Upload
truongtram
View
225
Download
2
Embed Size (px)
Citation preview
12/4/2017
1
Lecture 5a Slide 1
EE 5337
Computational Electromagnetics
Lecture #5a
Scattering Matrices for Semi‐Analytical Methods These notes may contain copyrighted material obtained under fair use rules. Distribution of these materials is strictly prohibited
InstructorDr. Raymond Rumpf(915) 747‐[email protected]
Outline
• Scattering matrix for a single layer
• Multilayer structures
• Longitudinally periodic structures
• Dispersion analysis
• Alternatives to scattering matrices
Lecture 5a Slide 2
12/4/2017
2
Lecture 5a Slide 3
Scattering Matrix for a Single Layer
R. C. Rumpf, “Improved Formulation of Scattering Matrices for Semi‐Analytical Methods That is Consistent with Convention,” PIERS B, Vol. 35, pp. 241‐261, 2011.
Lecture 5a Slide 4
Motivation for Scattering Matrices
Scattering matrices offer several important features and benefits:
• Unconditionally stable method.• Parameters have physical meaning.• Parameters correspond to those measured in the lab.• Can be used to extract dispersion.• Very memory efficient.• Can be used to exploit longitudinal periodicity.•Mature and proven approach.•Much greater wealth of literature available.
However, excellent alternatives to S‐matrices do exist!
12/4/2017
3
Lecture 5a Slide 5
Geometry of a Single Layer
thfield within layeri z i ψth
th
mode coefficients inside layer
mode coefficients outside layer
i
i
i
i
c
c
Indicates a point that lies on an interface, but associated with a particular side.
0iψ
0iψ
iL
i
i
c
c
1ψ
1ψ
0i ik Lψ
0i ik Lψ
2
2
c
c1
1
c
c
2ψ
2ψ
i iz ψ
i iz ψ
Medium 1 Layer i Medium 2
z
Lecture 5a Slide 6
Definition of A Scattering Matrix
11 121 1
21 222 2
S Sc c
S Sc c11
21
reflection
transmission
S
S
This is consistent with network theory and experimental convention.
12/4/2017
4
Lecture 5a Slide 7
Field Relations
Field inside the ith layer:
,
,
,
,
i i
i i
x i i
zy i i i i i
i i zx i i i i i
y i i
E z
E z ez
H z e
H z
λ
λ
W W c0ψ
V V c0
Boundary conditions at the first interface:
1
1 1 1
1 1 1
0i
i i i
i i i
ψ ψ
W WW W cc
V VV V cc
Boundary conditions at the second interface:
0
0
0 2
2 2 2
2 2 2
i i
i i
i i
k Li i i
k Li i i
k L
e
e
λ
λ
ψ ψ
W W W Wc c0
V V V Vc c0
Note: k0 has been incorporated to normalize Li.
Lecture 5a Slide 8
Derivation of the Scattering Matrix
1
1 1 1
1 1 1
i ii
i ii
W W W Wc c
V V V Vc c
0
0
1
2 2 2
2 2 2
i i
i i
k Li ii
k Li ii
e
e
λ
λ
W W W Wc c0
V V V Vc c0
Solve both boundary condition equations for the intermediate mode coefficients and .i
c ic
Both of these equations have the term1 1 1
1 1
1
2j j ij iji i ij i j i j
j j ij iji i ij i j i j
W W A BW W A W W V V
V V B AV V B W W V V
We set substitute this result into the first two equations and then set them equal to eliminate the intermediate mode coefficients.
0
0
1 1 2 21 2
1 1 2 21 2
1 1
2 2
i i
i i
k Li i i i
k Li i i i
e
e
λ
λ
A B A Bc c0
B A B Ac c0
We write this as two matrix equations and rearrange the terms until they have the form of a scattering matrix.
1 1
2 2
? ?
? ?
c c
c c
See HW2
12/4/2017
5
Lecture 5a Slide 9
The Scattering Matrix
The scattering matrix Si of the ith
layer is defined as:
After some algebra, the components of the scattering matrix are computed according to
1 1
2 2
i
c cS
c c
11 12
21 22
i ii
i i
S SS
S S
1 1
1 1
ij i j i j
ij i j i j
A W W V V
B W W V V
0i ik Li e λX
11 111 1 2 2 1 2 2 1 1
11 112 1 2 2 1 2 2 2 2
11 121 2 1 1 2 1 1 1 1
11 122 2 1 1 2 1 1 2 2
ii i i i i i i i i i i i
ii i i i i i i i i i i
ii i i i i i i i i i i
ii i i i i i i i i i i i
S A X B A X B X B A X A B
S A X B A X B X A B A B
S A X B A X B X A B A B
S A X B A X B X B A X A B
i is the layer number.j is either 1 or 2 depending on which external medium is being referenced.
iS
Lecture 5a Slide 10
Scattering Matrices in the Literature
For some reason, the computational electromagnetics community has: (1) deviated from convention, and (2) formulated inefficient scattering matrices.
1ic
1ic
ic
ic
11 121
21 22 1
i i
i i
S Sc c
S Sc c
12S
11S
21S
22S
reflection
transmission
• Here s11 is not reflection. Instead. it is backward transmission!
• Here s21 is not transmission. Instead, it is a reflection parameter!
• Scattering matrices can not be interchanged.• Scattering matrices are not symmetric so they take twice the memory to store and are more time‐consuming to calculate.
R. C. Rumpf, "Improved formulation of scattering matrices for semi‐analytical methods that is consistent with convention," PIERS B, Vol. 35, 241‐261, 2011.
12/4/2017
6
Lecture 5a Slide 11
Limitation of Conventional S‐Matrix Formulation
Note that the elements of a scattering matrix are a function of materials outside of the layer.
For example, there are only three unique layers in the multilayer structure below, yet 20 separate computations of scattering matrices are needed.
Three unique layers
20 layer stack
This makes it difficult to interchange scattering matrices arbitrarily.
Lecture 5a Slide 12
Solution
To get around this, we will surround each layer with external regions of zero thickness. This lets us connect the scattering matrices in any order because they all calculate fields that exist outside of the layers in the same medium. This will have no effect electromagnetically as long as we make the external regions have zero thickness between layers.
iL
Layer iGap Medium Gap Medium
12/4/2017
7
Lecture 5a Slide 13
Visualization of the Technique
Three unique layers
We calculate the scattering matrices for just the unique layers.
Then we just manipulate these same three scattering matrices to “build” the global scattering matrix.
Gaps between the layers are made to have zero thickness so they have no effect electromagnetically.
Faster! Simpler! Less memory needed!
Lecture 5a Slide 14
Revised Geometry of a Single Layer
0iψ
0iψ
iL
i
i
c
c
1ψ
1ψ
0i ik Lψ
0i ik Lψ
2
2
c
c1
1
c
c
2ψ
2ψ
i iz ψ
i iz ψ
Layer iGap Medium Gap Medium
r,g
r,g
r,g
r,g
same medium
12/4/2017
8
Lecture 5a Slide 15
Calculating Revised Scattering Matrices
The scattering matrix Si of the ith
layer is still defined as:
But the equations to calculate the elements reduce to
1 1
2 2
i
c cS
c c
11 12
21 22
i ii
i i
S SS
S S
1 1g g
1 1g g
i i i
i i i
A W W V V
B W W V V
0i ik Li e λX
11 111
11 112
21 12
22 11
ii i i i i i i i i i i i
ii i i i i i i i i i i
i i
i i
S A X B A X B X B A X A B
S A X B A X B X A B A B
S S
S S
• Layers are symmetric so the scattering matrix elements have redundancy.• Scattering matrix equations are simplified.• Fewer calculations.• Less memory storage.
iS
Lecture 5a Slide 16
Layers in TMM are Actually Four‐Port NetworksWe have written the scattering matrices as 22 block matrices.
11 12
21 22
S S
S S
For TMM, this actually expands to a 44 element scattering matrix.
11 12 13 14
21 22 23 2411 12
31 32 33 3421 22
41 42 43 44
13 1411 1211 12
23 2421 22
33 3431 3221 22
43 4441 42
s s s s
s s s s
s s s s
s s s s
s ss s
s ss s
s ss s
s ss s
S S
S S
S S
S S
Each mode provides an I/O mechanism and there are two modes on each side in each direction.
1
2
3
4
12/4/2017
9
Lecture 5a Slide 17
Scattering Matrices of Lossless Media
If a scattering matrix is composed of materials that have no loss and no gain, the scattering matrix must conserve power. That is, all incident power must either reflect or transmit.
This implies that the scattering matrix is unitary.
If the scattering matrix is unitary, it must obey the following rules:
1
1 1
H
H H
S S
S S SS S S SS I
Note: If the regions external to the layer are different from each other, the scattering matrices will not be unitary. This is because the field amplitudes will be different even though the field carries the same amount of power.
Hints About Stability in These Formulations
• Diagonal elements S11 and S22 tend to be the largest numbers. Divide by these instead of any off‐diagonal elements for best numerical stability.
• X describes propagation through an entire layer. Don’t divide by X or your code can become unstable.
Lecture 5a Slide 18
11 12
21 22
S SS
S S
12/4/2017
10
Lecture 5a Slide 19
Multilayer Structures
Lecture 5a Slide 20
Solution Using Scattering Matrices
The scattering matrix method consists of working through the device one layer at a time and calculating an overall scattering matrix.
1S 2S 3S 4S 5S
device 1 2 3 4 5 S S S S S S
Redheffer star product.NOT matrix multiplication!
12/4/2017
11
Lecture 5a Slide 21
Derivation of the Redheffer Star Product
We start with the equations for the two adjacent scattering matrices.
A A B B11 12 11 122 21 1
A A B B3 32 221 22 21 22
S S S Sc cc c
c cc cS S S S
We expand these into four matrix equations.
A A B B1 11 1 12 2 2 11 2 12 3
A A B B2 21 1 22 2 3 21 2 22 3
Eq. 1 Eq. 3
Eq. 2 Eq. 4
c S c S c c S c S c
c S c S c c S c S c
We substitute Eq. (2) into Eq. (3) to get an equation with only .2c
We substitute Eq. (3) into Eq. (2) to get an equation with only .2c
B A B A B11 22 2 11 21 1 12 3
A B A A B22 11 2 21 1 22 12 3
Eq. 5
Eq. 6
I S S c S S c S c
I S S c S c S S c
We eliminate and by substituting these equations into Eq. (1) and (4). We then rearrange terms into the form of a scattering matrix.
2c 2
c
1 1
3 3
? ?
? ?
c c
c cOverall, this is just algebra. We start with 4 equations and 6 unknowns and reduce it to 2 equations with 4 unknowns.
Lecture 5a Slide 22
Redheffer Star Product
Two scattering matrices may be combined into a single scattering matrix using Redheffer’s star product.
A A11 12
A A21 22
A
S SS
S S
B B11 12
B B21 22
B
S SS
S S AB A B S S S
The combined scattering matrix is then
AB AB11 12
AB AB21 22
AB
S SS
S S
1AB A A B A B A
11 11 12 11 22 11 21
1AB A B A B
12 12 11 22 12
1AB B A B A
21 21 22 11 21
1AB B B A B A B
22 22 21 22 11 22 12
S S S I S S S S
S S I S S S
S S I S S S
S S S I S S S S
R. Redheffer, “Difference equations and functional equations in transmission-line theory,” Modern Mathematics for the Engineer, Vol. 12, pp. 282-337, McGraw-Hill, New York, 1961.
12/4/2017
12
Lecture 5a Slide 23
Putting it All Together (1 of 2)
device S
First, we calculate the device scattering matrix by iterating through each layer of the device and combining the scattering matrices using the Redheffer star product.
1S
1L
1S
2S
2L
0
2S
3S
3L
0
3S
NS
NL
NS
Lecture 5a Slide 24
Putting it All Together (2 of 2)
global S
1S 2S 3S NS
1L 2L 3L NL
0 0
Second, we must connect the device scattering matrix to the external regions to get the global scattering matrix. We use connection scattering matrices to do this.
1S 2S 3S NS
refS
0
0
ref S
trnS
0
0
trn S
12/4/2017
13
Lecture 5a Slide 25
Reflection/Transmission Side Scattering Matrices
The reflection‐side scattering matrix is
ref 111 ref ref
ref 112 ref
ref 121 ref ref ref ref
ref 122 ref ref
2
0.5
S A B
S A
S A B A B
S B A
1 1ref g ref g ref
1 1ref g ref g ref
A W W V V
B W W V V
trn 111 trn trn
trn 112 trn trn trn trn
trn 121 trn
trn 122 trn trn
0.5
2
S B A
S A B A B
S A
S A B
1 1trn g trn g trn
1 1trn g trn g trn
A W W V V
B W W V V
The transmission‐side scattering matrix is
,I
,I
r
r
r,g
r,g
0limL
,II
,II
r
r
r,g
r,g
refs
trns
0limL
Lecture 5a Slide 26
Summary of Using Scattering Matrices
1S 2S 3S NS
1L 2L 3L NL
0 0
refS
0
0
trnS
0
0
ref tgl l rnoba 1 2
device
N
S
S S SS SS
Device in gap medium
deviceS
12/4/2017
14
Lecture 5a Slide 27
Longitudinally Periodic Devices
Lecture 5a Slide 28
Longitudinally Periodic Devices
Suppose we just calculated the scattering matrix for the unit cell of a longitudinally periodic device.
Unit Cell 1 Unit Cell 2 Unit Cell 3 Unit Cell 4 Unit Cell 4 Unit Cell 6 Unit Cell 7 Unit Cell 8
Unit Cell
1 A B C S S S S
There exists a very efficient way of calculating the global scattering matrix of a longitudinally periodic device without calculating and combining all the individual scattering matrices.
A B C
8 A B C A B C A B C A B C A B C A B C A B C A B C S S S S S S S S S S S S S S S S S S S S S S S S S
Both are inefficient!!! 8 1 1 1 1 1 1 1 1 S S S S S S S S S
12/4/2017
15
Lecture 5a Slide 29
Cascading and Doubling
We can quickly build an overall scattering matrix that describes hundreds and thousands of unit cells.
We start by calculating the scattering matrix for a single unit cell.
1 A B C S S S S A B C
Next, we keep connecting the scattering matrix to itself to keep doubling the number of unit cells it describes.
2 1 1 S S S
4 2 2 S S S
8 4 4 S S S
and so on… What if you have a non‐integer power of 2 number of layers?
Lecture 5a Slide 30
Block Diagram for Modified Cascading and Doubling Algorithm
Inputs 1 S-matrix of one unit cell
Number of times to repeat unit cellN
S
Convert N to binary10110
Initialize Algorithm
global bin 1
0 IS S S
I 0
Done?
Perform Doubling bin bin bin S S S
digit = 1?Update S(N)
global global bin S S S
yes
no
no
Output globalSLo
op th
rough
all binary d
igits startin
g with
the least sign
ificant d
igit.
12/4/2017
16
Lecture 5a Slide 31
Example of Cascading and Doubling Algorithm
Inputs to algorithm: 1 scattering matrix for a single unit cell
22 number of unit cells to combineN
S
Step 0 – Calculate scattering matrix for one unit cell
A B C 1 A B C S S S S
Step 2 – Initialize binary and global scattering matrices
global11
global12
global21
global22
S 0
S I
S I
S 0
bin 1S S
Step 1 – Convert N to binary
221’s2’s4’s16’s 8’s
1 1 10 0
Lecture 5a Slide 32
Example of Cascading and Doubling Algorithm
Step 3 – Loop through binary digits
2’s digit = 1 Update S(global) global global bin S S S S(global) now encompasses 2 unit cells
4’s digit = 1 Update S(global) global global bin S S S S(global) now encompasses 6 unit cells
116’s digit = Update S(global) global global bin S S S S(global) now encompasses 22 unit cells
22 101101’s digit = 0 Do not update S(global) S(global) still encompasses 0 unit cells
bin bin bin S S S S(bin) now represents 2 unit cells Double S(bin)
bin bin bin S S S S(bin) now represents 4 unit cells Double S(bin)
bin bin bin S S S S(bin) now represents 8 unit cells Double S(bin)
bin bin bin S S S S(bin) now represents 16 unit cells Double S(bin)
Oops! This algorithm performs one unnecessary doubling operation.How can we fix this?
bin bin bin S S S S(bin) now represents 32 unit cells Double S(bin)
8’s digit = 0 Do not update S(global) S(global) still encompasses 6 unit cells
12/4/2017
17
Lecture 5a Slide 33
Dispersion Analysis
Lecture 5a Slide 34
Dispersion Analysis (1 of 2)
An overall scattering matrix is calculated that describes the unit cell.
uc uc11 120 0
uc uc1 121 22N N
S Sc c
c cS S
The terms are rearranged in “almost” the form of a transfer matrix.
uc uc12 111 0
uc uc1 022 21
N
N
0 S S Ic c
c cI S S 0
If the device is infinitely periodic in the z direction, then the following periodic boundary condition must hold.
1 0
1 0
zzjN
N
ke
c c
c cHere kz is the “effective” propagation constant of the mode.
uc 1 same as on previous slideS S
12/4/2017
18
Lecture 5a Slide 35
Dispersion Analysis (2 of 2)
We substitute the periodic boundary condition into our rearranged equation to get
uc uc11 120 0
uc uc0 021 22
z zjke
S I 0 Sc c
c cS 0 I S
This is a generalized eigen‐value problem.
uc11 0
uc021
uc12
uc22
zzkje
S I cA x
cS 0Ax Bx
0 SB
I S
[V,D] = eig(A,B);
Eigen vectorsBloch modes
Eigen valueskz’s
Lecture 5a Slide 36
Who Cares?
Given kz, we can
1. Calculate the effective properties of the unit cell.
2. Construct band diagrams.
r,eff r,effzk This is an over simplification and beyond the scope of this course.
12/4/2017
19
Lecture 5a Slide 37
Alternatives to Scattering Matrices
Lecture 5a Slide 38
Transmittance Matrices (T‐Matrices)
The T‐matrix method is the transfer matrix method where forward and backward waves are distinguished.
lefttrn 11 12 inc
rightinc 21 22 ref
c T T c
c T T c
Benefits•Much faster (5 to 10 times)•Unconditionally stable
Drawbacks• Less memory efficient•Cannot exploit longitudinal periodicity• Less popular in the literature
M. G. Moharam, Drew A. Pommet, Eric B. Grann, “Stable implementation of the rigorous coupled-wave analysis for surface-relief gratings: enhanced transmittance matrix approach,” J. Opt. Soc. Am. A, Vol. 12, No. 5, pp. 1077-1086, 1995.
12/4/2017
20
Lecture 5a Slide 39
Hybrid Matrices (H‐Matrices)
The h‐matrix method is borrowed from electrical two‐port networks.
1 11 12 1
2 21 22 2
V h h I
I h h V
2 1
2 1
1 111 12
2 20 0
2 221 22
1 20 0
V I
V I
V Vh h
I V
I Ih h
I V
In the framework of fields, the h‐matrix is defined as
, 1 , 1
, 1 11 12 , 1
, ,21 22
, ,
x i x ii i
y i y i
i ix i x i
y i y i
E H
E H
H E
H E
H H
H H
Claimed Benefits• Improved numerical stability•More concise formulation• Simpler to implement• Improved numerical efficiency (30% better than ETM)•Unconditionally stable
Eng L. Tan, “Hybrid-matrix algorithm for rigorous coupled-wave analysis of multilayered diffraction gratings,” J. Mod. Opt., Vol. 53, No. 4, pp. 417-428, 2006.
Lecture 5a Slide 40
R‐Matrices
The R‐matrix method is essentially the impedance matrix framework borrowed from electrical two‐port networks.
1 11 12 1
2 21 22 2
V z z I
V z z I
2 1
2 1
1 111 12
1 20 0
2 221 22
1 20 0
I I
I I
V Vz z
I I
V Vz z
I I
In the framework of fields, the h‐matrix is defined as
, 1 , 1
, 1 11 12 , 1
, ,21 22
, ,
x i x ii i
y i y i
i ix i x i
y i y i
E H
E H
E H
E H
R R
R R
Claimed Benefits•Unconditionally stable• Improved numerical efficiency
Lifeng Li, “Bremmer series, R-matrix propagation algorithm, and numerical modeling of diffraction gratings,” J. Opt. Soc. Am. A, Vol. 11, No. 11, pp. 2829-2836, 1994.