Vous êtes sur la page 1sur 72

A study towards dispersion-effects using slab-waveguide

ERRORS ENCOUNTERED & STAGES OF DEVELOPMENT

Gaussian Beam (normal incidence) Eg=exp(-(y^2)/(2*0.001^2))/sqrt(2*pi*0.001^2) Eg= E0*sin(omega1*x)*sin(theta) COMSOL OUTPUT : Working Gaussian Beam (angular incidence) theta=45 deg BW= 2e-9 nm
Eg= exp((-4*(cos(theta))^2/(BW)^2*(y)^2)+imag(2*pi/lamda*y*sin(theta)))

COMSOL OUTPUT : Not Working

HINT FOR SIMULATING THE FRINGE PATTERN

FRINGE PATTERN GENERATION & MATHEMATICAL TREATMENT

Time : 0sec

Time : 1.569782e-4 sec

Time : 2.162206e-4 sec

Time : 2.196788e-4 sec

ERROR: TIME-STEP TOO SMALL TO EVALUATE PROPOSED FIX: NON DIMENSIONALIZATION OF MAXWELLS EQUATION

Introduction of MATLAB for numerical computations The time co-ordinate is stretched to a [0,1] scale & the time-step for the original time_array is used for timemapping

clc; clear all; %%%%%%%%%%%% Constants for the geometry, in COMSOL%%%%%%%%%%%%%%%%%%%%%%%%%%% mu_Si=1.0; epsilon_Si=11.8; n_Si=3.4255; res_Si=640; sigma_Si=1000; epsilon_PMMA=2.9; mu_PMMA=0.866; n_PMMA=1.4914; res_PMMA=1e-19; sigma_PMMA=1e-4; epsilon_air=1; mu_air=1; n_air=1; c_light=3e+8; res_air=1; sigma_air=3e-15; pi=3.414; %%%%%%%%%%%% Parameters which modulate the Diffraction amount %%%%%%%%%%%%%% BW=2e-9; E0=1; lamda=632e-9; height=3.8e-6; theta=30; %%%%%% %%%%%% %%%%%% %%%%%% %%%%%% Bandwidth of incident beam %%%%%%%%%%%%% Maximum Amplitude of incident beam %%%%% Wavelength of incident beam %%%%%%%%%%% height of the layer in Geometry %%%%%%% angle of incidence %%%%%%%

%%%%%%%%% number of Time-steps (based on time of propagation) %%%%%%% omega1=2*pi*c_light/lamda; freq=2*pi/omega1; delT=lamda/c_light; time_propagation=abs(2*height/(c_light*cos(theta))); %%%%%%% Actual time: 1.6423e-13 %%%%%%%%%% Ntime_step=ceil(time_propagation/delT); %%%%%%% Number of time-steps %%%%%%%%

%%%%%%%%%%% co-ordinate stretching (time-axis) %%%%%%%%%%%%%%%%%%%%%%%%%%% time_array=linspace(0,time_propagation,Ntime_step); N=linspace(0,1,length(time_array)); increment=N(12)-N(11); sprintf('%6f',increment)

Concept Re-defined
Understood that COMSOL was not able to calculate due to memory problem but due to the wrongly assumed Boundary Conditions & improper meshing Understood the concept of PML (perfectly matched layer) for a slab waveguide Followed the procedures for a Photonic Crystal (COMSOL->Modal Library) Introduced the weak-terms and other important co-efficients (h, q, r) in the Model Electric field excitation is achieved by applying a Gaussian beam on the boundary of the top layer Applied it for a Stationary Analysis. It has to be verified for Time-variant and Wave-propagation in any multi-layered slab waveguides

Concept Re-defined
1. Tried to export a different module in the geometry. 2. Combined the PDE co-efficient form with the Structural Mechanics module and simulated the combined modules together. 3. Understood that GardE gives the effective flux, outward direction. 4. Came back to the original PDE form and tried to incorporate the User Models EM Wave propagation Diffraction pattern 5. The above module was an example of Stationary analysis. Hence, simulated the module for a time-dependent analysis and simulated the diffraction patterns. (Double slit fringe diffraction) 6. Re- calculated the simulation parameters and the other solver-time based on the new method for calculation. 7. Reduced the geometry to non-dimensional form and found that the solver-step-time which I was making 1, (delT/t_prop) is absurd. Thus, the correct numerals [0:delT:N*(steps)] and simulated the geometry for the Counter fringe patterns.

CONTOUR VARIATIONS

(in [sec])

6.32e-6

4.425e-5

0.00000632

1.9592e-4

0.00080264

0.00632

0.014587

0.08216

0.7584

1.43464

1.61792

2.04768

2.99568

3.99424

4.99912

5.99678

6.99624

7.268

Speed of light in vaccum (c1): Wavelength of incident beam (lamda) : Height of the layer in Geometry (h1) : Angle of incidence (theta) : Frequency of incident beam (f): Angular Frequency of incident beam (w): Time period of the incident wave (T): Minimum step-time required (T): Time of propagation (Ttotal) : No. of steps required (N):

3e+8 [m/s] 632 nm 3.16e-4 m 30 [deg] 4.746835e+14[Hz] 2.982525e+15 [rad] 2.106667e-15[s] 2.10667e-16[s] 2.432584e-12[s] 11547

Non-dimentionalization
x->x/L t->c*t/L y->y/L L= 10 m Ttotal = 2.432584e-17 [s] delT = 0.00632 Number of time step for simulation (N) = 1154 Tcomsol = 7.29328 [s]

A new perspective
1. Even thought the calculations were exact, the desired diffraction effect was far from achievement. 2. Re-visited the concepts of SPP (surface plasmon polaritons) for the metallo-di electric slab waveguides. 3. Found that every parameter (mu, sigma, n, omega..) are spatially dependent terms and not constant. This led to a significant change in the Maxwells equation and the corresponding boundary conditions in the geometry. 4. Instead of sinusoidal excitation, a Gaussian beam excitation with a fixed wavelength is theoretically calculated for the model. 5. The surface propagation constant (ksp) and other necessary parameters were adjusted, and the geometry was simulated again, based on the procedures followed for obtaining the SPP diffraction effects at the metaldielectric interface. 6. Discussed the possibilities with Sir and the solution is yet to be verified.

CONTOUR

6.32e-6

9.48e-5

8.0264e-4 2.92616 0.014089

7.38808 5.89024

Normal Incidence

Static Analysis of PMMA layer excluded slab waveguide

Drop Tolerance : .0001


Sine/Gauss spatial contour (Ey)

Solver : GMRES
Sine/Gauss temporal contour (Ey)

Normal Incidence

Static Analysis of PMMA layer included slab waveguide

Gauss-spatial contour (Ey) Solver : GMRES Sine-temporal contour (Ey) Solver : GMRES

Drop Tolerance : .0001


Sine-spatial contour (Ey) Solver : GMRES Gauss-temporal contour (Ey) Solver : GMRES

____ Sine Pulse ____ Gaussian Pulse

Transient Analysis of PMMA layer excluded slab waveguide Gaussian Pulse

0s

1.00498 s

Solver : GMRES

Drop Tolerance : .0001

2.5596 s 2.50272 s

Contd..

Drop Tolerance : .0001

4.00056 s 5.03702 s

____ Sine Pulse ____ Gaussian Pulse

Solver : GMRES

6.29472 s 7.0468 s

Transient Analysis of PMMA layer excluded slab waveguide

Sine Pulse

0s

9.48e-5 s

Drop Tolerance : .0001 Solver : GMRES

1.96552 s

0.632 s

Contd..

Normal Incidence

4.97384 s

Solver : GMRES

Drop Tolerance : .0001 7.0468 s

Transient Analysis of PMMA layer included slab waveguide


Gaussian excitation

Gaussian Pulse

6.32e-6s

Drop Tolerance : .0001

Solver : GMRES

3.58344 s

2.212 s

Contd.. Drop Tolerance : .0001

4.2028 s 4.49984 s

Solver : GMRES

7.0468 s

5.03248 s

Transient Analysis of PMMA layer included slab waveguide

Sine Pulse Solver : GMRES

6.32e-6

Drop Tolerance : .0001

2.54064 s

1.34616 s

Contd..

Drop Tolerance : .0001

6.32e-6 s

Solver : GMRES

2.54064 s

1.34616 s

Contd..

Drop Tolerance : .0001

3.8236 s

5.0876 s

Solver : GMRES

7.0468 s

6.58544 s

Completely Lost the track.


1. The assumptions were all incorrect. 2. Surface plasmon and high related high-concept physics has nothing to do with this model. 3. Verified the mathematics and the physics concept behind the scene. 4. Went back to the original PDE co-efficient form. 5. Did a static analysis with a Neumann boundary condition. 6. Did a cross verification different kind of solver settings. 7. Drop Tolerence to 0.0001 while using GMRES solver for both time-dependent and frequency dependent analysis. 8. Got a hint that the mesh-parameters play a pivotal role in FE-analysis. 9. Tried to use a chirped signal but wasnt successful. 10.The modified results are shown below:

SOLVER: Direct (UMFPACK)

STATIONARY ANALYSIS

Case II: electric field exists at the silicon bottom layer. incident boundary electric field has both Ex and Ey silicon boundary (last boundary) Ex 0 & Ey =0

Case I: zero electric field at the silicon bottom layer. incident boundary electric field has both Ex and Ey silicon boundary (last boundary) Ex & Ey =0

Case II: electric field exists at the silicon bottom layer. incident boundary electric field has both Ex and Ey silicon boundary (last boundary) Ex 0 & Ey =0

Coming back to the right path.


1. Got salary for 2-months and bought new books. 2. Bought a new book for FEA for electro-magnetic simulation. 3. Revisited and re-learned the PDE in various form. 4. Learnt that various of Maxwells equation (integral, differential etc.) essentially represent the same thing and that it can be used for any kind of wave-propagation and not only for EM-wave simulations. 5. Got mathematical definition of mesh-parameter adjustment and how it is related for the correct simulation of a geometry. 6. Scraped every model (previously built) and started the simulation freshly. 7. Verified the results for Si-Air layered slab and then finally did a static analysis of the 3-layered geometry.

AN UNBELIEVEBALE MISCONCEPTION FINALLY LED TO THE UNDERLYING PHYSICS

Waveguide Models

Static Analysis

TRANSIENT ANALYSIS
MAXWELLS EQUATION:

SCALAR HELMHOLTZ

VECTORIAL REPRESENTATION

TRANSIENT EQUATION

Which can be effectively derived from the transient form we used in our solution

Thus we have to somehow retrieve and use the value of eigenvalue/propagation constant and use it in the solution for our geometry.

COMSOL TREATMENT FOR TRANSIENT ANALYSIS


HOW COMSOL IS DOING A TRANSIENT ANALYSIS

EXPLANATION

COMPUTATIONAL WINDOW

DETAILED ANALYSIS

FOCUS :CO-EFFICENTS OF THE GENERALIZED PDE

MISCONCEPTION FINALLY CLEARED UNDERSTOOD THE BASICS OF BOTH SLAB WAVEGUIDE & DIELECTRIC INTERFACE

2D 2-LAYER DIELECTRIC INTERFACE STATIONARY ANALYSIS

2D 2-LAYER DIELECTRIC INTERFACE

TRANSIENT ANALYSIS

1D 2-LAYER DIELECTRIC INTERFACE

TANGENTIAL ELECTRIC FIELD COMPONENT

DOUBTS FINALLY CLEARED


1. UNDERSTOOD WHY THE INCIDENT ELECTRIC FIELD IS IN TM-MODE & THE DIELECTRIC -PLANE OF WAVE PROPAGATION IS IN TE MODE 2. HOW WE ARE USING THE GAUSSIAN PULSE/WAVE AS A DIFFRACTION LIMITED WAVE FOR THE STUDY OF DISPERSION AND INTERFERENCE IN A DIELECTRIC MEDIUM 3. GOT THE KEY CONCEPT THAT THE DIRECTION OF ELECTRIC FIELD (LIGHT-WAVE) PROPAGATION IS ALWAYS ORTHOGONAL TO THE PLANE OF LIGHT PROPAGATION 4. UNDERSTOOD THE BASIC PRINCIPLES OF MODE-DECOMPOSITON AND THE RELATIVE IMPORTANCE OF IT IN CEM, FOR FIELD APPROXIMATIONS(FAR-FIELD & NEAR FIELD) 5. RE-DERIVED THE GOVERNING EQUATIONS (MAXWELLS VECTORIAL FORM) AND APPLIED THE PRINCIPLE OF SUPERPOSITION, TO CALCULATE THE TRANSVERSE WAVES IN TERMS OF TWO SIMILAR FIELD (H/E) USING A SCALAR HELMHOLTZ EQUATION 6. UNDERLYING MATHEMATICS OF TIME-HARMONIC FUNCTIONS, UNIFORM-HOMOGENEOUS PALNE WAVE, INTRINSIC IMPEDENCE, COMPLEX PERMITTIVITY, TENSORIAL REPRESENETATION 7. HOW TO RELATE THE NYQUIST-CRITERION, COURANTS CONDITION, STEP-TIME IN TIME MARCHING, SATISFYING A STABILITY CRITERION IN THE NEUMERICAL ANALYSIS

Vous aimerez peut-être aussi