SlideShare une entreprise Scribd logo
1  sur  15
Nanostructures
Research
Group

CENTER FOR SOLID STATE ELECTRONICS RESEARCH
Full-Band

                           Full-Wave

             Simulator
                            Simulator


 6
 4
 2
0
-2
-4
-6
               Γ             X U,K
               L
       L                                      Γ


     Nanostructures
Research
Group

     CENTER FOR SOLID STATE ELECTRONICS RESEARCH
When
devices
are
operated
at
high
frequencies:


       •  Coupling
between
fluctuation
in
charge
distribution
and
       
propagating
EM
fields
must
be
included
into
simulation
model.


       • 

As
operating
frequencies
increase,
period
of
EM
waves
       
approaches
relaxation
time
of
carriers
in
semiconductor
material.


       • 

Finite
amount
of
time
for
carrier
to
react
to
changes
in
applied
       
fields
(i.e.
changes
in
particle
velocities)



                                  Transport
directly
affected
                                  
by
EM
wave
propagation



    Nanostructures
Research
Group

    CENTER FOR SOLID STATE ELECTRONICS RESEARCH
Poisson
solvers
are
unable
to:

               •  directly
capture
inherent
“carrier-wave”
interaction.

               • 
account
for
existing
magnetic
fields
in
real
device.



Full-wave
solver
can:


             •  directly
solve
full
set
of
EM
field
equations.

             • 
account
for
externally
applied
sources
and
changes
in
the
             
field
due
to
charge
fluctuations.

             • 
directly
simulate
absorption/emission
of
EM
energy
in/out
of

             
system
(i.e.
optical
excitation,
radiative
processes,
THz
             
devices.)



  Nanostructures
Research
Group

  CENTER FOR SOLID STATE ELECTRONICS RESEARCH
M. Saraniti and S.M. Goodnick, IEEE TED, 47, 1909 (2000)
                                                    K. Kometer, G. Zandler, and P. Vogl, Phys. Rev., B46(3), 1382 (1992)


particle
dynamics


                                                                         choose
scattering


 Ensemble
Monte
Carlo
(EMC)

                                                                            
new
energy

      
computationally
slow

      
low
memory
requirements

                                                                          
find
new
k
with

                                                                        
dispersion
relation

                         VS.


Cellular
Monte
Carlo
(CMC)

       computationally
fast
                                              choose
new
k

      
high
memory
requirements



      Nanostructures
Research
Group

      CENTER FOR SOLID STATE ELECTRONICS RESEARCH
Idea:


    use
MC
scattering
in
regions
of
band
structure
where
scattering
is
low.


      Nearly
as
fast
as
CMC.

      Reduces
memory
usage.





                                                                        H ybrid/ MC perf ormance ratio




                                                                                                                       time per iter. [sec/ 5000 e ]
                                                                                                                       -
                6
                4
energ y [eV]




                2
               0
               -2
               -4                                                 EMC
               -6                                                 CMC
                                       X U,K
                          L                            L
                    L
                                                                                                         field [V/m]
                           wave vector

                        Nanostructures
Research
Group

                    CENTER FOR SOLID STATE ELECTRONICS RESEARCH
z                                            K.S. Yee, IEEE Trans. Antennas Propagat., 14(302) 1966
                                                              “Yee cell”
Maxwell’s
equations
                                                                                      • 

Most
direct
explicit
solution
of
                                                                             Ey
                                                                                                          
Maxwell’s
equations
available
(i.e.
                                             Ex                                  Ex
                                                                                                          
no
matrix
inversion
required).

    
                                                                   Hz

           ∂H                                                                                    Ez

∇ × E = −µ
                                                         Ex
                                                              Ey
                                                                                                          • 

A
complete
“full-wave”
method
            ∂t                                                          Hx                                
without
approximation
(i.e.
no
pre
                                                   Hy                                      Hy
                                                                                                          -selection
of
output
modes
or
         ∂E 
                                         Ez                                            Ex
                                                                                                          
solution
form
necessary.)

∇× H = ε     +J                                Ex
                                                         Hx
                                                                             Ey        Ex
                                                                                                      y
          ∂t                                                       Hz

                                                              Ey

                                         x
PML
Absorbing
Boundary
Conditions

   • 

Introduces
“artificial”
anisotropic
electric
   /magnetic*
conductivities
within
domain
   
boundaries
allowing
for
absorption
   /attenuation
waves.

   • 

Employs
a
numerical
“split-field”
approach
   
allowing
perfect
transmission
into
absorbing
   
layer
(regardless
of
frequency,
polarization,
or
   
angle
of
incidence).

                                                                                  J. P. Bérenger, IEEE Trans. Antennas Propagat., 44(110) 1996.
        Nanostructures
Research
Group

       CENTER FOR SOLID STATE ELECTRONICS RESEARCH
Sheen, et. al. , IEEE- MTT, 38(7), 1990.




Nanostructures
Research
Group

CENTER FOR SOLID STATE ELECTRONICS RESEARCH
• 
 
 Stability
limit,
called
the
CFL
criterion
severely
limits
maximum
timestep
for

solution
of
PDEs
on
a
finite
grid.


                                                        1
                        Δt FDTD ≤
                                                    2        2           2
                                               ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞
                                      υ max    ⎜    ⎟ +⎜
                                                       ⎜ Δy ⎟ + ⎜ Δz ⎟
                                                            ⎟
                                               ⎝ Δx ⎠ ⎝     ⎠ ⎝ ⎠

•       CFL
 criterion
 can
 be
 relaxed
 using
 newly
 reported
 ADI-FDTD

method.



           
 
 Requires
 both
 implicit
 and
 explicit
 field
 updates
 thus
 more
          
time
spent
per
FDTD
timestep.


           Allows
 for
 timesteps
 several
 orders
 of
 magnitude
 larger
 than
          
conventional
limit.


           Tradeoff
b/w
accuracy
and
chosen
timestep.

                                                            T. Namiki, IEEE MTT 47(10), 2003 (1999).
                                                            F. Zheng, et. al, Microwave Guided Wave Lett., 9(11), 441 (1999).
     Nanostructures
Research
Group

     CENTER FOR SOLID STATE ELECTRONICS RESEARCH
Steps
full-wave
simulation:

FDTD:

                                                     Initialization

          ∂H
∇ × E = −µ                                            1.    Obtain
steady-state
solution
for
specific
dc
            ∂t                                              
bias
point
(CMC/Poisson)
and
store
E
fields
           
         ∂E                                               
and
J.

∇× H = ε     +J
          ∂t                                          2.    Initialize
H
field
in
FDTD
solver
using:

                                                                           
CMC:

                                                                        ∇× E = 0
                                                                            dc  dc
                                                                        ∇× H = J
                   1    ⎛ N (i , j ,k ) ⎞
 J (i, j , k ) =        ⎜ ∑ S n vn ⎟
                 ΔxΔyΔz ⎜ n =1          ⎟             3.    Apply
excitation
source
and
begin
                        ⎝               ⎠                   
updating
fields:


                                                                         
                        J tot                                           ∂E 1
                                                                        ∂t ε
                                                                              [   ac  tot  dc
                                                                            = ∇× H − J − J(             )]
   CMC
                                 FDTD
                           ∂H
                                                                          
                                                                               1   
                                                                          = − ∇× E
                   (Etot , H tot   )                                     ∂t    µ
        Nanostructures
Research
Group

        CENTER FOR SOLID STATE ELECTRONICS RESEARCH
E AC + E DC
                                       Start  
                                                   (
                                                                                          H AC + H DC   )

                                Run
CMC
for
DC

                                  bias
point.
                                 Update
particles
using


                                                                               newly
computed
fields.

                                    DC
                                  E x,y,z (x, y, z)
                                    DC
                                  J x,y,z (x, y, z)                                       Total
                                                                                                  .
                                                                                        J x,y,z (x, y, z;t)



                 Apply
small-signal


                 excitation
source
                   Update
E,
H
Fields



                                                           AC
                                                         E x,y,z (x, y, z;t)

              FDTD
Solver
                                 AC
                                                         H x,y,z (x, y, z;t)



                                                             t = t MAX ?
                                                                               No   

                                                                   Yes     

                                                               Stop
Nanostructures
Research
Group

CENTER FOR SOLID STATE ELECTRONICS RESEARCH
• 

Transverse
E-fields
computed
via
2D
Poisson
solver

and
applied
to
source
plane
at
each
timestep.





                     −
                         (t −t0 )2
Vgs (t ) = Δυ gs e         T2




                                                         z

                                                              y


                                                                   x





       Nanostructures
Research
Group

       CENTER FOR SOLID STATE ELECTRONICS RESEARCH
⎡ ℑ( out (ω , zi )⎤
                                            V
 Voltage
gain:
            Gain = 20 log ⎢                 ⎥
                                         ⎣ ℑ( in (ω , z0 )⎦
                                             V

                                               − S 21
  Current
gain:
           h21 =
                                   (1 − S11 )(1 + S 22 )+ S12 S 21
                                                                                    Voltage
Gain

                             S11 : input reflection coefficient
                             S 22 : input reflection coefficient
                             S12 : reverse transmission coefficient
                             S 21 : forward transmission coefficient       5

                                                                           4

                                                                           3

                                                                       ]   2
                                                                       B
                                                                       d   1
0.1µm gate MESFET (125µm width)                                        [
80 x 25 x 30 uniform mesh                                              n 0
                                                                       i
Gaussian pulse excitation (0.1V peak AC amplitude)                     a
100,000 particles                                                      G -1
                                                                                             170 GHz
ΔtPoisson= 5x10-15 s                                                       -2
ΔtFDTD = 4x10-17 s                                                                  Current
gain

                                                                           -3
10-layer PML ABC
                                                                           -4
Simul. time = 6.5 days (3GHz 64-bit Xeon, 8GB RAM)
                                                                           -5
                                                                                0     50    100         150         200   250   300
                                                                                                  Frequency [GHz]
       Nanostructures
Research
Group

       CENTER FOR SOLID STATE ELECTRONICS RESEARCH
Start Time-Stepping                      (t =0 )



                                                                                     n+1 2
                                                                        Update E x           implicitly along y direction for all x, y, z
• 

Coupling
ADI-FDTD
with
CMC
simulator.
                              Update E y
                                                                                     n+1 2
                                                                                             implicitly along z direction for all x, y, z




                                                     Sub-Iteration #1
                                                                                     n+1 2
                                                                        Update Ez            implicitly along z direction for all x, y, z
• 

Timestep
is
split
into
(2)
sub-iterations.
                                                                                                   t = (n + 1 2)Δt
                                                                                                   n+1 2
                                                                                Update H x                 explicitly for all x, y, z
• 

E-fields
are
updated
implicitly
along                                       Update H y
                                                                                                   n+1 2
                                                                                                           explicitly for all x, y, z

specific
directions.
                                                          Update H z
                                                                                                   n+1 2
                                                                                                           explicitly for all x, y, z


• 

H-fields
are
updated
explicitly

throughout.
                                                           Update E x
                                                                                      n+1
                                                                                             implicitly along z direction for all x, y, z
                                                                                     n+1




                                                     Sub-Iteration #2
                                                                        Update E y          implicitly along x direction for all x, y, z
                                                                                     n+1
                                                                        Update Ez           implicitly along y direction for all x, y, z
                                                                                                                                                  t = (n + 1)Δt
                                                                                                    n+1
                                                                                Update H x                 explicitly for all x, y, z
                                                                                                    n+1
                                                                                Update H y                 explicitly for all x, y, z

   Larger
ΔtFDTD
possible

                                                                                                    n+1
                                                                                Update H z                 explicitly for all x, y, z



   Shorter
simulation
times
                                                                                                 NO (t < t max )
                                                                                                  Time-Stepping
                                                                                                    Complete?
                                                                                                                    YES (t = t max )


                                                                                               End Time-Stepping

       Nanostructures
Research
Group

       CENTER FOR SOLID STATE ELECTRONICS RESEARCH
Nanostructures
Research
Group

CENTER FOR SOLID STATE ELECTRONICS RESEARCH

Contenu connexe

Tendances

Lesson 12: Linear Approximation
Lesson 12: Linear ApproximationLesson 12: Linear Approximation
Lesson 12: Linear ApproximationMatthew Leingang
 
Lesson 16: Inverse Trigonometric Functions
Lesson 16: Inverse Trigonometric FunctionsLesson 16: Inverse Trigonometric Functions
Lesson 16: Inverse Trigonometric FunctionsMatthew Leingang
 
Solar Cells Lecture 4: What is Different about Thin-Film Solar Cells?
Solar Cells Lecture 4: What is Different about Thin-Film Solar Cells?Solar Cells Lecture 4: What is Different about Thin-Film Solar Cells?
Solar Cells Lecture 4: What is Different about Thin-Film Solar Cells?Tuong Do
 
Electromagnetic Wave
Electromagnetic WaveElectromagnetic Wave
Electromagnetic WaveYong Heui Cho
 
14.40 o8 s wimbush
14.40 o8 s wimbush14.40 o8 s wimbush
14.40 o8 s wimbushNZIP
 
Solar Cells Lecture 1: Introduction to Photovoltaics
Solar Cells Lecture 1: Introduction to PhotovoltaicsSolar Cells Lecture 1: Introduction to Photovoltaics
Solar Cells Lecture 1: Introduction to PhotovoltaicsTuong Do
 
PAC-Bayesian Bound for Gaussian Process Regression and Multiple Kernel Additi...
PAC-Bayesian Bound for Gaussian Process Regression and Multiple Kernel Additi...PAC-Bayesian Bound for Gaussian Process Regression and Multiple Kernel Additi...
PAC-Bayesian Bound for Gaussian Process Regression and Multiple Kernel Additi...Taiji Suzuki
 
Bouguet's MatLab Camera Calibration Toolbox
Bouguet's MatLab Camera Calibration ToolboxBouguet's MatLab Camera Calibration Toolbox
Bouguet's MatLab Camera Calibration ToolboxYuji Oyamada
 
Instantons and Chern-Simons Terms in AdS4/CFT3: Gravity on the Brane?
Instantons and Chern-Simons Terms in AdS4/CFT3: Gravity on the Brane?Instantons and Chern-Simons Terms in AdS4/CFT3: Gravity on the Brane?
Instantons and Chern-Simons Terms in AdS4/CFT3: Gravity on the Brane?Sebastian De Haro
 
TU3.T10.2.pdf
TU3.T10.2.pdfTU3.T10.2.pdf
TU3.T10.2.pdfgrssieee
 
IGARSS_AMASM_woo_20110727.pdf
IGARSS_AMASM_woo_20110727.pdfIGARSS_AMASM_woo_20110727.pdf
IGARSS_AMASM_woo_20110727.pdfgrssieee
 
Camera calibration
Camera calibrationCamera calibration
Camera calibrationYuji Oyamada
 
Solar Cells Lecture 2: Physics of Crystalline Solar Cells
Solar Cells Lecture 2: Physics of Crystalline Solar CellsSolar Cells Lecture 2: Physics of Crystalline Solar Cells
Solar Cells Lecture 2: Physics of Crystalline Solar CellsTuong Do
 
Anuj 10mar2016
Anuj 10mar2016Anuj 10mar2016
Anuj 10mar2016Anuj012
 
A Genetic Programming Challenge: Evolving the Energy Function for Protein Str...
A Genetic Programming Challenge: Evolving the Energy Function for Protein Str...A Genetic Programming Challenge: Evolving the Energy Function for Protein Str...
A Genetic Programming Challenge: Evolving the Energy Function for Protein Str...Natalio Krasnogor
 

Tendances (20)

Lesson 12: Linear Approximation
Lesson 12: Linear ApproximationLesson 12: Linear Approximation
Lesson 12: Linear Approximation
 
Lesson 16: Inverse Trigonometric Functions
Lesson 16: Inverse Trigonometric FunctionsLesson 16: Inverse Trigonometric Functions
Lesson 16: Inverse Trigonometric Functions
 
Solar Cells Lecture 4: What is Different about Thin-Film Solar Cells?
Solar Cells Lecture 4: What is Different about Thin-Film Solar Cells?Solar Cells Lecture 4: What is Different about Thin-Film Solar Cells?
Solar Cells Lecture 4: What is Different about Thin-Film Solar Cells?
 
Electromagnetic Wave
Electromagnetic WaveElectromagnetic Wave
Electromagnetic Wave
 
14.40 o8 s wimbush
14.40 o8 s wimbush14.40 o8 s wimbush
14.40 o8 s wimbush
 
Jokyokai2
Jokyokai2Jokyokai2
Jokyokai2
 
Solar Cells Lecture 1: Introduction to Photovoltaics
Solar Cells Lecture 1: Introduction to PhotovoltaicsSolar Cells Lecture 1: Introduction to Photovoltaics
Solar Cells Lecture 1: Introduction to Photovoltaics
 
PAC-Bayesian Bound for Gaussian Process Regression and Multiple Kernel Additi...
PAC-Bayesian Bound for Gaussian Process Regression and Multiple Kernel Additi...PAC-Bayesian Bound for Gaussian Process Regression and Multiple Kernel Additi...
PAC-Bayesian Bound for Gaussian Process Regression and Multiple Kernel Additi...
 
Bouguet's MatLab Camera Calibration Toolbox
Bouguet's MatLab Camera Calibration ToolboxBouguet's MatLab Camera Calibration Toolbox
Bouguet's MatLab Camera Calibration Toolbox
 
Electricmotor3
Electricmotor3Electricmotor3
Electricmotor3
 
Instantons and Chern-Simons Terms in AdS4/CFT3: Gravity on the Brane?
Instantons and Chern-Simons Terms in AdS4/CFT3: Gravity on the Brane?Instantons and Chern-Simons Terms in AdS4/CFT3: Gravity on the Brane?
Instantons and Chern-Simons Terms in AdS4/CFT3: Gravity on the Brane?
 
Electromagnetics
ElectromagneticsElectromagnetics
Electromagnetics
 
TU3.T10.2.pdf
TU3.T10.2.pdfTU3.T10.2.pdf
TU3.T10.2.pdf
 
IGARSS_AMASM_woo_20110727.pdf
IGARSS_AMASM_woo_20110727.pdfIGARSS_AMASM_woo_20110727.pdf
IGARSS_AMASM_woo_20110727.pdf
 
Camera calibration
Camera calibrationCamera calibration
Camera calibration
 
Solar Cells Lecture 2: Physics of Crystalline Solar Cells
Solar Cells Lecture 2: Physics of Crystalline Solar CellsSolar Cells Lecture 2: Physics of Crystalline Solar Cells
Solar Cells Lecture 2: Physics of Crystalline Solar Cells
 
Anuj 10mar2016
Anuj 10mar2016Anuj 10mar2016
Anuj 10mar2016
 
Smith Chart
Smith ChartSmith Chart
Smith Chart
 
A Genetic Programming Challenge: Evolving the Energy Function for Protein Str...
A Genetic Programming Challenge: Evolving the Energy Function for Protein Str...A Genetic Programming Challenge: Evolving the Energy Function for Protein Str...
A Genetic Programming Challenge: Evolving the Energy Function for Protein Str...
 
17.04.2012 seminar trions_kochereshko
17.04.2012 seminar trions_kochereshko17.04.2012 seminar trions_kochereshko
17.04.2012 seminar trions_kochereshko
 

En vedette

Fully-Polarimetric_Phased_Array_Far_Field_Modeling_2015SECONWorkshop_JHucks
Fully-Polarimetric_Phased_Array_Far_Field_Modeling_2015SECONWorkshop_JHucksFully-Polarimetric_Phased_Array_Far_Field_Modeling_2015SECONWorkshop_JHucks
Fully-Polarimetric_Phased_Array_Far_Field_Modeling_2015SECONWorkshop_JHucksJoseph Hucks, Ph.D.
 
Agilent impedance measurements handbook
Agilent impedance measurements handbookAgilent impedance measurements handbook
Agilent impedance measurements handbookMohammed Benlamlih
 
Intro to MATLAB GUI
Intro to MATLAB GUIIntro to MATLAB GUI
Intro to MATLAB GUIAsjad Ali
 
Concurrent 2.4/5-GHz Multi-Loop MIMO Antennas with Wide 3-dB Beamwidth Radiat...
Concurrent 2.4/5-GHz Multi-Loop MIMO Antennas with Wide 3-dB Beamwidth Radiat...Concurrent 2.4/5-GHz Multi-Loop MIMO Antennas with Wide 3-dB Beamwidth Radiat...
Concurrent 2.4/5-GHz Multi-Loop MIMO Antennas with Wide 3-dB Beamwidth Radiat...Saou-Wen Su
 
Phased array radar antennas - Anten mảng pha
Phased array radar antennas - Anten mảng phaPhased array radar antennas - Anten mảng pha
Phased array radar antennas - Anten mảng phaTuấn Trần
 
2010 APS_ Broadband Characteristics of A Dome Dipole Antenna
2010 APS_ Broadband Characteristics of A Dome Dipole Antenna2010 APS_ Broadband Characteristics of A Dome Dipole Antenna
2010 APS_ Broadband Characteristics of A Dome Dipole AntennaJing Zhao
 
Ajal 2 ppt
Ajal 2 pptAjal 2 ppt
Ajal 2 pptAJAL A J
 
An Internal Wideband Monopole Antenna for UMTS/WLAN Dual-Mode Mobile Phone
An Internal Wideband Monopole Antenna for UMTS/WLAN Dual-Mode Mobile PhoneAn Internal Wideband Monopole Antenna for UMTS/WLAN Dual-Mode Mobile Phone
An Internal Wideband Monopole Antenna for UMTS/WLAN Dual-Mode Mobile PhoneSaou-Wen Su
 
Multi-Funtion Phased Array Radar
Multi-Funtion Phased Array RadarMulti-Funtion Phased Array Radar
Multi-Funtion Phased Array RadarMistral Solutions
 
Smartphone Hardware Architecture
Smartphone Hardware ArchitectureSmartphone Hardware Architecture
Smartphone Hardware ArchitectureYong Heui Cho
 
Fundamentals of Array Antenna
Fundamentals of Array AntennaFundamentals of Array Antenna
Fundamentals of Array AntennaYong Heui Cho
 
Noc ajal final
Noc ajal  finalNoc ajal  final
Noc ajal finalAJAL A J
 
Active Phased Array Radar Systems
Active Phased Array Radar SystemsActive Phased Array Radar Systems
Active Phased Array Radar SystemsReza Taryghat
 

En vedette (20)

WIMAX
WIMAXWIMAX
WIMAX
 
Fully-Polarimetric_Phased_Array_Far_Field_Modeling_2015SECONWorkshop_JHucks
Fully-Polarimetric_Phased_Array_Far_Field_Modeling_2015SECONWorkshop_JHucksFully-Polarimetric_Phased_Array_Far_Field_Modeling_2015SECONWorkshop_JHucks
Fully-Polarimetric_Phased_Array_Far_Field_Modeling_2015SECONWorkshop_JHucks
 
Semiconductor
SemiconductorSemiconductor
Semiconductor
 
Agilent impedance measurements handbook
Agilent impedance measurements handbookAgilent impedance measurements handbook
Agilent impedance measurements handbook
 
Chapter 14
Chapter 14Chapter 14
Chapter 14
 
Intro to MATLAB GUI
Intro to MATLAB GUIIntro to MATLAB GUI
Intro to MATLAB GUI
 
Concurrent 2.4/5-GHz Multi-Loop MIMO Antennas with Wide 3-dB Beamwidth Radiat...
Concurrent 2.4/5-GHz Multi-Loop MIMO Antennas with Wide 3-dB Beamwidth Radiat...Concurrent 2.4/5-GHz Multi-Loop MIMO Antennas with Wide 3-dB Beamwidth Radiat...
Concurrent 2.4/5-GHz Multi-Loop MIMO Antennas with Wide 3-dB Beamwidth Radiat...
 
Phased-Array Radar Talk Jorge Salazar
Phased-Array Radar Talk Jorge SalazarPhased-Array Radar Talk Jorge Salazar
Phased-Array Radar Talk Jorge Salazar
 
Phased array radar antennas - Anten mảng pha
Phased array radar antennas - Anten mảng phaPhased array radar antennas - Anten mảng pha
Phased array radar antennas - Anten mảng pha
 
2010 APS_ Broadband Characteristics of A Dome Dipole Antenna
2010 APS_ Broadband Characteristics of A Dome Dipole Antenna2010 APS_ Broadband Characteristics of A Dome Dipole Antenna
2010 APS_ Broadband Characteristics of A Dome Dipole Antenna
 
Ajal 2 ppt
Ajal 2 pptAjal 2 ppt
Ajal 2 ppt
 
An Internal Wideband Monopole Antenna for UMTS/WLAN Dual-Mode Mobile Phone
An Internal Wideband Monopole Antenna for UMTS/WLAN Dual-Mode Mobile PhoneAn Internal Wideband Monopole Antenna for UMTS/WLAN Dual-Mode Mobile Phone
An Internal Wideband Monopole Antenna for UMTS/WLAN Dual-Mode Mobile Phone
 
array and phased array antennna
array and phased array antennnaarray and phased array antennna
array and phased array antennna
 
Ajal 3
Ajal 3Ajal 3
Ajal 3
 
Matlab GUI
Matlab GUIMatlab GUI
Matlab GUI
 
Multi-Funtion Phased Array Radar
Multi-Funtion Phased Array RadarMulti-Funtion Phased Array Radar
Multi-Funtion Phased Array Radar
 
Smartphone Hardware Architecture
Smartphone Hardware ArchitectureSmartphone Hardware Architecture
Smartphone Hardware Architecture
 
Fundamentals of Array Antenna
Fundamentals of Array AntennaFundamentals of Array Antenna
Fundamentals of Array Antenna
 
Noc ajal final
Noc ajal  finalNoc ajal  final
Noc ajal final
 
Active Phased Array Radar Systems
Active Phased Array Radar SystemsActive Phased Array Radar Systems
Active Phased Array Radar Systems
 

Similaire à Global Modeling of High-Frequency Devices

Em theory lecture
Em theory lectureEm theory lecture
Em theory lecturej sarma
 
Phys e8(2000)1
Phys e8(2000)1Phys e8(2000)1
Phys e8(2000)1FISICO2012
 
Spectroscopic ellipsometry
Spectroscopic ellipsometrySpectroscopic ellipsometry
Spectroscopic ellipsometrynirupam12
 
SCF methods, basis sets, and integrals part III
SCF methods, basis sets, and integrals part IIISCF methods, basis sets, and integrals part III
SCF methods, basis sets, and integrals part IIIAkefAfaneh2
 
Waveguiding Structures Part 1 (General Theory).pptx
Waveguiding Structures Part 1 (General Theory).pptxWaveguiding Structures Part 1 (General Theory).pptx
Waveguiding Structures Part 1 (General Theory).pptxPawanKumar391848
 
Efficient mode-matching analysis of 2-D scattering by periodic array of circu...
Efficient mode-matching analysis of 2-D scattering by periodic array of circu...Efficient mode-matching analysis of 2-D scattering by periodic array of circu...
Efficient mode-matching analysis of 2-D scattering by periodic array of circu...Yong Heui Cho
 
Slide of computer networks introduction to computer networks
Slide of computer networks introduction to computer networksSlide of computer networks introduction to computer networks
Slide of computer networks introduction to computer networksMohammedAbbas653737
 
Vibration energy harvesting under uncertainty
Vibration energy harvesting under uncertaintyVibration energy harvesting under uncertainty
Vibration energy harvesting under uncertaintyUniversity of Glasgow
 
481 lecture10
481 lecture10481 lecture10
481 lecture10david s
 
Micro-nanosystems for electrical metrology and precision instrumentation
Micro-nanosystems for electrical metrology and precision instrumentationMicro-nanosystems for electrical metrology and precision instrumentation
Micro-nanosystems for electrical metrology and precision instrumentationdie_dex
 

Similaire à Global Modeling of High-Frequency Devices (20)

Fields Lec 5&amp;6
Fields Lec 5&amp;6Fields Lec 5&amp;6
Fields Lec 5&amp;6
 
ch09.pdf
ch09.pdfch09.pdf
ch09.pdf
 
Em theory lecture
Em theory lectureEm theory lecture
Em theory lecture
 
Ennaoui cours rabat part ii
Ennaoui cours rabat part iiEnnaoui cours rabat part ii
Ennaoui cours rabat part ii
 
quantum dots
quantum dotsquantum dots
quantum dots
 
Phys e8(2000)1
Phys e8(2000)1Phys e8(2000)1
Phys e8(2000)1
 
Spectroscopic ellipsometry
Spectroscopic ellipsometrySpectroscopic ellipsometry
Spectroscopic ellipsometry
 
Chapter 3 wave_optics
Chapter 3 wave_opticsChapter 3 wave_optics
Chapter 3 wave_optics
 
SCF methods, basis sets, and integrals part III
SCF methods, basis sets, and integrals part IIISCF methods, basis sets, and integrals part III
SCF methods, basis sets, and integrals part III
 
Waveguiding Structures Part 1 (General Theory).pptx
Waveguiding Structures Part 1 (General Theory).pptxWaveguiding Structures Part 1 (General Theory).pptx
Waveguiding Structures Part 1 (General Theory).pptx
 
Efficient mode-matching analysis of 2-D scattering by periodic array of circu...
Efficient mode-matching analysis of 2-D scattering by periodic array of circu...Efficient mode-matching analysis of 2-D scattering by periodic array of circu...
Efficient mode-matching analysis of 2-D scattering by periodic array of circu...
 
Electro magnetic waves
Electro magnetic wavesElectro magnetic waves
Electro magnetic waves
 
Electromagnetic.pdf
Electromagnetic.pdfElectromagnetic.pdf
Electromagnetic.pdf
 
Slide of computer networks introduction to computer networks
Slide of computer networks introduction to computer networksSlide of computer networks introduction to computer networks
Slide of computer networks introduction to computer networks
 
1 1 4
1 1 41 1 4
1 1 4
 
1 1 4
1 1 41 1 4
1 1 4
 
Vibration energy harvesting under uncertainty
Vibration energy harvesting under uncertaintyVibration energy harvesting under uncertainty
Vibration energy harvesting under uncertainty
 
4 b5lecture62008
4 b5lecture620084 b5lecture62008
4 b5lecture62008
 
481 lecture10
481 lecture10481 lecture10
481 lecture10
 
Micro-nanosystems for electrical metrology and precision instrumentation
Micro-nanosystems for electrical metrology and precision instrumentationMicro-nanosystems for electrical metrology and precision instrumentation
Micro-nanosystems for electrical metrology and precision instrumentation
 

Global Modeling of High-Frequency Devices

  • 2. Full-Band

 Full-Wave
 Simulator
 Simulator
 6 4 2 0 -2 -4 -6 Γ X U,K L L Γ Nanostructures
Research
Group
 CENTER FOR SOLID STATE ELECTRONICS RESEARCH
  • 3. When
devices
are
operated
at
high
frequencies:
 •  Coupling
between
fluctuation
in
charge
distribution
and 
propagating
EM
fields
must
be
included
into
simulation
model.
 • 

As
operating
frequencies
increase,
period
of
EM
waves 
approaches
relaxation
time
of
carriers
in
semiconductor
material.
 • 

Finite
amount
of
time
for
carrier
to
react
to
changes
in
applied 
fields
(i.e.
changes
in
particle
velocities)
 Transport
directly
affected 
by
EM
wave
propagation
 Nanostructures
Research
Group
 CENTER FOR SOLID STATE ELECTRONICS RESEARCH
  • 4. Poisson
solvers
are
unable
to:
 •  directly
capture
inherent
“carrier-wave”
interaction.
 • 
account
for
existing
magnetic
fields
in
real
device.
 Full-wave
solver
can:
 •  directly
solve
full
set
of
EM
field
equations.
 • 
account
for
externally
applied
sources
and
changes
in
the 
field
due
to
charge
fluctuations.
 • 
directly
simulate
absorption/emission
of
EM
energy
in/out
of
 
system
(i.e.
optical
excitation,
radiative
processes,
THz 
devices.)
 Nanostructures
Research
Group
 CENTER FOR SOLID STATE ELECTRONICS RESEARCH
  • 5. M. Saraniti and S.M. Goodnick, IEEE TED, 47, 1909 (2000) K. Kometer, G. Zandler, and P. Vogl, Phys. Rev., B46(3), 1382 (1992) particle
dynamics
 choose
scattering
 Ensemble
Monte
Carlo
(EMC)
 
new
energy
  
computationally
slow
  
low
memory
requirements
 
find
new
k
with
 
dispersion
relation
 VS.
 Cellular
Monte
Carlo
(CMC)
   computationally
fast
 choose
new
k
  
high
memory
requirements
 Nanostructures
Research
Group
 CENTER FOR SOLID STATE ELECTRONICS RESEARCH
  • 6. Idea:

 use
MC
scattering
in
regions
of
band
structure
where
scattering
is
low.
   Nearly
as
fast
as
CMC.
   Reduces
memory
usage.
 H ybrid/ MC perf ormance ratio time per iter. [sec/ 5000 e ] - 6 4 energ y [eV] 2 0 -2 -4 EMC -6 CMC X U,K L L L field [V/m] wave vector Nanostructures
Research
Group
 CENTER FOR SOLID STATE ELECTRONICS RESEARCH
  • 7. z K.S. Yee, IEEE Trans. Antennas Propagat., 14(302) 1966 “Yee cell” Maxwell’s
equations
 • 

Most
direct
explicit
solution
of Ey 
Maxwell’s
equations
available
(i.e.  Ex Ex 
no
matrix
inversion
required).
  Hz ∂H Ez ∇ × E = −µ Ex Ey • 

A
complete
“full-wave”
method ∂t Hx 
without
approximation
(i.e.
no
pre  Hy Hy -selection
of
output
modes
or  ∂E  Ez Ex 
solution
form
necessary.)
 ∇× H = ε +J Ex Hx Ey Ex y ∂t Hz Ey x PML
Absorbing
Boundary
Conditions
 • 

Introduces
“artificial”
anisotropic
electric /magnetic*
conductivities
within
domain 
boundaries
allowing
for
absorption /attenuation
waves.
 • 

Employs
a
numerical
“split-field”
approach 
allowing
perfect
transmission
into
absorbing 
layer
(regardless
of
frequency,
polarization,
or 
angle
of
incidence).
 J. P. Bérenger, IEEE Trans. Antennas Propagat., 44(110) 1996. Nanostructures
Research
Group
 CENTER FOR SOLID STATE ELECTRONICS RESEARCH
  • 8. Sheen, et. al. , IEEE- MTT, 38(7), 1990. Nanostructures
Research
Group
 CENTER FOR SOLID STATE ELECTRONICS RESEARCH
  • 9. • 
 
 Stability
limit,
called
the
CFL
criterion
severely
limits
maximum
timestep
for 
solution
of
PDEs
on
a
finite
grid.
 1 Δt FDTD ≤ 2 2 2 ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ υ max ⎜ ⎟ +⎜ ⎜ Δy ⎟ + ⎜ Δz ⎟ ⎟ ⎝ Δx ⎠ ⎝ ⎠ ⎝ ⎠ •  CFL
 criterion
 can
 be
 relaxed
 using
 newly
 reported
 ADI-FDTD 
method.

  
 
 Requires
 both
 implicit
 and
 explicit
 field
 updates
 thus
 more 
time
spent
per
FDTD
timestep.
  Allows
 for
 timesteps
 several
 orders
 of
 magnitude
 larger
 than 
conventional
limit.
  Tradeoff
b/w
accuracy
and
chosen
timestep.
 T. Namiki, IEEE MTT 47(10), 2003 (1999). F. Zheng, et. al, Microwave Guided Wave Lett., 9(11), 441 (1999). Nanostructures
Research
Group
 CENTER FOR SOLID STATE ELECTRONICS RESEARCH
  • 10. Steps
full-wave
simulation:
 FDTD:
  Initialization
  ∂H ∇ × E = −µ 1.  Obtain
steady-state
solution
for
specific
dc ∂t 
bias
point
(CMC/Poisson)
and
store
E
fields   ∂E  
and
J.
 ∇× H = ε +J ∂t 2.  Initialize
H
field
in
FDTD
solver
using:
  CMC:
 ∇× E = 0  dc  dc ∇× H = J 1 ⎛ N (i , j ,k ) ⎞ J (i, j , k ) = ⎜ ∑ S n vn ⎟ ΔxΔyΔz ⎜ n =1 ⎟ 3.  Apply
excitation
source
and
begin ⎝ ⎠ 
updating
fields:
   J tot ∂E 1 ∂t ε [  ac  tot  dc = ∇× H − J − J( )] CMC
 FDTD
 ∂H  1    = − ∇× E (Etot , H tot ) ∂t µ Nanostructures
Research
Group
 CENTER FOR SOLID STATE ELECTRONICS RESEARCH
  • 11. E AC + E DC Start 
 ( H AC + H DC ) Run
CMC
for
DC
 bias
point.
 Update
particles
using

 newly
computed
fields.
 DC E x,y,z (x, y, z) DC J x,y,z (x, y, z) Total . J x,y,z (x, y, z;t) Apply
small-signal

 excitation
source
 Update
E,
H
Fields
 AC E x,y,z (x, y, z;t) FDTD
Solver
 AC H x,y,z (x, y, z;t) t = t MAX ? No 
 Yes 
 Stop Nanostructures
Research
Group
 CENTER FOR SOLID STATE ELECTRONICS RESEARCH
  • 12. • 

Transverse
E-fields
computed
via
2D
Poisson
solver 
and
applied
to
source
plane
at
each
timestep.
 − (t −t0 )2 Vgs (t ) = Δυ gs e T2 z
 y
 x
 Nanostructures
Research
Group
 CENTER FOR SOLID STATE ELECTRONICS RESEARCH
  • 13. ⎡ ℑ( out (ω , zi )⎤ V Voltage
gain:
 Gain = 20 log ⎢ ⎥ ⎣ ℑ( in (ω , z0 )⎦ V − S 21 Current
gain:
 h21 = (1 − S11 )(1 + S 22 )+ S12 S 21 Voltage
Gain
 S11 : input reflection coefficient S 22 : input reflection coefficient S12 : reverse transmission coefficient S 21 : forward transmission coefficient 5 4 3 ] 2 B d 1 0.1µm gate MESFET (125µm width) [ 80 x 25 x 30 uniform mesh n 0 i Gaussian pulse excitation (0.1V peak AC amplitude) a 100,000 particles G -1 170 GHz ΔtPoisson= 5x10-15 s -2 ΔtFDTD = 4x10-17 s Current
gain
 -3 10-layer PML ABC -4 Simul. time = 6.5 days (3GHz 64-bit Xeon, 8GB RAM) -5 0 50 100 150 200 250 300 Frequency [GHz] Nanostructures
Research
Group
 CENTER FOR SOLID STATE ELECTRONICS RESEARCH
  • 14. Start Time-Stepping (t =0 ) n+1 2 Update E x implicitly along y direction for all x, y, z • 

Coupling
ADI-FDTD
with
CMC
simulator.
 Update E y n+1 2 implicitly along z direction for all x, y, z Sub-Iteration #1 n+1 2 Update Ez implicitly along z direction for all x, y, z • 

Timestep
is
split
into
(2)
sub-iterations.
 t = (n + 1 2)Δt n+1 2 Update H x explicitly for all x, y, z • 

E-fields
are
updated
implicitly
along Update H y n+1 2 explicitly for all x, y, z 
specific
directions.
 Update H z n+1 2 explicitly for all x, y, z • 

H-fields
are
updated
explicitly 
throughout.
 Update E x n+1 implicitly along z direction for all x, y, z n+1 Sub-Iteration #2 Update E y implicitly along x direction for all x, y, z n+1 Update Ez implicitly along y direction for all x, y, z t = (n + 1)Δt n+1 Update H x explicitly for all x, y, z n+1 Update H y explicitly for all x, y, z Larger
ΔtFDTD
possible
 n+1 Update H z explicitly for all x, y, z Shorter
simulation
times
 NO (t < t max ) Time-Stepping Complete? YES (t = t max ) End Time-Stepping Nanostructures
Research
Group
 CENTER FOR SOLID STATE ELECTRONICS RESEARCH