the channel is densely occupied by electrons

so that the ground energy level in the quantum well enters to the channel
carrier density 𝑛𝑐ℎ formula.

𝐸𝐹 − 𝐸0
𝑛𝑐ℎ = 𝐷𝑜𝑆2𝐷 × 𝑈𝑇 × Ln (1 + 𝑒 𝐾𝐵 𝑇 )

7
 Table 2.1 List of symbols used in the model
Symbol Unit Short Description

𝑵𝒄𝒉 , 𝑵𝒈 𝒎−𝟐 Channel and Gate charge density

q C Electron charge

𝑈𝑇 V Thermal Voltage

𝝍, 𝝍𝒔 V Potential profile and Surface potential

𝑽𝒄𝒉 V Shift in the quasi-fermi level

𝑬𝑪 eV Bottom of the conduction Band

𝑬𝑽 eV Top of the valence band

𝑬𝒇 eV Fermi level

𝑬𝟎 , 𝑬𝟏 eV Bottom of the conduction Band

𝒈𝒎 , 𝒈𝒅𝒔 S Input and output transconductances

𝑰𝒅𝒔 m Drain to source current

𝑽
𝜀 Electric Field
𝒎
𝒖𝟏 - Experimentally determined parameter

𝑪
𝐷 Electric displacement field
𝒎𝟐
𝑭
𝑪𝒃 Equivalent barrier capacitance
𝒎𝟐
𝛾 𝑽𝟏/𝟐 Body factor

𝑛𝑞 - The n factor

𝑉𝐴 , 𝜓𝑝 V Offset voltage and pinch-off voltage

𝑄𝑆𝑃 C Specific charge

𝐼𝑆𝑃 A Specific current

8
Here 𝐷𝑜𝑆2𝐷 is the density of states of a 2D system obtained by 𝐷𝑜𝑆2𝐷 =
𝑚∗⁄ where 𝑚∗ is the electron effective mass and ℏ is the reduced plank
𝜋ℏ2
constant. As we know, the ground energy level in the triangular-like quantum
well has direct connection with the channel density.

𝐸0 − 𝐸𝐶 (0) = 𝑞𝑉0 =  √𝑛𝑐ℎ 2

3

 is estimated from Shubnikov-de Haas oscillations or by cyclotron resonance

experiments and 𝑉0 is the relative ground level potential.

From what could be expressed from Fig. 2, 𝐸𝐹 − 𝐸0 can be written:

𝐸𝐹 − 𝐸0 = 𝐸𝐹 − 𝐸𝑉 (𝑥2 ) + 𝐸𝑉 (𝑥2 ) − 𝐸𝐶 (𝑥2 ) + 𝐸𝐶 (𝑥2 ) − 𝐸𝐶 (0) + 𝐸𝐶 (0)

− 𝐸0

𝐸𝐹 − 𝐸0 𝑁𝐴 𝐸𝑔 𝐸𝐶 (0) − 𝐸0
= 𝑉𝐹0 = − 𝑈𝑇 𝐿𝑛 ( ) − + + 𝜓𝑠
𝑞 𝑁𝑉 𝑞 𝑞
𝑁𝐴 𝐸𝑔
= − 𝑈𝑇 𝐿𝑛 ( ) − + 𝜓𝑠 − 𝑉0
𝑁𝑉 𝑞
Based on all the calculation given in the reference paper, the main equation is
written:

𝑛𝑐ℎ 𝑞𝑛𝑐ℎ 3
𝑈𝑇 𝐿𝑛 (𝑒 𝐷𝑜𝑆2𝐷 𝑈𝑇 − 1) + + 𝑢1 √𝑛𝑐ℎ 2 = 𝜓𝑝 − 𝑉
𝐶𝑏 𝑛𝑞

V is the charge-based relation can be expressed:

𝑁𝐴 𝐸𝑔 
𝑉 = 𝑈𝑇 𝑙𝑛 ( ) + + 𝑉𝑐ℎ , 𝑢1 =
𝑁𝑉 𝑞 𝑞

9
 Since, the drain-source current consists of drift and diffusion processes the
normalized current equation is written:

𝑉
𝐼 = (𝑞𝑐ℎ 2 + 𝑞𝑐ℎ )| 𝑆𝐵 = 𝑞𝑆 2 + 𝑞𝑆 − 𝑞𝐷 2 − 𝑞𝐷
𝑉𝐷𝐵

Where 𝑞𝑆 and 𝑞𝐷 are respectively the normalized channel charge in source and
drain.
By imposing Newton-Raphson numerical approximation technique on the main
equation, channel density can be calculated. The current derivation equation
then can be achieved by calculating 𝑞𝑆 and 𝑞𝐷 from 𝑛𝑐ℎ .

Based on what was said at the beginning, in this context, HEMTs are considered
as a generalized MOSFETs, Table 3 is the evidence of the claim.

Table 3.1 A useful comparison between EPFL HEMT model basic equations and MOSFET model

Equation
EPFL HEMT Model MOSFET Model
Type

Charge- 𝑛𝑐ℎ 𝑞𝑛𝑐ℎ 3

Based 𝑈𝑇 𝑙𝑛(𝑒 𝐷𝑜𝑆2𝐷𝑈𝑇 − 1) + + 𝑢1 √𝑛𝑐ℎ 2 = 𝜓𝑝 − 𝑉 𝟐𝒒𝒊 + 𝒍𝒏(𝒒𝒊 ) = 𝝂𝒑 − 𝝂
𝐶𝑏 𝑛𝑞
Expression

Charge 𝑄𝑐ℎ = 𝐶𝑏 𝑛𝑞 (𝜓𝑠 − 𝜓𝑝 ) 𝑸𝒄𝒉 = 𝑪𝒃 𝒏𝒒 (𝝍𝒔 − 𝝍𝒑 )

Linearization
Scheme

Current
𝑰 = 𝒒𝑺 𝟐 + 𝒒𝑺 − 𝒒𝑫 𝟐 − 𝒒𝑫 𝑰 = 𝒒𝑺 𝟐 + 𝒒𝑺 − 𝒒𝑫 𝟐 − 𝒒𝑫
Derivation

Now considering that two main standard compact models were elected by CMC,
it is worth to give a simple but fair comparison between these two. And the EPFL
HEMT model in Table 4.

10
 Table 4.1 A comparison between three proposed HEMT compact models

EPFL HEMT Model ADM-HEMT Model MVSG-HEMT Model

Charge/Potential-based Charge-based Surface potential-based Charge-based

Type Semi-impirical near

Physics-based Physics-based
ballistic

Q-V and Q-I expressions Unique function for Q-v and Unique function
Piecewise function
Q-I

Simplicity design Simple for hand calculation complex complex

Similarity to MOSFET’s model Inherits from regular silicon

× ×
MOSFETs

Compatibility with RF gm/I

design methodology  × ×

Design-oriented  × ×

Validated y TCAD  × ×

11
Chapter 2
List of parameters:

In this chapter a table format of all the parameters which have been used
in the Verilog-A code of the model is prepared along with a brief but useful
description on their use.

General Parameters:
Table 1.2 General Parameters

Parameter Default Value Unit Short Description

𝑻𝑵𝒐𝒎 300 K Temperature

T 300 K Temperature

M 1.0 - Parallel multiple device number

NS 1.0 - Series multiple device number

𝑪𝒃 0.0026 𝑭 AlGaN Capacitance per Unit Area

𝒎𝟐

𝒙𝒋 100n m Depth of junction (source and drain)

𝑵𝑨 𝟒 × 𝟏𝟎𝟐𝟎 - Dopping of GaN Layer

𝑵𝑫 𝟑 × 𝟏𝟎𝟐𝟏 - Dopping of AlGaN Layer

12
 𝑵𝑫𝑺 𝟓 × 𝟏𝟎𝟐𝟒 - Doping of Source/Drain

xx 0.3 - Percentage of Al in AlxxGa(1-xx)N

𝝈𝒑𝒐𝒍 0.0120 - Polarization Charge Density in AlGaN/Gan Interface

𝝓𝑩 1.1 - Gate Schottky Barrier

𝒙𝟏 -30n m AlGaN Layer Thickness

𝝁 0.85 Permeability

𝟏
IBA 50𝑴 First impact ionization coefficient
𝒎

𝑽
IBB 400M Second impact ionization coefficient
𝒎

IBN 1.0 V Saturation voltage factor for impact ionization

Process Geometrical Scaling Factors:

This group of parameters include information about length and width of
the device. The SCALE parameter affects both dimensions
but the XL and XW only optically effect respectively on the length and width of
the HEMT device. In the case of the devices with small length or width, the
model uses a set of parameters to calculate the effective dimensions. DL is the
difference between the effective and the drawn length while DLC is considered
to better tune the length in dynamic behavior. DW and DWC parameters woks
as the same way on the device width. Finally, WDL and LDW are defined as
scaling factors for the narrow and short channel effect.

Table 2.2 Process Geometrical Scaling Factor

Parameter Default Value Unit Short Description

L 1𝝁 m Gate’s Length

𝑳𝒂𝒓𝒅 7𝝁 m Channel length

13
 𝑳𝒂𝒓𝒔 2𝝁 m Channel length

𝑳𝒌 𝟎. 𝟒𝝁 m Reverse short channel effect characteristic length

W 3050 𝝁 m Gate’s Width

DL -10n m Difference between effective and drawn gate length

DW -10n m Difference between effective and drawn gate width

WDL 0.0 𝒎𝟐 Width Scaling for narrow devices of 𝑳𝒆𝒇𝒇

LDW 0.0 𝒎𝟐 Length Scaling for short devices of 𝑾𝒆𝒇𝒇

Tuning the difference of effective gate length

DLC 0.0 m
between current and capacitance behaviour

Tuning the difference of effective gate width

DWC 0.0 m
between current and capacitance behaviour

Velocity Saturation and CLM Related Parameters:

The velocity saturation effect (VSE) defines the saturation behavior of the
short channel transistor. The critical velocity of the VSE appears in parameter
UCRIT. DELTA is defined to form a variable order and may take a value strictly
between 1 and 2 which its value is different in terms of P-type and N-type
devices.
The modulation of the effective length can be controlled by 𝜆. Moreover, the
ACLM is a parameter which can change if fitting is not adequate.

Table 3.2 Velocity Saturation and CLM Related Parameters

Parameter Default Value Unit Short Description

𝐕
UCRIT 8M Critical velocity of velocity saturation ef f ect
𝒎

Depletion length coef f icient (channel length

𝝀 3.5 -
modulation)

𝑾𝝀 0.0 m Width scaling f or LAMBDA

14
 Fine Tuning Parameter f or Chanel Length
ACLM 1.0 -
Modulation

𝝀𝝎 𝟎. 𝟐𝝁 -

𝜷𝒄𝒍𝒎 1.0 -

DELTAA 1.0 - order of VSAT model between [1,2]

Vertical Field Mobility Effect:

Reducing the length of the transistor needs to include mobility reduction.
the effective mobility which is the division of drift velocity to the longitudinal
electric field specifically relates the vertical electric field of the channel. E0 is the
mobility reduction coefficient and ETA is a constant and differs in N-type to P-
type devices.

Table 4.2 Vertical Field Mobility Effect

Parameter Default Value Unit Short Description

𝑽 Parameter For Mobility Reduction Due to Vertical

E0 𝟐𝟓𝑴
𝒎 Field Effect

𝑻𝒉𝒆𝒕𝒉𝒂𝟑 1.0 - Vertical Field Effect on saturation Velocity

Matching Parameters:
The mismatch between transistors with similar geometries is deliberately
employed with a statistical variation. This variation is related to the square root
of the area of the gate. The coefficients have variety of range of instances, such
as threshold voltage (AVA), mobility (AKP), and body factor (AGAMMA).

Table 5.2 Matching Parameter

15
Parameter Default Value Unit Short Description

AVT0 1𝝁 Vm Area related threshold voltage mismatch parameter

AKP 1𝝁 m Area related gain mismatch parameter

AGAMMA 1𝝁 𝑽𝟐 𝒎 Matching Parameter for Body Factor (Gamma)

Drain Induced Barrier Lowering Parameters:

Drain induced barrier lowering effect (DIBL) is a short-channel effect.
Actually, the pinch-off voltage can be affected by the drain potential and due to
this effect a reduction in threshold voltage can occur at high drain voltage.

Table 7.2 Drain Induced Barrier Lowering Parameters

Parameter Default Value Unit Short Description

ETAD 1.7 - DIBL Coef f icient

ETAD2 0 - DIBL Coef f icient

WETA 0.0 - Short-channel effect coefficient

LETA 0.0 - Short-channel effect coefficient

Temperature Parameters:
Doubtlessly, temperature can change the behavior of HEMT devices.
Many parameters are directly related to temperature so there is a need to
define parameters that can be used for modeling this effect. In Table 8.2 all the
temperature- related parameters are displayed.
Table 8.2 Temperature Parameters

Parameter Default Value Unit Short Description

𝑽
TCV 0.0 Temperature Scaling f or Threshold Voltage
𝒌

BEX 1.6 - Temperature Scaling f or Mobility

ETAMOB 1.0 - Mobility reduction f actor due to temp

16
ETAUC 0.0 - Critical Field reduction f actor due to temp

Longitudinal critical f ield temperature

UCEX 1.0 -
exponent

UE0 0.5 - Vertical Field temperature exponent

THETA3 1.0 - Vertical Field Ef f ect on saturation Velocity

ETAX 0.2 -

gg 1.6f

𝟏
IBBT 0.9m Saturation voltage factor for impact ionization
𝒌

𝒌
𝑹𝒕𝒉 45 Thermal resistance
𝒘

𝒌
𝑪𝒕𝒉 1n Thermal capacitance
𝒘

temperature dependence of 2-DEG charge density

𝒏𝒔𝟎 5× 𝟏𝟎𝟏𝟕
in the drain or source side access region

low field mobility parameter for access region at

𝝁𝟎𝒂𝒄𝒄 0.155
room temperature

𝑽𝒔𝒂𝒕𝟎𝒂𝒄𝒄 3.4K saturation velocity extracted at TNOM

𝑹𝒄𝒔 4.1m

𝑲𝒕𝒓𝒔 0.0

𝑲𝒕𝒓𝒅 0.0

𝑹𝒄𝒅 4.1m

𝑲𝒕𝒉𝟏 0.0

𝑲𝒕𝒉𝟐 0.0

𝑲𝒕𝒉𝟑 0.5

𝒈𝒔𝒂𝒓 2.0

17
 𝜼𝒎𝒐𝒃 1 - mobility reduction factor due to temp

𝜼𝑼𝑪 1 - Critical Field reduction factor due to temp

Velocity Overshoot Parameters:

The average velocity of electrons increases in short-channel devices tends
to exceeding steady-state saturation velocity and the carrier behavior in the
channel would be quasi-ballistic, mainly cause the velocity overshoot.
Table 9.2 Velocity Overshoot

Parameter Default Value Unit Short Description

Lomega 0.5𝝁 - Velocity overshoot coefficient

18
Chapter 3
Equations:

This chapter focuses on equation which have been used in the

model process. For more details, please, look at the references
mentioned at the end of the draft.

3.1 Basic Voltages:

𝑉𝐺 = 𝑉𝑔𝑎𝑡𝑒 − 𝑉𝑏𝑢𝑙𝑘 3.1.1

𝑉𝑆 = 𝑉𝑆𝑎𝑟 − 𝑉𝑏𝑢𝑙𝑘 3.1.2

𝑉𝐷 = 𝑉𝑑𝑎𝑟 − 𝑉𝑏𝑢𝑙𝑘 3.1.3

𝑉𝐷𝐴𝑅 = 𝑉𝐷 − 𝑉𝑑𝑎𝑟 3.1.4

𝑉𝑆𝐴𝑅 = 𝑉𝑠𝑎𝑟 − 𝑉𝑆 3.1.5

3.2 Helping Equations:

1 3.2.1
MAX smooth (a, b, c) = (a + b + √(a − b)2 + c) = MX s (a, b, c)
2
𝐴𝐵𝑆(𝑥) = √𝑥 2 + 0.4𝜇 3.2.2

3.3 General Equations:

𝑻
𝑼𝑻 = 𝟎. 𝟎𝟐𝟓𝟗 × 3.3.1
𝑻𝑵𝑶𝑴
𝐆𝐚𝐍
𝐭 𝐆𝐚𝐍 =
𝐂𝐛 3.3.2

𝐭 𝐀𝐥𝐆𝐚𝐍 = −𝐱𝟏
3.3.3

19
𝐋𝐂 = √𝐭 𝐆𝐚𝐍 𝐱𝐣
3.3.4

𝒘𝝀
𝝀𝒈 = 𝝀 × (𝟏 + ) 3.3.5
𝒘

𝑳𝑪𝝀 = 𝑳𝑪 × 𝝀𝒈 3.3.6

𝑾𝑬𝑻𝑨
𝒕𝑮𝒂𝑵𝑾 = 𝟑 × 3.3.7
𝑻𝑨𝒍𝑮𝒂𝑵
𝑳𝑬𝑻𝑨
𝒕𝑮𝒂𝑵𝑳 = 3.3.8
𝑻𝑨𝒍𝑮𝒂𝑵
−𝑬𝒈𝑮𝒂𝑵
𝑵𝒊𝑮𝒂𝑵 = √𝑵𝒄𝑮𝒂𝑵 × 𝑵𝑽 × 𝒆𝟐×𝑲𝑩 ×𝑻 3.3.9

𝑵𝑫𝑺
𝝋𝑩 = 𝑼𝑻 × 𝒍𝒏( ) 3.3.10
𝑵𝒊𝑮𝒂𝑵
𝑵𝑨
𝝋𝑻 = 𝑼𝑻 × 𝒍𝒏( ) 3.3.11
𝑵𝒊𝑮𝒂𝑵
𝑵𝑨 𝑬𝒈𝑮𝒂𝑵
𝝋𝟎 = 𝑼𝑻 × 𝒍𝒏( ) + 3.3.12
𝑵𝑽 𝒒

𝝋𝑻𝑯𝑹 = 𝝋𝟎 – 𝝋𝑻 3.3.13

−∆𝑬𝑪 𝒙𝟏 × 𝝈𝒑𝒐𝒍 𝒒 × 𝑵𝑫 × 𝒙𝟏 𝟐
𝑽𝑨𝟏 = + 𝝋𝑩 + − 3.3.14
𝒒 𝜺𝑨𝒍𝑮𝒂𝑵 𝟐 × 𝜺𝑨𝒍𝑮𝒂𝑵
𝟎. 𝟐𝐦𝐞
𝐃𝐎𝐒𝟐𝐃 =
𝟏
( 𝟏. 𝟎𝟓𝟒𝟓𝟗𝟏𝟎−𝟑𝟒 )𝟐 3.3.15
𝐪

𝒕𝟐
𝑬𝒈𝑮𝒂𝑵 = 𝟑. 𝟓𝟎𝟕 × 𝒒 − (𝟎. 𝟗𝟎𝟗 × 𝟏𝟎−𝟑 ) × ×𝒒 3.3.16
𝒕 + 𝟖𝟑𝟎

𝒕𝟐
𝑬𝒈𝑨𝒍𝑵 = 𝟔. 𝟐𝟑 × 𝒒 − (𝟏. 𝟕𝟗𝟗 × 𝟏𝟎−𝟑 ) × ×𝒒 3.3.17
𝒕 + 𝟏𝟒𝟔𝟐

𝑬𝒈𝑨𝒍𝑮𝒂𝑵 = 𝑬𝒈𝑨𝒍𝑵 × 𝒙𝒙 + 𝑬𝒈𝑮𝒂𝑵 × (𝟏 − 𝒙𝒙) × 𝟏. 𝟑 × 𝒙𝒙 × (𝟏 − 𝒙𝒙) × 𝒒 3.3.18

𝝐𝑨𝒍𝑮𝒂𝑵 = (𝟖. 𝟓 × 𝒙𝒙 + 𝟖. 𝟗 × (𝟏 − 𝒙𝒙)) × 𝝐𝟎 3.3.19

∆𝑬𝑪 = 𝟎. 𝟕 × (𝑬𝒈𝑨𝒍𝑮𝒂𝑵 − 𝑬𝒈𝑮𝒂𝑵 ) 3.3.20

20
 𝜺𝑨𝒍𝑮𝒂𝑵
𝑪𝒃𝟏 = − 3.3.21
𝒙𝟏

−𝒙𝟏
𝜸= × √𝟐 × 𝜺𝑮𝒂𝑵 × 𝑵𝑨𝟎 × 𝒒 3.3.22
𝜺𝑨𝒍𝑮𝒂𝑵

𝑽𝑻𝑶 = 𝑽𝑨𝟏 – 𝝋𝑻𝑯𝑹 – 𝜸 × √𝝋𝑻𝑯𝑹 3.3.23

𝐍𝟎 = 𝐃𝐎𝐒𝟐𝐃  𝐔𝐓 3.3.24
𝑵𝑨 𝑬𝒈𝑮𝒂𝑵
𝑽𝒔𝒔 = 𝑽𝒔 + 𝑼𝑻 × 𝒍𝒏 ( )+ 3.3.25
𝑵𝑽 𝒒
𝑽 𝑮∗ = 𝑽𝑮 − 𝑽𝑻𝑶𝑺 + 𝝋𝑻 3.3.26

𝑽𝑮∗ + √𝑽𝑮∗ 𝟐 + 𝟑𝟐 × 𝑼𝑻 𝟐 3.3.27

𝑽 𝑮′ =
𝟐
𝜸𝟐𝒔 𝜸𝒔
𝑽𝒑𝒊𝒏𝒄𝒉𝟎 = 𝑽𝑮′ − 𝜸𝒔 × (√𝑽𝑮′ + − ) 3.3.28
𝟒 𝟐
𝑻𝑮𝑨𝑵𝑾
𝑾𝑬𝑻𝑨𝑾 = 3.3.29
𝑾𝒆𝒇𝒇
𝑻𝑮𝑨𝑵𝑳
𝑳𝑬𝑻𝑨𝑾 = 3.3.30
𝑳𝒆𝒇𝒇
𝜸∗
= 𝜸𝒔 − 𝑳𝑬𝑻𝑨𝑳

𝟐 𝟐 𝟐 𝟐
√𝝋𝑻 + 𝑽𝑺 + √(𝝋𝑻 + 𝑽𝑺 ) + 𝟏𝟔𝑼𝒕 √𝝋𝑻 + 𝑽𝑫 + √(𝝋𝑻 + 𝑽𝑫 ) + 𝟏𝟔𝑼𝒕
× + 3.3.31
𝟐 𝟐

( )
+ 𝑾𝑬𝑻𝑨𝑾 × √𝑽𝒑𝟎 + 𝝋𝑻 − 𝟎. 𝟏𝑽𝑻𝑶𝒔

𝜸∗ + √𝜸∗ 𝟐 + 𝟎. 𝟏𝑼𝑻 3.3.32

′
𝜸 =
𝟐
𝜸′ 𝟐 𝜸′ 3.3.33
𝑽𝒑𝒊𝒏𝒄𝒉 = 𝑽𝑮′ − 𝜸 × (√𝑽𝑮′ +
′
− )
𝟒 𝟐
𝜸
𝒏𝒒 = 𝟏 + 3.3.34
𝟐 × √𝑽𝒑𝒊𝒏𝒄𝒉 + 𝟒 × 𝑼𝑻
𝑸𝒔𝒑 = −𝟐 × 𝒏𝒒 × 𝑪𝒃 × 𝑼𝑻 3.3.35
𝝁 𝟐
𝒘𝒆𝒇𝒇𝑪
𝑰𝒔𝒑 = 𝟐 × 𝒏𝒒 × × 𝑪 𝒃 × 𝑼 𝑻 × 3.3.36
(𝟏 + 𝑬𝑻𝑨𝑴𝑶𝑩 × (𝑹𝒂𝒕𝒊𝒐𝑻 − 𝟏))𝑩𝑬𝑿 𝒍𝒆𝒇𝒇𝑪
𝑽𝒅𝒅 = 𝑽𝑫 + 𝑽𝑺 3.3.37
𝑸𝒔𝒍 = 𝑪𝑶𝑹𝑬(𝑽𝑺 , 𝟎, 𝑽𝒑𝒊𝒏𝒄𝒉 , 𝒏𝒒 , 𝑪𝒃 , 𝑫𝑶𝑺𝟐𝑫 , 𝒈𝒈, 𝒒, 𝑼𝑻 , 𝑵𝑽 , 𝑬𝒈𝑮𝒂𝑵 , 𝑵𝑨 , 𝝋𝑻 ) 3.3.38

21
 𝑸𝒔𝒍
𝑷𝒔𝒖𝒓𝒇 = 𝑽𝒑𝒊𝒏𝒄𝒉 − 3.3.39
𝑪𝒃 × 𝒏 𝒒
𝑸𝒔𝒍𝟐 = 𝑪𝑶𝑹𝑬(𝑽𝑺 , 𝟎, 𝑷𝒔𝒖𝒓𝒇 , 𝒏𝒒 , 𝑪𝒃 , 𝑫𝑶𝑺𝟐𝑫 , 𝒈𝒈, 𝒒, 𝑼𝑻 , 𝑵𝑽 , 𝑬𝒈𝑮𝒂𝑵 , 𝑵𝑨 , 𝝋𝑻 ) 3.3.40
𝑸𝒔𝒍𝟐
𝑷𝒔𝒖𝒓𝒇𝟐 = 𝑷𝒔𝒖𝒓𝒇 − 3.3.41
𝑪𝒃 × 𝒏 𝒒
𝒅𝒔𝒎𝒊𝒏 = 𝑫𝑺𝑴(𝑷𝒔𝒖𝒓𝒇𝟐 , 𝑽𝑫 , 𝑳𝒆𝒇𝒇 , 𝝋𝑻 , 𝝋𝑩 , 𝑻𝑨𝒍𝑮𝒂𝑵 , 𝑵𝑨 , 𝜺𝑮𝒂𝑵 , 𝒒, 𝝅, 𝑬𝑻𝑨𝑫, 𝑬𝑻𝑨𝑫𝟐) 3.3.42
𝒔𝒊𝒎𝒊𝒏 = 𝑷𝒔𝒖𝒓𝒇𝟐 + 𝒅𝒔𝒎𝒊𝒏 3.3.43
𝜸
𝜸𝑩 = 3.3.44
√𝑼𝑻
𝑽𝒑𝒊𝒏𝒄𝒉
𝑽𝒑𝒙 = 3.3.45
𝑼𝑻
𝑸𝒔𝒔 = 𝑪𝑶𝑹𝑬(𝑽𝑺 , 𝑫𝑰𝑩𝑳𝒙, 𝑽𝑷𝒊𝒏𝒄𝒉 , 𝒏𝒒 , 𝑪𝒃 , 𝑫𝑶𝑺𝟐𝑫 , 𝒈𝒈, 𝒒, 𝑼𝑻 , 𝑵𝑽 , 𝑬𝒈𝑮𝒂𝑵 , 𝑵𝑨 , 𝝋𝑻 ) 3.3.46
𝑸𝒔𝒔
𝒒𝒔𝒔 = 3.3.47
𝑸𝒔𝒑
𝒊𝒇𝒇 = 𝒒𝒔𝒔 𝟐 + 𝒒𝒔𝒔 3.3.48

3.4 Temperature related Equations:

∆𝑻 = 𝑻 − 𝑻𝑵𝑶𝑴 3.4.1
𝑻 3.4.2
𝑹𝒂𝒕𝒊𝒐𝑻 =
𝑻𝑵𝑶𝑴
𝑽𝑻𝑶𝑻 = 𝑽𝑻𝑶 + 𝑻𝑪𝑽 × ∆𝑻 3.4.3
𝑼𝑪𝑹𝑰𝑻𝑻 = 𝑼𝑪𝑹𝑰𝑻 × (𝟏 − 𝑬𝑻𝑨𝑼𝑪 × (𝑹𝒂𝒕𝒊𝒐𝑻 − 𝟏))𝑼𝒄𝒆𝒙 3.4.4
𝑰𝑩𝑩𝑻𝑻 = 𝑰𝑩𝑩 × (𝟏 + 𝑰𝑩𝑩𝑻 × ∆𝑻 ) 3.4.5
𝟏 3.4.6
𝒊𝒏𝒗𝒖𝒄𝒓𝒊𝒕 =
𝒖𝒄𝒓𝒊𝒕𝑻
𝑳𝑪𝒖𝒄𝒓𝒊𝒕 = 𝑳𝑪 × 𝑼𝑪𝑹𝑰𝑻𝑻 3.4.7
𝑳𝑪𝑰𝑩𝑩 = 𝑳𝑪 × 𝑰𝑩𝑩𝑻 3.4.8
𝑰𝑩𝑨 3.4.9
𝑰𝑩𝑨𝑰𝑩𝑩 = 𝑰𝑩𝑩
𝑻
𝟏 3.4.10
𝝁𝒕 = 𝝁 × 𝑩𝑬𝑿
(𝟏 + 𝜼𝑴𝑶𝑩 × (𝒓𝒂𝒕𝒊𝒐𝑻 − 𝟏))

3.5 DSM Function Equations:

𝑾𝑫𝑳 3.5.1
𝑳𝒆𝒇𝒇 = 𝑳 + 𝑫𝑳 +
𝑾
𝑳𝑫𝑾 3.5.2
𝑾𝒆𝒇𝒇 = 𝑾 + 𝑫𝑾 +
𝑳
𝑳𝒆𝒇𝒇𝑪 = 𝑳𝒆𝒇𝒇 + 𝑫𝑳𝑪 3.5.3
𝑾𝒆𝒇𝒇𝑪 = 𝑾𝒆𝒇𝒇 + 𝑫𝑾𝑪 3.5.4

3.6 Denormalization Equations:

𝟏 3.6.1
𝑨𝑾𝑳 =
√𝑾𝒆𝒇𝒇 × 𝑳𝒆𝒇𝒇

22
 𝑨𝑾𝑳 × (𝑨𝑽𝑻𝑶 − 𝟏𝟎−𝟔 ) + 𝑽𝑻𝑶𝑻 , 𝑨𝑽𝑻𝑶 ≠ 𝟏𝟎−𝟔 3.6.2
𝑽𝑻𝑶𝑺 = {
𝑽𝑻𝑶𝑻 , 𝑶𝒕𝒉𝒆𝒓𝒘𝒊𝒔𝒆
−𝟔
𝑮𝑨𝑴𝑴𝑨 + (𝑨𝑮𝑨𝑴𝑴𝑨 − 𝟏𝟎 ) × 𝑨𝑾𝑳, 𝑨𝑮𝑨𝑴𝑴𝑨 ≠ 𝟏𝟎−𝟔 3.6.3
𝑮𝑨𝑴𝑴𝑨𝑺 = {
𝑮𝑨𝑴𝑴𝑨, 𝑶𝒕𝒉𝒆𝒓𝒘𝒊𝒔𝒆

3.7 DSM Function Equations:

𝜺𝑮𝒂𝑵 3.7.1
𝑾𝒅 = √𝟐 × × (𝑽𝒑𝒊𝒏𝒄𝒉 + 𝝋𝑻 )
𝒒 × 𝑵𝑨
𝟐 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 3.7.2
𝒃= × (𝝋𝑩 − 𝑽𝒑𝒊𝒏𝒄𝒉 ) × 𝒄𝒐𝒔( )× × (−𝟏
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝟐
+ 𝒄𝒐𝒔( )) + ( ) × (𝝋𝑩 − 𝑽𝒑𝒊𝒏𝒄𝒉 )
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝟐
× 𝒔𝒊𝒏( )× × 𝒔𝒊𝒏( )+
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵
× (𝝋𝑩 − 𝑽𝑷𝒊𝒏𝒄𝒉 ) × 𝒄𝒐𝒔( ) × (𝒔𝒊𝒏( )
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
×( ) − (𝒕 × 𝒄𝒐𝒔( )×
𝒏𝝅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅
𝟐 𝝋𝑩 − 𝑽𝒑𝒊𝒏𝒄𝒉 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐
+ × × 𝒔𝒊𝒏( )×( )
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵
− 𝒄𝒐𝒔( )×( ) − (𝒕𝑨𝒍𝑮𝒂𝑵 × 𝒔𝒊𝒏( ))
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵
× + × (𝝋𝑩 − 𝝋𝑻 ) × 𝒄𝒐𝒔( ) × (𝟐
𝒏𝝅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝝅𝒏𝒘𝒅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐
× 𝒔𝒊𝒏( )𝟐 × + × (𝝋𝑩 − 𝝋𝑻 )
𝟐 × (𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 ) 𝒏𝝅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝝅𝒏𝒘𝒅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐
× 𝒔𝒊𝒏( ) × 𝒔𝒊𝒏( )× )+
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 𝒏𝝅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
× (−𝝋𝑻 − 𝑽𝒑𝒊𝒏𝒄𝒉 ) × 𝒄𝒐𝒔( )× − (𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 )
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅
𝝅𝒏𝒘𝒅
× (𝟒 × 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐 × 𝒔𝒊𝒏( )𝟐 + 𝟒 × 𝒘𝒅 𝟐
𝟐 × (𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 )
𝝅𝒏𝒘𝒅
× 𝒔𝒊𝒏( )𝟐 + 𝟖 × 𝒕𝑨𝒍𝑮𝒂𝑵 × 𝒘𝒅
𝟐 × (𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 )
𝝅𝒏𝒘𝒅
𝒔𝒊𝒏( )𝟐 𝟐
𝟐 × (𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 )
× 𝟑 𝟐 𝟑
)+ × (−𝝋𝑻 − 𝑽𝒑𝒊𝒏𝒄𝒉 )
𝒏 × 𝒘𝒅 × 𝝅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵
× 𝒔𝒊𝒏( ) × (−𝟐 × (𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 )𝟐 × (𝒕𝑨𝒍𝑮𝒂𝑵
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝝅𝒏𝒘𝒅 𝝅𝒏𝒘𝒅
× 𝒔𝒊𝒏( )) + (𝒘𝒅 × 𝒔𝒊𝒏( )) − 𝟐𝝅𝒏𝒘𝒅
𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅
(𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 )𝟐
×
𝒏𝟑 × 𝒘𝒅 𝟐 × 𝝅𝟑

23
 𝟐 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 3.7.3
𝒄= × (𝝋𝑩 + 𝑽𝑫𝑺 − 𝑽𝒑𝒊𝒏𝒄𝒉 ) × 𝒄𝒐𝒔( )× × (−𝟏
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝟐
+ 𝒄𝒐𝒔( )) + ( ) × (𝝋𝑩 + 𝑽𝑫𝑺 − 𝑽𝒑𝒊𝒏𝒄𝒉 )
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝟐
× 𝒔𝒊𝒏( )× × 𝒔𝒊𝒏( )+
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵
× (𝝋𝑩 + 𝑽𝑫𝑺 + 𝑽𝑫𝑺 − 𝑽𝒑𝒊𝒏𝒄𝒉 ) × 𝒄𝒐𝒔( ) × (𝒔𝒊𝒏( )
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
×( ) − (𝒕𝑨𝒍𝑮𝒂𝑵 × 𝒄𝒐𝒔( )×
𝒏𝝅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅
𝟐 𝝋𝑩 + 𝑽𝑫𝑺 − 𝑽𝒑𝒊𝒏𝒄𝒉 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵
+ × × 𝒔𝒊𝒏( )
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐
×( ) − 𝒄𝒐𝒔( )×( ) − (𝒕𝑨𝒍𝑮𝒂𝑵
𝒏𝝅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐
× 𝒔𝒊𝒏( )) × + × (𝝋𝑩 + 𝑽𝑫𝑺 − 𝝋𝑻 )
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝝅𝒏𝒘𝒅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
× 𝒄𝒐𝒔( ) × (𝟐 × 𝒔𝒊𝒏( )𝟐 ×
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐 × (𝒕 + 𝒘𝒅 ) 𝒏𝝅
𝟐 𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵
+ × (𝝋𝑩 + 𝑽𝑫𝑺 − 𝝋𝑻 ) × 𝒔𝒊𝒏( )
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝝅𝒏𝒘𝒅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐
× 𝒔𝒊𝒏( )× )+ × (−𝝋𝑻 − 𝑽𝒑𝒊𝒏𝒄𝒉 )
𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 𝒏𝝅 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
× 𝒄𝒐𝒔( )× − (𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 ) × (𝟒 × 𝒕𝑨𝒍𝑮𝒂𝑵 𝟐
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒏𝝅
𝝅𝒏𝒘𝒅 𝝅𝒏𝒘𝒅
× 𝒔𝒊𝒏( )𝟐 + 𝟒 × 𝒘𝒅 𝟐 × 𝒔𝒊𝒏( )𝟐 + 𝟖
𝟐 × (𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 ) 𝟐 × (𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 )
𝝅𝒏𝒘𝒅
𝒔𝒊𝒏( )𝟐 𝟐
𝟐 × (𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 )
× 𝒕𝑨𝒍𝑮𝒂𝑵 × 𝒘𝒅 × )+ × (−𝝋𝑻
𝒏𝟑 × 𝒘𝒅 𝟐 × 𝝅𝟑 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝒏𝝅𝒕𝑨𝒍𝑮𝒂𝑵
− 𝑽𝒑𝒊𝒏𝒄𝒉 ) × 𝒔𝒊𝒏( ) × (−𝟐 × (𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 )𝟐 × (𝒕𝑨𝒍𝑮𝒂𝑵
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝝅𝒏𝒘𝒅 𝝅𝒏𝒘𝒅
× 𝒔𝒊𝒏( )) + (𝒘𝒅 × 𝒔𝒊𝒏( )) − 𝟐𝝅𝒏𝒘𝒅
𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅
(𝒕𝑨𝒍𝑮𝒂𝑵 + 𝒘𝒅 )𝟐
×
𝒏𝟑 × 𝒘𝒅 𝟐 × 𝝅𝟑

𝑳𝒆𝒇𝒇 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒄𝟏 + 𝒄𝟐 + 𝒄𝟑 3.7.4

𝒚𝒎𝒊𝒏 = − × 𝒍𝒏( )
𝟐 𝟐𝝅 𝒃𝟏 + 𝒃𝟐 + 𝒃𝟑

24
 𝒕𝒆𝒎𝒑 3.7.5
𝝅(𝑳𝒆𝒇𝒇 − 𝒚𝒎𝒊𝒏 ) 𝝅(𝒚 )
𝒃𝟏 × 𝐬𝐢𝐧𝐡 ( 𝒘 + 𝒕 ) + 𝒄𝟏 + 𝐬𝐢𝐧𝐡 (𝒘 + 𝒕𝒎𝒊𝒏 )
𝒅 𝑨𝒍𝑮𝒂𝑵 𝒅 𝑨𝒍𝑮𝒂𝑵 𝝅(𝒕𝑨𝒍𝑮𝒂𝑵 )
= × 𝐬𝐢𝐧𝐡 ( )
𝝅(𝑳𝒆𝒇𝒇 ) 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝐬𝐢𝐧𝐡 (𝒘 + 𝒕 )
𝒅 𝑨𝒍𝑮𝒂𝑵

𝟐𝝅(𝑳𝒆𝒇𝒇 − 𝒚𝒎𝒊𝒏 ) 𝟐𝝅(𝒚 )

𝒃𝟐 × 𝐬𝐢𝐧𝐡 ( 𝒘 + 𝒕 ) + 𝒄𝟐 × 𝐬𝐢𝐧𝐡 (𝒘 + 𝒕𝒎𝒊𝒏 )
𝒅 𝑨𝒍𝑮𝒂𝑵 𝒅 𝑨𝒍𝑮𝒂𝑵 𝟐𝝅(𝒕𝑨𝒍𝑮𝒂𝑵 )
+ × 𝐬𝐢𝐧𝐡 ( )
𝟐𝝅(𝑳𝒆𝒇𝒇 ) 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝐬𝐢𝐧𝐡 (𝒘 + 𝒕 )
𝒅 𝑨𝒍𝑮𝒂𝑵

𝟑𝝅(𝑳𝒆𝒇𝒇 − 𝒚𝒎𝒊𝒏 ) 𝟑𝝅(𝒚𝒎𝒊𝒏 )

𝒃𝟑 × 𝐬𝐢𝐧𝐡 ( ) + 𝒄𝟑 × 𝐬𝐢𝐧𝐡 ( )
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵 𝟑𝝅(𝒕𝑨𝒍𝑮𝒂𝑵 )
+ × 𝐬𝐢𝐧 ( )
𝟑𝝅(𝑳𝒆𝒇𝒇 ) 𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵
𝐬𝐢𝐧𝐡 ( )
𝒘𝒅 + 𝒕𝑨𝒍𝑮𝒂𝑵

−𝟎. 𝟎𝟓 + 𝒕𝒆𝒎𝒑, 𝑳𝒆𝒇𝒇 ≥ 𝟓 × 𝟏𝟎−𝟔 3.7.6

𝒅𝒔𝒎 = {
𝑬𝑻𝑨𝑫 × 𝒕𝒆𝒎𝒑 × (𝟏 + 𝑽𝒅𝒔𝒊𝒏 × 𝑬𝑻𝑨𝑫𝟐 ), 𝒙 < 𝟓 × 𝟏𝟎−𝟔

3.8 Core Function Equations:

𝑵𝑨 𝑬𝒈𝑮𝒂𝑵 3.8.1
𝑽𝑺 = 𝑼𝑻 × 𝒍𝒏( ) +
𝑵𝑽 𝒒

𝑽𝒑𝒊𝒏𝒄𝒉 + 𝑽𝒔𝒉 − 𝑽𝑺 + 𝝋𝑻 − 𝑽𝒄𝒉 3.8.2

𝑽𝒑𝟏 =
𝑼𝑻
𝒒 3.8.3
𝑪𝒃 × 𝒏𝒒 × 𝑫𝑶𝑺𝟐𝑫 × 𝑼𝑻
𝜶=
𝑼𝑻
𝒈 𝟑 3.8.4
𝜷= × √𝒏𝟎 𝟐
𝑼𝑻

𝟓 × 𝒆−𝑽𝒑𝟏 , 𝑽𝒑𝟏 ≤ 𝟎 3.8.5

𝟏
, 𝟎 < 𝑽𝒑𝟏 ≤ 𝟑𝟎
𝑽𝒑𝟏
𝒚𝟎 = 𝜶 + 𝜷 + 𝟎. 𝟓
𝟑
, 𝑶𝒕𝒉𝒆𝒓𝒘𝒊𝒔𝒆
𝑽𝒑𝟏
{𝜶 + 𝜷 + 𝟎. 𝟓
𝟐×𝜷× 𝟑√𝒚𝟎 3.8.6
𝒚𝟎 + 𝜶 + 𝟑
, 𝑽𝒑𝟏 ≤ 𝟎
𝜷× √𝒚𝟎 −𝟐
𝟑×(𝟏+𝑽𝒑𝟏 +𝒍𝒏(𝒚𝟎 )− )
𝟑
𝟑
𝟐×𝜷× √𝒚𝟎
𝒚𝟎 + 𝜶 + , 𝟎 < 𝑽𝒑𝟏 ≤ 𝟑𝟎
𝒚𝟏 = 𝟑
𝜷× √𝒚𝟎 −𝟐
𝟑×(𝟏+𝑽𝒑𝟏 +𝒍𝒏(𝒚𝟎 )− )
𝟑
𝟑
𝟐×𝜷× √𝒚𝟎
𝟏+𝜶+ 𝟑
, 𝑶𝒕𝒉𝒆𝒓𝒘𝒊𝒔𝒆
𝜷× √𝒚𝟎 −𝟐
{ 𝟑×(𝟎.𝟓+𝑽𝒑𝟏 −
𝟑
)

25
 𝟐 × 𝜷 × 𝟑√𝒚𝟏 3.8.7
𝒚𝟏 + 𝜶 + 𝟑
, 𝑽𝒑𝟏 ≤ 𝟎
𝜷 × √𝒚𝟏 −𝟐
𝟑 × (𝟏 + 𝑽𝒑𝟏 + 𝒍𝒏(𝒚𝟏 ) − 𝟑 )
𝟐 × 𝜷 × 𝟑√𝒚𝟏
𝒚𝟐 = 𝒚𝟏 + 𝜶 + 𝟑
𝜷 × √𝒚𝟏 −𝟐
, 𝟎 < 𝑽𝒑𝟏 ≤ 𝟑𝟎
𝟑 × (𝟏 + 𝑽𝒑𝟏 + 𝒍𝒏(𝒚𝟏 ) − 𝟑 )
𝟑
𝟐 × 𝜷 × √𝒚𝟏
𝟏+𝜶+ 𝟑
, 𝑶𝒕𝒉𝒆𝒓𝒘𝒊𝒔𝒆
𝜷 × √𝒚𝟏 −𝟐
{ 𝟑 × (𝟎. 𝟓 + 𝑽𝒑𝟏 − 𝟑 )
𝟎 , 𝑽𝒑𝟏 ≤ 𝟎 3.8.8
𝟎 , 𝟎 < 𝑽𝒑𝟏 ≤ 𝟑𝟎
𝒚𝟑 = 𝟐 × 𝜷 × 𝟑√𝒚𝟐
𝟏+𝜶+ 𝟑
, 𝑶𝒕𝒉𝒆𝒓𝒘𝒊𝒔𝒆
𝜷 × √𝒚𝟐 −𝟐
𝟑 × (𝟎. 𝟓 + 𝑽𝒑𝟏 − 𝟑 )
{
𝟑 3.8.9
𝜷 × √𝒚𝟐 −𝟐
𝟏 + 𝑽𝒑𝟏 + 𝒍𝒏(𝒚𝟐 ) − , 𝑽𝒑𝟏 ≤ 𝟎
𝟐 × 𝜷 × 𝟑√𝒚𝟐
𝟑 × (𝒚𝟐 + 𝜶 + 𝟑 )
𝟑
𝜷 × √𝒚𝟐 −𝟐
𝒙𝟑 = 𝟏 + 𝑽𝒑𝟏 + 𝒍𝒏(𝒚𝟐 ) − 𝟐 × 𝜷 × 𝟑√𝒚𝟐
, 𝟎 < 𝑽𝒑𝟏 ≤ 𝟑𝟎
𝟑 × (𝒚𝟐 + 𝜶 + )
𝟑
𝟑
𝜷 × √𝒚𝟑 −𝟐
𝟏. 𝟏𝟓 + 𝑽𝒑𝟏 − , 𝑶𝒕𝒉𝒆𝒓𝒘𝒊𝒔𝒆
𝟐 × 𝜷 × 𝟑√𝒚𝟑
{ 𝟑 × (𝟏 + 𝜶 + )
𝟑
𝒄𝒐𝒓𝒆 = 𝑸𝒄𝒉 = 𝒒 × 𝒏𝟎 × 𝒙𝟑 3.8.10

3.9 Geometrical Equations:

𝑻𝟐 3.9.1
𝑬𝒈𝑮𝒂𝑵 = 𝟑. 𝟓𝟎𝟕 × 𝒒 − (𝟎. 𝟗𝟎𝟗 × 𝟏𝟎−𝟑 ) × ×𝒒
𝑻 + 𝟖𝟑𝟎

−𝟑
𝑻𝟐 3.9.2
𝑬𝒈𝑮𝒂𝑵 = 𝟑. 𝟓𝟎𝟕 × 𝒒 − (𝟎. 𝟗𝟎𝟗 × 𝟏𝟎 )× ×𝒒
𝑻 + 𝟖𝟑𝟎

𝑬𝒈𝑨𝒍𝑮𝒂𝑵 = 𝑬𝒈𝑨𝒍𝑵 × (𝒙𝒙) + 𝑬𝒈𝑮𝒂𝑵 × (𝟏 − 𝒙𝒙) × 𝟏. 𝟑(𝒙𝒙) × (𝟏 − 𝒙𝒙) × 𝒒 3.9.3

∆𝑬𝑪 = 𝟎. 𝟕 × (𝑬𝒈𝑨𝒍𝑮𝒂𝑵 − 𝑬𝒈𝑮𝒂𝑵 ) 3.9.4

3.10 Velocity Saturation Equations:

𝑼𝑪𝑹𝑰𝑻𝑻 × 𝑳 3.10.1
𝑼𝑻
𝑽𝑪 = 𝒒
𝟏 + 𝑻𝑯𝑬𝑻𝑨𝟑 × ( 𝒔𝒔 )
𝟐 + 𝒒𝒃

26
 𝟐 3.10.2
𝒆𝒄𝒍𝒎 =
𝑽𝑪

𝟐 × 𝒆𝒄𝒍𝒎 × 𝒊𝒇𝒇 3.10.3

𝒒𝒔𝒂𝒕 =
𝒆𝒄𝒍𝒎 + 𝟐 + 𝟐𝒆𝒄𝒍𝒎 × 𝒒𝒔𝒔 + √(𝒆𝒄𝒍𝒎 + 𝟐 )𝟐 + 𝟖𝒆𝒄𝒍𝒎 × 𝒒𝒔𝒔

3.10.4

𝑽𝑫𝑺𝑺𝑨𝑻 = 𝑴𝑨𝑿𝒔 𝒗𝒑 − (𝟐𝒒𝒔𝒂𝒕 + 𝒍𝒏(𝒒𝒔𝒂𝒕 ))

(
𝟏 + 𝒆𝒄𝒍𝒎 ∗ (𝒒𝒔𝒔 − 𝒒𝒔𝒂𝒕 )
×
𝟐
𝟏 + 𝟐 (𝒆𝒄𝒍𝒎 × ((𝟐 − 𝑫𝑬𝑳𝑻𝑨𝑨) × (𝒒𝒔𝒔 − 𝒒𝒔𝒂𝒕 )) )
√
(𝟎. 𝟏 + 𝒆𝒄𝒍𝒎 × (𝟐 − 𝑫𝑬𝑳𝑻𝑨𝑨) × (𝒒𝒔𝒔 − 𝒒𝒔𝒂𝒕 ))

𝑽𝑺
+ 𝒆𝒄𝒍𝒎 (𝒒𝒔𝒔 − 𝒒𝒔𝒂𝒕 ) − , 𝟑, 𝟒
𝑼𝑻

)
𝑨𝑪𝑳𝑴 𝟒 × 𝒒𝒔𝒂𝒕 + 𝑫𝑬𝑳𝑻𝑨𝑨 3.10.5
𝒅𝒗 = ×( )
𝑫𝑬𝑳𝑻𝑨𝑨 𝒒𝒔𝒔 + 𝟏

3.10.6
𝑽𝑫 𝑽𝑺 𝟒 × 𝒅𝒗
𝑽𝒅′ = 𝟎. 𝟓 × √( − × √𝟏 + + 𝑽𝑫𝑺𝑺𝑨𝑻) 𝟐 + 𝟒 × 𝒅𝒗 × 𝑽𝑫𝑺𝑺𝑨𝑻 − 𝟎. 𝟓
𝑼𝑻 𝑼𝑻 𝑽𝑫𝑺𝑺𝑨𝑻

𝑽𝑫 𝑽𝑺 𝟒 × 𝒅𝒗 𝑽𝒔
× √( − × √𝟏 + − 𝑽𝑫𝑺𝑺𝑨𝑻 )𝟐 + 𝟒 × 𝒅𝒗 × 𝑽𝑫𝑺𝑺𝑨𝑻 +
𝑼𝑻 𝑼𝑻 𝑽𝑫𝑺𝑺𝑨𝑻 𝑼𝑻
𝐔𝐓 3.10.7
𝐔𝐓_𝐕𝐜 =
𝐕𝐃𝟏

3.10.8
𝑼𝑻_𝑽𝑪
𝑽𝒅𝒔𝒔 = 𝑽𝑪 × √𝟎. 𝟐𝟓 + √𝒊𝒇𝒇 × − 𝟎. 𝟓
𝑼𝑻
𝑽𝑫𝟏 = 𝑼𝑪𝑹𝑰𝑻 × 𝑳 3.10.9

3.10.10
∆𝑽 = (𝟏𝟔 × 𝑼𝑻 𝟐 ) × 𝝀 × (√𝒊𝒇𝒇 − 𝑽𝒅𝒔𝒔 ) + 𝟏𝟓. 𝟔 × 𝟏𝟎−𝟑

3.10.11
𝑽𝒊𝒑 = √𝑽𝒅𝒔𝒔 𝟐 + ∆𝑽
− √(𝟎. 𝟓 × (𝑽𝑫 − 𝑽𝒔) − 𝑽𝒅𝒔𝒔 ) × ((𝟎. 𝟓 × (𝑽𝑫 − 𝑽𝒔) − 𝑽𝒅𝒔𝒔 + ∆𝑽)
𝑸𝑫𝑫 = 𝑪𝑶𝑹𝑬(𝑽𝑫𝑷𝑹𝑰𝑴𝑬 3.10.12
× 𝑼𝑻 , 𝑫𝑰𝑩𝑳𝒙, 𝑽𝑷𝒊𝒏𝒄𝒉 , 𝒏𝒒 , 𝑪𝒃 , 𝑫𝑶𝑺𝟐𝑫 , 𝒈𝒈, 𝒒, 𝑼𝑻 , 𝑵𝑽 , 𝑬𝒈𝑮𝒂𝑵 , 𝑵𝑨 , 𝝋𝑻 )

27
 𝑸𝑫𝑫 3.10.13
𝒒𝒅𝒅 =
−𝑸𝒔𝒑

𝑳𝑪_𝑼𝑪𝑹𝑰𝑻 = 𝑳𝑪 × 𝑼𝑪𝑹𝑰𝑻 3.10.14

𝑽𝒊𝒑 3.10.15
∆𝑳 = 𝑳𝒄𝝀 × 𝒍𝒏(𝟏 + 𝑽𝑫𝑺 − )
𝑳𝑪_𝑼𝑪𝑹𝑰𝑻

𝑳𝒆𝒒 = 𝑳 − ∆𝑳 3.10.16

3.11 Vertical Field Effect Equations:

−𝑸𝒔𝒑 3.11.1
𝒕𝒉𝒆𝒕𝒂𝒙 =
𝑬𝟎 × 𝜺𝑮𝒂𝑵

𝑼𝑻 3.11.2
𝟐×
𝑼𝑪𝑹𝑰𝑻𝑻 × 𝑳
𝜼𝒙 =
𝝀
𝟏 + 𝟐 × 𝑳𝒘
𝟏 3.11.3
𝑲𝟎 =
𝑽𝒑𝒊𝒏𝒄𝒉
𝟏 + 𝟐√ 𝜸

𝒌𝟏 = 𝒕𝒉𝒆𝒕𝒂𝒙 × (𝟎. 𝟓 − 𝒌𝟎 ) 3.11.4

𝑽𝑷𝒊𝒏𝒄𝒉 3.11.5
𝑼𝑻
𝒌𝟐 = 𝟏 + 𝑻𝒉𝒆𝒕𝒂𝒙 ×
𝑽
𝟏 + 𝟐√ 𝑷𝒊𝒏𝒄𝒉
𝜸
𝒌𝟑 = 𝒌𝟏 × 𝒒𝒅𝒅 + 𝒌𝟐 3.11.6

𝒌𝟒 = 𝒌𝟏 × 𝒒𝒔𝒔 + 𝒌𝟐 3.11.7

𝒌𝟑 3.11.8
𝒌𝟓 =
𝒌𝟒

𝒌𝟐 3.11.9
𝒌𝟔 = (𝟐 × − 𝟏)
𝒌𝟏

𝒍 𝟐(𝒒𝒔𝒔 − 𝒒𝒅𝒅 ) + 𝒌𝟔 × 𝒍𝒏(𝒌𝟓 ) 3.11.10

𝑰𝒅𝒔 = 𝑰𝒔𝒑 × 𝜷𝒄𝒍𝒎 × ×
𝒍𝒆𝒒 𝒌𝟏

28
3.12 Access Resistance Equations:
𝑻 3.12.1
𝒏𝒔𝒂𝒄𝒄 = 𝒏𝒔𝟎 × (𝟏 − 𝒌𝒕𝒉𝟏 ( − 𝟏))
𝑻𝑵𝑶𝑴

𝒗𝒔𝒂𝒕𝒂𝒄𝒄 = 𝒗𝒔𝒂𝒕𝟎𝒂𝒄𝒄 ∗ (𝟏 + 𝒌𝒕𝒉𝟐 × (𝑻 − 𝑻𝑵𝑶𝑴 )) 3.12.2

𝑻 𝒌𝒕𝒉𝟑 3.12.3
𝝁𝒂𝒄𝒄 = 𝝁𝟎𝒂𝒄𝒄 × ( )
𝑻𝑵𝑶𝑴

𝒒𝒂𝒄𝒄 = 𝒒 × 𝒏𝒔𝒂𝒄𝒄 × 𝑾𝒆𝒇𝒇𝒄 3.12.4

𝑳𝒂𝒓𝒅 3.12.5
𝑹𝒅𝟎 =
𝒒𝒂𝒄𝒄 × 𝝁𝒂𝒄𝒄

𝑳𝒂𝒓𝒔 3.12.6
𝑹𝒔𝟎 =
𝒒𝒂𝒄𝒄 × 𝝁𝒂𝒄𝒄

𝑰𝒂𝒄𝒄𝒔𝒂𝒕 = 𝒒𝒂𝒄𝒄 × 𝒗𝒔𝒂𝒕𝒂𝒄𝒄 3.12.7

𝟏 3.12.8
𝒈𝒔𝒂𝒓 𝒈
𝑰𝒅𝒔 𝒔𝒂𝒓
𝑹𝒅 = 𝑹𝒅𝟎 / (𝟏 − ( ) )
𝑰𝒂𝒄𝒄𝒔𝒂𝒕
𝟏 3.12.9
𝒈𝒔𝒂𝒓 𝒈
𝑰𝒅𝒔 𝒔𝒂𝒓
𝑹𝒔 = 𝑹𝒔𝟎 / (𝟏 − ( ) )
𝑰𝒂𝒄𝒄𝒔𝒂𝒕
𝑹𝒄𝒔 3.12.10
𝑹𝒄𝒐𝒔 =
𝑾𝒆𝒇𝒇𝒄

𝑹𝒄𝒅 3.12.11
𝑹𝒄𝒐𝒅 =
𝑾𝒆𝒇𝒇𝒄

𝟏 3.12.13
𝒈𝒔𝒂𝒓 𝒈
𝑻 𝑰𝒅𝒔 𝒔𝒂𝒓
𝑹𝒅 = 𝑹𝒄𝒐𝒅 / (𝟏 + 𝒌𝒕𝒓𝒅 ( ) + 𝑹𝒅𝟎 /(𝟏 − ( ) )
𝑻𝑵𝑶𝑴 𝑰𝒂𝒄𝒄𝒔𝒂𝒕
𝟏 3.12.14
𝒈𝒔𝒂𝒓 𝒈
𝑻 𝑰𝒅𝒔 𝒔𝒂𝒓
𝑹𝒔 = 𝑹𝒄𝒐𝒔 / (𝟏 + 𝒌𝒕𝒓𝒔 ( ) + 𝑹𝒔𝟎 /(𝟏 − ( ) )
𝑻𝑵𝑶𝑴 𝑰𝒂𝒄𝒄𝒔𝒂𝒕

3.13 Transconductance Equations:

−𝒒 𝟏 𝒒 3.13.1
𝑫𝒊𝒇𝒒𝒅_𝒗𝒈 = 𝒏 ×( 𝒒 + 𝒈 )−𝟏
𝒒 ×𝑸𝒔𝒑 𝑸𝒔𝒑 × 𝒅𝒅 𝑪𝒃 ×𝒏𝒒 +𝟐×
𝒒 𝟑 𝒒
𝟑× √−𝑸𝒔𝒑 × 𝒅𝒅
𝑫𝑶𝑺𝟐𝑫 ×(𝟏−𝒆𝑫𝑶𝑺𝟐𝑫 ×𝑼𝑻 ) 𝒒

29
 −𝒒 3.13.2
𝑫𝒊𝒇𝒒𝒅𝒔𝒂𝒕_𝒗𝒈 =
𝒏𝒒 × 𝑸𝒔𝒑
𝟏
×( 𝒒
𝑸𝒔𝒑 × 𝒅𝒔𝒂𝒕
𝒒
𝑫𝑶𝑺𝟐𝑫 × (𝟏 − 𝒆𝑫𝑶𝑺𝟐𝑫 ×𝑼𝑻 )
𝒒
+ 𝒈 )−𝟏
𝑪𝒃 × 𝒏 𝒒 + 𝟐 ×
𝟑 𝒒𝒅𝒔𝒂𝒕
𝟑 × √−𝑸𝒔𝒑 × 𝒒
−𝒒 3.13.3
𝑫𝒊𝒇𝒒𝒔_𝒗𝒈 =
𝒏𝒒 × 𝑸𝒔𝒑
𝟏 𝒒
×( 𝒒 + 𝒈 )−𝟏
𝑸𝒔𝒑 × 𝒔𝒔 𝑪𝒃 × 𝒏 𝒒 + 𝟐 ×
𝒒
𝒆𝑫𝑶𝑺𝟐𝑫 ×𝑼𝑻 ) 𝟑 𝒒
𝑫𝑶𝑺𝟐𝑫 × (𝟏 − 𝟑 × √−𝑸𝒔𝒑 × 𝒔𝒔
𝒒
−𝒒 3.13.4
𝑫𝒊𝒇𝒒𝒅_𝒗𝒈 =
𝒏𝒒 × 𝑸𝒔𝒑
𝟏
×( 𝒒
𝑸𝒔𝒑 × 𝒅𝒅
𝒒
𝑫𝑶𝑺𝟐𝑫 × (𝟏 − 𝒆𝑫𝑶𝑺𝟐𝑫 ×𝑼𝑻 )
𝒒
+ 𝒈 )−𝟏
𝑪𝒃 × 𝒏 𝒒 + 𝟐 ×
𝟑 𝒒
𝟑 × √−𝑸𝒔𝒑 × 𝒒𝒅𝒅
𝑫𝒊𝒇𝒍𝒏 3.13.5
𝑲𝟐
= 𝟐 × ( − 𝟏)
𝑲𝟏
(𝑲𝟏 × 𝑫𝒊𝒇𝒒𝒅_𝒗𝒈 × (𝑲𝟏 × 𝒒𝒔𝒔 + 𝑲𝟐 ) − 𝑲𝟏 × 𝑫𝒊𝒇𝒒𝒔_𝒗𝒈 × (𝑲𝟏 × 𝒒𝒅𝒅 + 𝑲𝟐 ))
×
(𝑲𝟏 × 𝒒𝒅𝒅 + 𝑲𝟐 ) × (𝑲𝟏 × 𝒒𝒔𝒔 + 𝑲𝟐 )
𝑳 3.13.6
𝒈𝒎 = 𝑰𝒔𝒑 × 𝜷𝒄𝒍𝒎 × 𝑳 × (𝟐 × (𝑫𝒊𝒇𝒒𝒔𝒗𝒈 − 𝑫𝒊𝒇𝒒𝒅𝒗𝒈 ) + 𝑫𝒊𝒇𝒍𝒏 ) × (𝑲𝟏 ×
𝒆𝒒
𝑲 𝑲 ×𝒒 +𝑲
(𝟏 + 𝜼 × (𝒒𝒔𝒔 − 𝒒𝒅𝒔𝒂𝒕 )) − 𝟐(𝒒𝒔𝒔 − 𝒒𝒅𝒅 ) + 𝟐(𝑲𝟐 − 𝟏) × 𝒍𝒏( 𝑲𝟏 ×𝒒𝒅𝒅+𝑲 𝟐)) × 𝑲𝟏 ×
𝟏 𝟏 𝒔𝒔 𝟐
𝜼×(𝑫𝒊𝒇𝒒𝒔_𝒗𝒈 −𝑫𝒊𝒇𝒒𝒅_𝒗𝒈 )
( )
𝑲𝟏 ×(𝟏+𝜼×(𝒒𝒔𝒔 −𝒒𝒅𝒔𝒂𝒕 ))𝟐
𝟏 𝒒 3.13.7
𝑭 =( 𝒒 + 𝒈𝒈 )−𝟏
𝑸𝒔𝒑 × 𝒅𝒅 𝑪𝒃 × 𝒏 𝒒 + 𝟐 ×
𝒒
𝒆𝑫𝑶𝑺𝟐𝑫 ×𝑼𝑻 ) 𝟑 𝒒
𝑫𝑶𝑺𝟐𝑫 × (𝟏 − 𝟑 × √−𝑸𝒔𝒑 × 𝒒𝒅𝒅
𝟏 𝒒 3.13.8
𝑭𝟏 = ( 𝒒𝒔𝒂𝒕 + 𝒈𝒈 )−𝟏
𝑸𝒔𝒑 × 𝑪𝒃 × 𝒏 𝒒 + 𝟐 ×
𝒒
𝑫𝑶𝑺𝟐𝑫 × (𝟏 − 𝒆𝑫𝑶𝑺𝟐𝑫 ×𝑼𝑻 ) 𝟑 𝒒𝒔𝒂𝒕
𝟑 × √−𝑸𝒔𝒑 × 𝒒
𝟐×𝑲 3.13.9
( 𝑲 𝟐 − 𝟏) × 𝑲𝟏
𝟏
𝑭𝟐 =
𝑲𝟏 × 𝒒𝒅𝒅 + 𝑲𝟐
𝟐 × 𝑲𝟐 3.13.10
( − 𝟏) × 𝑲𝟏
𝑲𝟏
𝑭𝟑 =
𝑲𝟏 × 𝒒𝒔𝒔 + 𝑲𝟐

30
 𝒒 𝑳 3.13.11
𝒈𝒅𝒔𝟏 = × 𝑰𝒔𝒑 × 𝜷𝒄𝒍𝒎 × × (𝑭 × (−𝟐 + 𝑭𝟐)) × (𝑲𝟏 × (𝟏 + 𝜼 × (𝒒𝒔𝒔 − 𝒒𝒔𝒂𝒕 ))
𝑸𝒔𝒑 𝑳𝒆𝒒
𝑲𝟐 𝑲𝟏 × 𝒒𝒅𝒅 + 𝑲𝟐
+ (𝟐(𝒒𝒔𝒔 − 𝒒𝒅𝒅 ) + 𝟐( − 𝟏) × 𝒍𝒏( )) × 𝑭𝟏
𝑲𝟏 𝑲𝟏 × 𝒒𝒔𝒔 + 𝑲𝟐
𝑲𝟏 × 𝜼
×( )
(𝑲𝟏 × (𝟏 + 𝜼 × (𝒒𝒔𝒔 − 𝒒𝒔𝒂𝒕 ))𝟐
𝟏 𝒒 3.13.12
𝑭𝒂 = ( 𝒒𝒔𝒔 + 𝒈𝒈 )−𝟏
𝑸𝒔𝒑 × 𝑪 × 𝒏 + 𝟐 ×
𝒒 𝒃 𝒒
𝑫𝑶𝑺𝟐𝑫 × (𝟏 − 𝒆𝑫𝑶𝑺𝟐𝑫 ×𝑼𝑻 ) 𝟑 𝒒
𝟑 × √−𝑸𝒔𝒑 × 𝒒𝒔𝒔
𝒒 𝑳 3.13.13
𝒈𝒎𝒔𝟏 = × 𝑰𝒔𝒑 × 𝜷𝒄𝒍𝒎 × × (𝑭𝒂 × (−𝟐 + 𝑭𝟑)) × (𝑲𝟏 × (𝟏 + 𝜼 × (𝒒𝒔𝒔
𝑸𝒔𝒑 𝑳𝒆𝒒
𝑲𝟐 𝑲𝟏 × 𝒒𝒅𝒅 + 𝑲𝟐
− 𝒒𝒔𝒂𝒕 )) + (𝟐(𝒒𝒔𝒔 − 𝒒𝒅𝒅 ) + 𝟐( − 𝟏) × 𝒍𝒏( )) × 𝑭𝟏
𝑲𝟏 𝑲𝟏 × 𝒒𝒔𝒔 + 𝑲𝟐
𝑲𝟏 × 𝜼
×( )
(𝑲𝟏 × (𝟏 + 𝜼 × (𝒒𝒔𝒔 − 𝒒𝒔𝒂𝒕 ))𝟐

3.14 Self-Heating Equations:

𝑷𝒅𝒊𝒔𝒔 = −(𝑰𝒅𝒔 × (𝑽𝒅𝒂𝒓 − 𝑽𝒔𝒂𝒓 ) + 𝑰𝒅𝒔 × 𝑹𝟐𝒅 + 𝑰𝒅𝒔 × 𝑹𝟐𝒔 ) 3.14.1

3.15 Transcapacitance:
𝑸𝒔𝒑 𝟐 3.15.1
𝜼 = −𝑾𝒆𝒇𝒇𝒄 × 𝑳𝒆𝒇𝒇𝒄 × ( )
𝒊𝒇𝒊𝒓

̅ 𝒎𝒅 = 𝜼 × (𝒊𝒇𝒇 × (𝟎. 𝟔𝟔𝟕 ∗ 𝒒𝟑𝒅𝒅 + 𝟎. 𝟓 × 𝒒𝟐𝒅𝒅 ) − (𝟎. 𝟒 × 𝒒𝟓𝒅𝒅 + 𝟎. 𝟕𝟓 × 𝒒𝟒𝒅𝒅 + 𝟎. 𝟑𝟑

𝑸 3.15.2
× 𝒒𝟑𝒅𝒅 ) − 𝒊𝒇𝒇 × (𝟎. 𝟔𝟔𝟕 × 𝒒𝟑𝒔𝒔 + 𝟎. 𝟓 × 𝒒𝟐𝒔𝒔 ) + (𝟎. 𝟒 × 𝒒𝟓𝒔𝒔 + 𝟎. 𝟕𝟓
× 𝒒𝟒𝒔𝒔 + 𝟎. 𝟑𝟑 × 𝒒𝟑𝒔𝒔 ))
̅ 𝒎𝒔 = 𝜼 ∗ (−𝒊𝒓 × (𝟎. 𝟔𝟔𝟕 × 𝒒𝟑𝒅𝒅 + 𝟎. 𝟓 × 𝒒𝟐𝒅𝒅 ) + (𝟎. 𝟒 × 𝒒𝟓𝒅𝒅 + 𝟎. 𝟕𝟓 × 𝒒𝟒𝒅𝒅 + 𝟎. 𝟑𝟑
𝑸 3.15.3
× 𝒒𝟑𝒅𝒅 ) + 𝒊𝒓 × (𝟎. 𝟔𝟔𝟕 × 𝒒𝟑𝒔𝒔 + 𝟎. 𝟓 × 𝒒𝟐𝒔𝒔 ) − (𝟎. 𝟒 × 𝒒𝟓𝒔𝒔 + 𝟎. 𝟕𝟓 × 𝒒𝟒𝒔𝒔
+ 𝟎. 𝟑𝟑 × 𝒒𝟑𝒔𝒔 ))
𝒒
(𝑸𝒔𝒑 × 𝒅𝒅 ) 3.15.4
𝒂𝟏 = 𝟏 − 𝒆 𝒒×𝒏𝟎

𝟏 3.15.5
𝒂𝟐 =
𝑫𝑶𝑺𝟐𝑫 × 𝒂𝟏

𝒈𝒈 3.15.6
𝒂𝟑 = 𝟐 ×
𝟑× 𝒏𝟎.𝟑𝟑𝟒
𝒄𝒉𝑫𝑫

𝒒 3.15.7
−
𝒏𝒒 × 𝑸𝒔𝒑
𝒅𝒊𝒇𝒒𝒅𝑽𝑮 = 𝒒
𝒂𝟐 + 𝑪 ∗ 𝒏 + 𝒂𝟑
𝒃 𝒒
𝒒𝒔𝒔
𝑸𝒔𝒑 ×
𝒒×𝑫𝑶𝑺𝟐𝑫 ×𝑼𝑻
3.15.8
𝒃𝟏 = 𝟏 − 𝒆

31
 𝟏 3.15.9
𝒃𝟐 =
𝑫𝑶𝑺𝟐𝑫 × 𝒃𝟏

𝒈𝒈 3.15.10
𝒃𝟑 = 𝟐 ×
𝟑× 𝒏𝟎.𝟑𝟑𝟒
𝒄𝒉𝑺𝑺

𝒒 3.15.11
(− 𝒏 × 𝑸 )
𝒒 𝒔𝒑
𝒅𝒊𝒇𝒒𝒔𝑽𝑮 = 𝒒
𝒃𝟐 + 𝑪 × 𝒏 + 𝒃𝟑
𝒃 𝒒
𝒇𝒅 = (𝒊𝒇𝒇 − 𝒒𝟐𝒅𝒅 − 𝒒𝒅𝒅 ) × (𝟐 × 𝒒𝟐𝒅𝒅 + 𝒒𝒅𝒅 ) 3.15.12

𝒈𝒔 = (−𝒊𝒓 + 𝒒𝟐𝒔𝒔 + 𝒒𝒔𝒔 ) × (𝟐 × 𝒒𝟐𝒔𝒔 + 𝒒𝒔𝒔 ) 3.15.13

𝑪𝑫𝑮 = 𝜼 × (𝒅𝒊𝒇𝒒𝒅 × 𝒇𝒅 − 𝒅𝒊𝒇𝒒𝒔 × 𝒇𝒔 ) + 𝜼 × (𝟐 × 𝒒𝒔𝒔 + 𝟏) × 𝒅𝒊𝒇𝒒𝒔 3.15.15

𝑽𝑮 𝑽𝑮 𝑽𝑮
× ((𝟎. 𝟔𝟔𝟕 × (𝒒𝟑𝒅𝒅 ) + 𝟎. 𝟓 × (𝒒𝟐𝒅𝒅 )) − (𝟎. 𝟔𝟔𝟕 × (𝒒𝟑𝒔𝒔 ) + 𝟎. 𝟓
𝑸𝑫
× (𝒒𝟐𝒔𝒔 ))) − (𝟐 × ) ∗ (𝒅𝒊𝒇𝒒𝒔 × (𝟐 × 𝒒𝒔𝒔 + 𝟏) − 𝒅𝒊𝒇𝒒𝒅𝑽𝑮 × (𝟐
𝒊𝒇𝒊𝒓 𝑽𝑮
× 𝒒𝒅𝒅 + 𝟏))
𝑪𝑺𝑮 = 𝜼 × (𝒅𝒊𝒇𝒒𝒅𝑽𝑮 × 𝒈𝒅 − 𝒅𝒊𝒇𝒒𝒔𝑽𝑮 × 𝒈𝒔 ) − 𝜼 × (𝟐 × 𝒒𝒅𝒅 + 𝟏) × 𝒅𝒊𝒇𝒒𝒅𝑽𝑮 3.15.16
× ((𝟎. 𝟔𝟔𝟕 × (𝒒𝒅𝒅 × 𝟑) + 𝟎. 𝟓 × (𝒒𝟐𝒅𝒅 )) − (𝟎. 𝟔𝟔𝟕 × (𝒒𝟑𝒔𝒔 ) + 𝟎. 𝟓
𝑸𝑺
× (𝒒𝟐𝒔𝒔 ))) − (𝟐 × ) ∗ (𝒅𝒊𝒇𝒒𝒔𝑽𝑮 × (𝟐 × 𝒒𝒔𝒔 + 𝟏) − 𝒅𝒊𝒇𝒒𝒅𝑽𝑮 × (𝟐
𝒊𝒇𝒊𝒓
× 𝒒𝒅𝒅 + 𝟏))
𝑪𝑮𝑮 = −(𝑪𝑫𝑮 + 𝑪𝑺𝑮 ) 3.15.17

3.16 Substrate Current:

𝑽𝒊𝒃 = 𝑽𝑫 − 𝑽𝑺 − 𝟐 × 𝑰𝑩𝑵 × 𝑽𝑫𝑺𝑺 3.16.1

𝑳𝒄𝑰𝑩𝑩 = 𝑳𝑪 × 𝑰𝑩𝑩𝑻 3.16.2

𝑳𝒄𝑰𝑩𝑩𝑽𝒊𝒃 = −𝑳𝒄𝑰𝑩𝑩 × 𝒊𝒏𝒗𝑽𝒊𝒃 3.16.3

𝑰𝑩𝑨 3.16.4
𝑰𝑩𝑨𝑰𝑩𝑩 =
𝑰𝑩𝑩𝑻

𝑰𝒔𝒖𝒃 = 𝑰𝑩𝑨𝑰𝑩𝑩 × 𝑽𝒊𝒃 × 𝒆𝑳𝒄𝑰𝑩𝑩𝑽𝒊𝒃 × 𝑰𝒅 3.16.5

32
 Appendix:
Validation:
Figure A.1 shows the verilog Ams data of EPFL HEMT Model from ADS along
with experimental MVSG data.

I D - V DS G ds
600 0.25

500
0.2

400
0.15

300

0.1

200

0.05
100

0
0
MIT: Line
HEMT: *

-100 -0.05
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10

Figure A.1 EPFL HEMT Model: * and MIT: -- Figure A.2 Gds of EPFL HEMT Model from ADS

I D - V Gs Gm
900 0.2

800 0.18

0.16
700

0.14
600

0.12
500
0.1

400
0.08

300
0.06

200
0.04

100 0.02

0 0
-5 -4 -3 -2 -1 0 1 2 3 4 -5 -4 -3 -2 -1 0 1 2 3 4

Figure A.3 EPFL HEMT Model: -- and MIT: * Figure A.4 Gm of EPFL HEMT Model from ADS

33
 Effect of parameters on 𝑉𝑑𝑠 :

0.9
0.9
E0 : 2.5 10 6
UCRIT: 1 10 6
0.8 7
E0 : 2.5 10 0.8 UCRIT: 4 10 6

E0 : 2.5 10 8 UCRIT: 8 10 6

0.7 0.7

0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0
-5 -4 -3 -2 -1 0 1 2 3 -5 -4 -3 -2 -1 0 1 2 3

0.9
0.9
lambda: 3
lambda: 3.5
0.8 rcs: 4.1 10 - 2
lambda: 4
0.8 rcs: 4.1 10 - 3
rcs: 4.1 10 - 4
0.7
0.7

0.6
0.6

0.5
0.5

0.4
0.4

0.3
0.3

0.2
0.2

0.1
0.1

0
-5 -4 -3 -2 -1 0 1 2 3 0
-5 -4 -3 -2 -1 0 1 2 3

0.9 0.9

rcd: 4.1 10 - 2 rth: 35

0.8 rcd: 4.1 10 - 3 0.8 rth: 45
rcd: 4.1 10 - 4 rth: 55

0.7 0.7

0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0
-5 -4 -3 -2 -1 0 1 2 3 -5 -4 -3 -2 -1 0 1 2 3

34
 Effect of parameters on 𝑉𝐺𝑠 :

0.9
0.9

0.8
0.8

0.7
0.7

0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2
E0 : 2.5 10 6 UCRIT: 1 10 6
E0 : 2.5 10 7 E0 : 4 10 6
0.1 0.1
E0 : 2.5 10 8 E0 : 8 10 6

0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10

0.9 0.9

rcd: 4.1 10 - 2
0.8 0.8 rcd: 4.1 10 - 3
rcd: 4.1 10 - 4

0.7 0.7

0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

lambda: 3
lambda: 3.5
0.1 0.1
lambda: 4

0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10

0.9 0.9

rcs: 4.1 10 - 2
0.8 rcs: 4.1 10 - 3 0.8
rcs: 4.1 10 - 4

0.7 0.7

0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

rth: 35
0.1 0.1 rth: 45
rth: 55

0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10

35
References:
F. Jazaeri and J. Sallese, "Charge-Based EPFL HEMT Model," in IEEE Transactions on
Electron Devices, vol. 66, no. 3, pp. 1218-1229, March 2019.

F. Jazaeri, M. Shalchian and J. Sallese, "Transcapacitances in EPFL HEMT Model," in

IEEE Transactions on Electron Devices, vol. 67, no. 2, pp. 758-762, Feb. 2020.

F. Jazaeri, et el, "Free Carrier Mobility Extraction in FETs," in IEEE Transactions on

Electron Devices, vol. 64, no. 12, pp. 5279-5283, Dec. 2017.

36