noc19-ee21 Lecture 22-Slab Waveguides

IIT Kanpur July 201
20 followers
2 Video Views·Dec 3, 2023
Slab waveguide, geometric optics, propagation layer, substrate, cladding, modes, Bessel functions, modes, symmetric slab.
Hello and welcome to NPTEL mooc on electromagnetic waves in guided and wireless media and in
this module we study very simplified element of many integrated optical circuits called
as slab waveguide. We have already introduced the idea behind a wave guide in the previous
module, where we imagined that we take a glass slab, okay, and then somehow couple light
into this slab. We also saw hot to couple light into that. And then if you think of
this light propagating inside, light of course being an electromagnetic wave, if it is propagating
inside this slab, provided that the slab refractive index, which we will take it to be N1 is actually
greater than the refractive index outside, which we will take it as N2, which is less
than N1, then these rays will undergo total internal reflection and as they bounce back
and forth between these walls, right, we see that effectively information or energy of
the wave is actually propagating along the length of the slab, so along the slab, which
we will take that axis to be Z axis. So this wave would be propagating in this direction.
However, we also saw that the ray picture that we have seen does not really tell us
more information, especially no information is given or obtained in terms of what the
electric field component would look like and what the magnetic field components or the
magnetic field would look like corresponding to these rays. So if you want to obtain this
information, you need to use Maxwell's equation. Moreover the geometric optics approach that
we looked at in the previous module does not tell us even the existence of what is called
as the evanescent wave, which we saw that will actually result when you go and approach
this problem in the electromagnetic terms, right, that is to say apply Maxwell's equations
and you get a fuller picture than what is possible with the geometric optics. Sure in
some cases geometric optics approach, such as the one that we discussed is actually quite
you know, simplified and it is easy to use that picture to… or qualitatively understand
what is going on, but if you need quantitative answers, you have to you know, like usually
use electromagnetic equations that is Maxwell's equations and that is what we are going to
do. Now before we study the propagation in this slab wave guide, let me point out that
this slab, which we have taken is usually made out of, in the case of an integrated
optical circuit, it is usually made out of substrates, which are silica or other type
of substrate, for example it could be a lithium niobate and you don’t actually get only
a slab, you get multi layer structure, but propagation usually happens in only one layer,
okay so the other layers for many different reasons, but propagation layer will actually
happen in, typically in one of the ones. I mean you try to make light confined into one
powerticular channel or one powerticular waveguide layer itself. Okay, so I am simplifying the
problem. I don’t want to deal with many layers, rather than that we will simplify
the problem to a single layer structure, in which we have channel formed by a material,
whose refractive index is higher than the material refractive index outside. Now in
practice, it is quite difficult to, you know, or it's not that difficult, but it's… in
practice you won't normally see such a symmetric slabs, especially in the integrated circuits,
optical integrated circuits. What you actually see is another substrate or another material
with a refractive index N3, which again is less than N1, but N3 can be greater than N2
or it can be less than N2, okay. So depending on this, the mode structures will vary slightly,
but again this will introduce an additional layer of pro… you know, complication to
our understanding of simple wave guides. So I am going to give you only the principle
behind the simplest of these wave guides called as the symmetric slab wave guides, okay. So
in… in our module, we will assume that outside medium is characterized by N2 itself and you
can think of a slab surrounded by these two, by a medium, or this slab actually being carved
out of in a medium, whose refractive index is N2 and N2 is less than N1. Now with Maxwell's
equations, we need to first determine what kind of coordinate systems we are going to
deal with. So we are going to use rectangular coordinate system that seems very natural
in this case. So in that case we take this, anyway we have already taken this to be the
Z axis. What we will assume is that this axis over which these two, you know the slab boundaries
are defined are taken to be at X = +A, and X= -A. So we center the coordinate system
at this point, clearly Y will also be the coordinate, but that Y will be coming out
of this powerticular board, so you have X Y and Z as the coordinate system. Now immediately
we will have to deal with two things, right. It is clear that even if you go to the ray
picture, it is clear that what we are dealing with is an obliquely incident wave, right.
So if we are dealing with obliquely incident, we have choices for transverse electric polarization
or transverse magnetic polarization, meaning that light of course can be polarized in anyway,
but you can split that polarization into transverse electric and transverse magnetic, analyze
them sepowerately and then put the results back together. So this polarization decoupling
can be done because of the properties that any general polarized light can be split into
two orthogonal polarizations of which TE and TM are and then analyze them individually
and then put the results back together. So we are going to make a choice between TE or
TM and to keep the mathematics simple, but still essential in showing you all the steps,
I will assume that I am dealing with TE polarization, okay. Note that at this time we have not said
anything more about what would be the electric field and what would be the magnetic field
for these polarized light and how would they behave in this region, the slab region or
the region outside and for later consistency with the fiber, we will call this region as
core and we will call this as cladding, okay. Although this is not really core and cladding
in the fiber sense, but because our next model is going to deal with fiber propagation that
is propagation of light in fibers, we will keep the terms similar to that and we call
that densely, dense refractive index region as core and the light refractive index region
as the cladding, which surrounds the core, okay. So accordingly the core width in our
case is 2 times A where A is the half core width, okay. So to summarize we are going
to deal with transverse electric polarized light. So far we have not said anything about
the electric field and the magnetic field form or the solutions of the electric and
magnetic fields in this scenario and we have assumed a symmetric slab wave guide. Now for
transverse electric polarization, if you take this as a K vector that I have already drawn,
then the magnetic field should be present, right. So the magnetic field if I take would
be going along this powerticular direction, don’t worry what the actual form of magnetic
field is, we are going to derive this as the module progresses. However, for this case
the electric field has to lie in, you know, it is to come out of this page, right it has
to lie along Y. So electric field is polarized along Y and then the magnetic field will have
two components, one is HX and the other one is HZ. So if you look at HZ component, that
HZ component would go something like this, it would be along the Z component and then
HX component would be in this manner, right. Clearly on to this interface that we have
consider, X=A, and X= -A, at these two boundaries HZ will be tangential to the boundary or tangential
to the interface at X=A or X= -A, whereas HX will be the normal component. And electric
field because it is already in the Y direction and Y direction is powerallel to this X=A
plane or X= -A plane, it is already tangential in this case. So when you apply boundary conditions
at the interface X=+A and at X=-A, you are going to apply boundary conditions on the
tangential components, namely you are going to apply boundary conditions on EY and HZ,
okay so we are going to see that one later on, but we will start with a simplest case.
We will still assume uniform plane waves, because well you know, in this case we can
still assume it to be uniform plane waves or rather we will assume it to be plane waves,
they are not really uniform in this coordinate system, because there is a phase e power – j
K0 X, which would be different, right Anyway, so disregarding that short remark, let us
write down what is the wave equation and we already know the wave equation for plane waves,
right, and that would be del square E + K0 square N square E=0. Clearly I am dealing
with only a single polarized case that is E is basically EY along the Y direction, therefore
I can drop this vector and then make this into a scalar equation, okay. So I am going
to get a scalar equation with EY. Of course you will recognize this as Helmholtz equation,
okay, which was actually possible, I mean for us it was possible to write this equation,
because we assumed waves to be plane wave kind of a thing, right. So with the wave vector
K given by K0xN. N is of course refractive index. Now here we have to think a little
bit. Does this N refer to N1 or N2? In fact you have to have two such equations, one for
core and the other one for cladding, okay so you will have another equation for the
cladding. So you will have K0 square N2 square EY, okay. And this equation has to be solved
sepowerately and this equation has to be solved sepowerately and you meet both equations or
match both equations at the boundary by applying appropriate boundary conditions, okay.
What would be the solution for the equation in N1, sorry in the core region. Well first
look at del square in rectangular coordinate system is del square by del X square + del
square by del Y square + del square by del Z square. Now we are dealing with plane waves.
We want the waves to propagate along Z direction. So I will assume that whether I am dealing
with core or cladding, it does not matter, the electric and magnetic fields will always
have an E to the power – J beta Z type of a propagation, right. So we have all, you
know, we are propagating along the +Z direction and we know this very well from our earlier
discussion that a plane wave propagating along Z will be described in this manner E Power
– J beta Z type of a solution. What it means is that this del square by del Z square can
be replaced by – beta square, okay. Because our structure actually is, you know, taken
to be kind of infinite in the X and Y plane and none of the components are actually dependent
on Y, where you can make this Del square by Del Y square term go to 0, okay so I am going
to make this Del square by Del Y square term go to 0 and what I am now left with this Del
square operator is Del square by Del X square – beta square, okay. I can put this operator
on to EY, so what do I get, and because now EY is going to be functioned only of X, okay
of course it will also be function of Z, but that Z function is already taken to be in
the form of E Power – J beta Z, okay. So I can replace all these partial derivatives
with the knowledge that along Z is this behavior, however, along X we still don’t know what
is the behavior, so we are going to find that out, right. So I can write down the equation
which says D square EY/DX square + K0 square. For the moment I am going to consider the
solutions in core region, so I can write this as K0 square N1 square – beta square EY
= 0, okay. This is your ordinary second order differential equation, nothing fancy about
this, the solutions of course can be in the form of an exponential signals, right, or
exponential functions. So if you go back to your high school, or rather your differential
equations' course, you will see this that you can write this as –S square EY or you
can, you can write this is as – S square EY, where S square = K0 square and 1 square
– beta square, okay. And the solutions for this one will be E to the power + or – JSX,
right or you have solutions in the form of cos SX or sin SX, right. If S square were
to be negative, that is when beta square is actually greater than K0 square N1 square
than this would be a positive quantity, which we will call as some gamma square, okay, and
the solutions would have been E to the power + or – gamma x, okay. The nature of these
two solutions are very clear, one of them is an oscillatory solution. So a cosin X would
go in this manner, a sin X would go in this manner, a sin SX, where as E power + or – gamma
X one would be decaying, the other one would be increasing, right. Of course we don’t
have any gain, so normally we can rule these out in our, you know, solution set, but these
three seem to be the solutions that we are looking for, but exactly which solution needs
to be used depends on the sign of this term S, okay. In the core region, we don’t want
any decay along X, meaning that we want in the core a solution to be in the form of cosin
or a sin wave. That actually means that K0 square N1 square must be greater than beta
square and instead of calling this S square as K0 square N1 square – beta square, we
will go with the more conventional notation that is used in the optics, we will call this
as kappa F square or we will call this simply as kappa square. So we will call this as kappa
square and we demand that this kappa square be actually a positive quantity, okay, which
means that the solutions for EY in the core will be of the form A cosin kappa X + B sin
Kappa X, we don’t know what the A and B constants are, we can find this out or we
are going to make one more simplification, we will assume arbitrarily that B = 0. I am
ruling out this solution because only A cos kappa X is also a solution of this equation,
okay, with an unknown value of A. And if we retain only the cosin type of solutions with
respect to EY as a function of X, these are called as even mode solutions, obviously because
cosin is an even function and because I have ruled out this sin function, they are called
as odd mode solutions. However, I can, I mean, I can obtain odd mode solutions by ruling
out the even mode solution arbitrarily, okay. In practice you can have both modes and the
overall structure can be of a combination of these two, but we will soon see that these
two modes don’t exactly propagate at the same propagation constant and therefore they
can be discriminated against, okay. I am taking the even mode only for simplicity and I will
leave this as an exercise for you to repeat the entire calculation assuming that we are
dealing with a odd mode functions or an odd function, okay. So yes we have now shown that
the solution in the core would be of this form.
What would be the solution for cladding, what would be the solution that you would expect
for cladding? Well I know that if at all this is in the form of a sinusoidal function or
a cosin function, right, then outside we must have an evanescent wave, which of course decays
as you move away from the interface, right so that is what we have already seen. And
the same behavior should be expected even in the down, you know, in this side of the
cladding as well. So what we hope for the solutions in the cladding would be of the
forms C E power – gamma X –A, I will tell you what gamma is in a minute, and then you
have D E Power – gamma, E Power gamma + X + A. The reason why we have taken X – A
is just to simply the exponential function, because at X=A this will be just C and at
X=-A this will be equal to D. Of course from symmetric conditions, you would expect C to
be = D and that is actually true, okay. So the solution will decay exponentially on both
sides with the same decay coefficient and with the same starting points at the boundary.
Now what is gamma? Remember the solutions in the cladding are actually the solutions
of Helmholtz coil equation D square EY/DX square + K0 square N2 square – beta square,
okay. If you ask or if you demand that beta is greater than K0 square N2 square, okay,
then you see that, sorry, beta is greater than K0 N2 not N2 square. When beta is greater
than K0 N2, clearly that you are going to, it is clear that you are going to get a decaying
solutions for EY, which is what you have actually gotten here, okay. I have shifted the coordinate
system on the exponential by making sort of E Power – gamma X, I have made it into E
Power – gamma X – A. If I don’t shift the solution then that extra term will also
be present in the solution, but that would be a constant, okay. So it doesn’t matter
if you just take E Power - gamma X, but to simplify the boundary conditions, I have just
moved the coordinate system for the exponential to be along or to be starting at X=+A or at
–A. Now we have the full set of solution. We can now go to the next step. What is my
next step? Next step graphically I have already shown. Basically what you are trying to do
is to find out what would be the values of A and C for a given kappa and gamma or you
are even adjusting kappa and gamma in such a way that the solution smoothly changes over
from this cosin wave in the core on to an exponentially decaying wave in the cladding,
okay. So that is what graphically we are doing, but mathematically what that would mean is
at X=A boundary, you have A cos kappa A, which would be equal to C, which is obtained by
making X= taking X=A in these equations. So what you have is A cos kappa A=C. At this
point I don’t know what A and C are, but that is fine, we will leave this expression
here. Now I won't get much by applying the boundary condition at X=-A instead what I
want is a boundary condition on HZ, right or I need form for Hz. But how do I find HZ?
Well I know Maxwell's equation, it tells me that Del cross E=-J omega Mu0 times H, correct.
And I know E is along EY, expanding this Del cross E and equating the terms on the left
hand and the right hand side, which again I will leave as an exercise, because we don’t
have much of a time, you can show that HZ will be equal to -1/J omega Mu0, D EY/DX,
okay. So you can show that this is what you are going to get. And once you know what is,
I mean, once you know the expression for HZ, and you know the expression for EY in the
core and cladding separately, so you can differentiate it. So for the magnetic field component HZ,
you will get a solution, which would be kappa by J omega Mu0 A sin Kappa X and on the cladding
side you get gamma C by J omega Mu0 times E to the power – gamma X-E, okay. So these
are the solutions. Here I will leave it as an exercise for you to show that the solution
will be –gamma C by J omega Mu0 E to the power gamma X+A. Please note that this region
is basically X greater than –A, okay. And this region is basically X greater than A
and of course this region is where –A is less than X less than A. Now you can apply
boundary condition at X=A for the tangential magnetic field, that is HZ component and once
you apply the boundary condition for HZ component what you get is kappa A sin kappa A from this
expression, should be equal to gamma times C, okay. Because at X=A this exponential will
be 1 and J omega Mu0 on both sides will actually cancel off, so you will get this kappa A sin
kappa A = gamma C. We already know that C is given by A cos kappa A, so I can replace
in this expression C/A, so I have gamma A cos kappa A, indeed arranging the terms you
get tan kappa A to be = , so A will cancel on both sides, sin by cosin is tan kappa A,
you get gamma by kappa, okay. So this is an interesting expression, unfortunately it does
not have straight forward solutions.
The solutions that you are going to get are all what is called as numerical solutions,
because you have a function of the form tan X=X and this solution is rather you know easily
found using graphical methods or numerical methods. Graphically what this is telling
you is that you have X in this manner and then you have say tan kappa X, so you have
this expression and then again this one, these are the solutions that you are going to get
for tan, right. So if you plot this as a function of X, these are the solutions and all the
intersections that you are going to get will be the solutions. And of course what you have
observed is that there are multiple solutions for this tan X = X kind of a solution, right.
In our case what we have is tan kappa A = gamma by kappa, okay. If I multiply on both sides
by A and then define kappa A as equal to U and gamma A as W okay then I can even rewrite
this equation as tan U=W/U, okay. And what is U? U is basically kappa times A and what
is kappa is square root of K0 square N1 square – beta square times A. And now W of course
is given by square root of beta square – K0 square N2 square times A. And what you observe
here is that this equation that we have written tan U=W/U implicitly contains beta. So the
solution method that we normally adopt is to express this W in terms of U and you can
actually do that by writing U square + W square as what is called as V square. What is this
U square + W square, that is basically K0 square N1 square – N2 square times A square,
which we will define as parameter V and V is an important parameter that is determined
only by the wave guide geometry and materials. In this case it is only determined by the
wave guide, material property, and the operation wave length lambda, right. Wave guide has
a radius A, which is known lambda is the operating wave length, which is also N1 and N2 of the
wave guide are also known or usually can be designed knowing certain other properties
of this V. But the important point is that the solution that you obtain will be of two
steps, okay. After you express W as V square – U square under root. So on the left hand
side you have a function of U, right hand side also you have a function only of U, and
you can plot the solution as a function of U, or rather plot the left hand side and the
right hand side as a function of U, okay. So on the left hand side it's just a tan function,
it will have different values in this manner, so this will be 0, this will be say, so this
is a pi by 2, this is at pi and so on it is a periodic solution, right. And on the left
hand side what will happen at, when VU… U is very small, when U goes to 0, this numerator
is finite, denominator is infinite, so you will start somewhere of the infinity value
and as you start to increase, it will go to 0 at U=V. So if you know what is the value
of V, as determined by let's say your properties of the material or properties of the wave
guide, you would actually get a value of V here, which is determined only by the wave
guide and now your solution has to be until this point, right. So you start off with infinity
and then you go to the solution until at this point, where it kind of decays off. How many
intersections do you see? You see two intersections one intersection here that corresponds to
the fundamental mode that would be propagating and this next one is called as the higher
order mode that would be propagating. So in fact you get multiple modes which are propagating.
The first intersection of this one will be denoted as TE1 mode, the next one will be
denoted as TE2 mode and so and these are all the even modes that we have considered, okay.
In the odd mode scenario things are slightly different, there the relation on the right
hand side or the left hand side actually changes slightly and without going too much into details,
you can plot the odd mode cases, and what you observe is an intersection, which would
be present in this manner as well until this value of V here and what you notice is that
you will first get an even mode, then you will get an odd mode, then you will get an
even mode. Odd modes are basically sin functions, okay and the higher order modes, that cosin
functions are the even order functions.
So if you were to look at the fundamental mode assuming that this your boundary for
the core end cladding, the odd mode would look, sorry even mode would look something
like this. This is a cosin wave and then it's exponentially decaying everywhere, and the
even mode would look something like this, okay. So these are the, even I have not drawn
the pictures nicely, but this is how the even and odd modes would look like and these are
the fundamental modes. The higher order modes will be, you know will have more maxima in
between. So this completes our basic understanding of wave guide modes. So to… take… the
take away message from this module is that mode is essentially solution of wave equation
or equivalently Maxwell's equation, meaning…solution meaning the E and H patterns that I am actually
interested in or the E and H solution that I am interested in and this should not only
be the solution of Maxwell's equations, but they should also satisfy boundary conditions,
okay. In the free space case, these modes were essentially plane waves, they were actually
in the form of E Power – J beta Z for a Z propagating wave or E Power plus J beta
Z for a Z propagating wave or E power + J beta Z for a Z propagating waves, okay. There
were no boundaries. However, the modes you can think of, the free space modes were the
plane waves, okay. There were no boundaries, however, the modes you can think of, the free
space modes were the plane waves.
However, the mode structure in the case of a guided thing is actually very different.
So in the guided media, the mode structure depends on the geometry. For the case of a
planar wave guide that we have considered or what is called as slab wave guide, the
mode structure is either cosin or sin with many-many periods within the core and decaying
exponentially in the cladding, of course propagating along the wave guide itself, right. So what
you have to understand is that this wave guide mode or mode in general is a property of geometry.
When there is no geometry, when there is no constraints, we get a plane wave, but when
we start putting constraints, the solution changes over, and depending on the type of
the constraint that you put in, that is, if… if you send in parallel plates, the solutions
inside will be given by sinusoidal functions, it will be cosin or sinusoidal. However, if
you instead of shrinking this, I mean, putting this one in this manner, if you make cylinders,
the functions will still look like sin and cosin, but they will be mathematically actually
bessel functions, okay. So the type of geometry determines the form of the function or form
of the electric and magnetic fields that can propagate within the given wave guide and
the solution of Maxwell's equation, which also satisfies boundary conditions is called
as a mode and we have just successfully solves for modes of a symmetric slab wave guide.
Thank you very much.
