noc18-ee28-Lecture 17-Further discussion on slab waveguides

IIT Kanpur July 201
20 followers
1 Video View·Dec 3, 2023
Hello and welcome to NPTEL MOOC on fiber optic communication systems and techniques.
In the previous module, we discussed a systematic procedure to derive properties of modes of
a slab waveguide.
We first applied that systematic procedure for a metallic waveguide and since metallic
waveguides are not used at optical frequencies for the reasons that we mentioned, we then
discussed slab waveguides and then we showed that the acceptable modes of a slab waveguide
can be partitioned into transverse electric and transverse magnetic mode.
Of course, in a real world scenario when you launch a mode, all these modes can be excited
depending on how you launch a particular mode.
Some of the mode launching procedures for fiber optic, optical fibers we are going to
consider later on.
We will not consider such a mode launching discussion for the slab waveguide anyway.
We saw that when we solve the equations, so we obtained a certain equation which related
the transverse wave number kappa f and the decay constant outside the slab waveguide
gamma into a transcendental equation.
So this relationship as we told you was what is called as the eigenvalue equation because
hidden in this kappa f and gamma is the propagation constant beta.
Because kappa f and beta are related to each other, gamma and beta themselves are related
to each other, by solving this characteristic equation, we can obtain what is the corresponding
propagation constant of that particular mode which is propagating in a given waveguide.
The systematic procedure also had this advantage that in addition to just knowing the value
of beta, which anyway you could have found from the transverse resonance condition, you
now also have the capacity to actually look at how the modes themselves look like or essentially
what is the functional dependence of the modes of the different components, electric field
and magnetic field corresponding to whatever components that go for the transverse electric
and transverse magnetic modes.
We did all this analysis only for transverse electric modes.
For transverse magnetic modes, the steps are very similar, except that for the transverse
magnetic mode, you start with the Z component of the electric field and then derive the
other components in terms of that Z component, solve the wave equation for EZ, obtain a characteristic
equation, find the propagation constant and from the relationship of EZ with other components,
you will then be able to calculate the dependence of the modes on the coordinate x.
So, what I want to do now is to kind of reconsider that characteristic equation, tell you a couple
of properties of that one and then show you a couple of methods to solve that transcendental
equation.
One method would be what is called as the numerical method.
Numerical method is fastest, gives you accurate answers much more than what you would get,
but for many problems or at least until the advent of computers, people used what is called
as the graphical approach.
For most cases, graphical approach does yield very close answers to numerical and it is
far, far simpler to use the graphical approach.
And of course, it also provides you in addition to just the numbers, it also provides you
the intuition as to which mode starts where and which mode ends or what is the dependence
of a given mode with respect to the frequency or wavelength, which is very important when
you are analyzing the pulse propagation through this slab waveguides.
We start from something that we already know.
What I am going to do now is to just recapitulate those equations which you already have seen
earlier.
So, for the transverse electric even modes, so this is for the transverse electric even
modes.
When we say even modes, what we actually mean by that is that if you were to look at the
x axis, the slab waveguide of course is propagate, the modes are propagating along the z axis,
but it is bounded by two planes.
So, inside this plane, you have n1, outside this plane you have n2 and this plane will
be one plane kept at minus h by 2, the other plane is kept at plus h by 2 as we have seen.
So, you can see that I have kind of only being considered, considering the x component in
this one, because along the z component, you know that whatever the basic mode, it would
simply propagate along the z axis.
So, when we say even modes, what we mean is that from the center of the axis, which of
course will be from the center of the axis, which will be at x equal to 0, if you were
to plot the modes, you would have modes in the form of cos kappa f h and we have already
seen that this kappa f has to satisfy a certain relationship in order to, so we need to solve
this equation to find out kappa f, but once you find kappa f, then it is possible for
you to sketch these modes.
This is not kappa f h, this is kappa f x inside the film and outside the film or that is outside
the slab, you have decaying modes.
So, you have either e power plus gamma x, which would of course decay outside or in
the cover region, because x is positive in that region, you will have e power minus gamma
x for that one and you have e power plus gamma x for x less than minus h by 2, which essentially
is the substrate and please remember, all our analysis has been done for what is called
as symmetric waveguide.
Symmetric waveguide means N2 on both sides is the same, we will mention couple of points
about asymmetric modes later on, but for the symmetric modes that we have considered, we
have done a fairly systematic analysis to obtain the equations, the characteristic equation.
So, as I was saying, what is an even mode for us?
An even mode is one which has a cosine dependence, because as the mode changes, it would be cos
kappa f x, if we did not have these planes at plus h by 2 and minus h by 2, then this
mode would actually have gone and become a cosine like this, but unfortunately, that
is how that is not possible, because you have these planes, which will stop your cosine
function just at this boundary, at this point, it would have stopped.
Now, we have seen that when we apply the boundary conditions, what essentially we are doing
is to match the fields, which would be decaying outside the film and at the point where at
x equal to h by 2, where the film meets the cover or at minus h by 2, where the film meets
the substrate, these decaying, exponentially decaying solutions must smoothly merge with
the solution in the film.
So, you can see that there is a single maxima here and this essentially is what is called
as the fundamental mode of the, fundamental TE mode of the symmetric slab waveguide.
So, all that we did, the complicated math that went into the systematic procedure was
to essentially get this point matched.
So, this was the matching condition that we had to impose by all that mathematics, but
what visually was happening was, you had a field in the film and then that would smoothly
change to a decaying field in the cover or in the substrate.
So, higher order TE modes are equally easy to sketch or probably difficult to sketch
depending on how good you are with your cosine functions.
So, let me try and sketch that one.
So, this was half cycle.
So, the other solution that might be possible is this one.
So, in this case, I have one half cycle, another half cycle.
So, I actually have almost complete cycle here, but then the field must decay outside.
So, it is not really going over nicely.
So, please excuse that one, but the field outside would still be decaying.
Of course, the decay rate depends on what is the value of gamma.
So, in this way, you can actually build up higher order modes.
So, this could be called as TE0 mode, this could be called as TE2 mode, in that way,
you can actually keep building up your modes.
So, this is something that we have seen in the previous class, but the physical picture
you might not have seen.
So, what would we call as the odd modes?
The odd modes would essentially be like a sine kappa fx.
So, they would actually go like this.
So, you have the odd, this one and then outside, it would decay in the substrate as well as
in the cover region.
So, this would be the TE odd modes.
I am not nice in sketching all these things, but it is essentially cosine and sine functions,
outside it is the exponentially decaying functions.
So, if you can use MATLAB or any other software that you are comfortable with, you can actually
keep sketching all these modes, except of course that you need to know what is the value
of gamma and you need to know what is the value of kappa f.
Of course, this can be found by finding beta, but beta is actually hidden in this equation.
What was the equation?
Let us complete this.
Tan kappa f h by 2 was given by gamma divided by kappa f.
This is for the TE even modes and for the odd modes, the left-hand side would essentially
remain the same, but the right-hand side becomes minus kappa f by gamma for the TE odd modes.
In both cases, the solutions are not simple because you have a transcendental equation.
For the TM modes, the left-hand side essentially is the same.
So, you have tan kappa f h by 2 being equal to n1 square by n2 square gamma by kappa f
and this would be for the TM even modes and minus n1 square by n2 square kappa f by gamma
for the TM odd modes.
Now, you can actually see that the right-hand side is essentially same as the case for the
TE even, but these values are slightly larger corresponding to the TE case because n1 is
usually larger than n2.
Of course, you need that scenario because you want waves to be guided in the film region.
So, for n1 being greater than n2, n1 square by n2 square will be greater than 1 and therefore,
this term that you have here, so n1 square by n2 square gamma times kappa f is slightly
larger than gamma by kappa f of the TE modes.
So, now I have not derived the TM even and TM odd mode characteristic equation.
I will leave that for your consideration and now what I want to do is to consider one of
these equations because the procedure for the other equations would essentially be the
same.
So, I want to consider one of these equations which I will consider for the transverse electric
even mode and I want to solve this equation.
So, how do I solve this equation?
My goal to solve this equation would be to find out beta, but I am not going to do this
in a straightforward manner.
That is, I am not going to find out the value of beta, but I will first find out the value
of kappa f which satisfies these two equations.
I express everything in terms of kappa f and then I will be able to compute beta because
the relationship between beta and kappa is kappa f square plus beta square equals k0
square n1 square.
So, now my job of course would be to express gamma in terms of kappa that is rather very
easily done because gamma square at least is given by beta square minus k0 square n2
square.
Of course, beta square is nothing but k0 square n1 square minus kappa f square, kappa
f will be treated as the variable in these equations.
So, minus k0 square n2 square as it is, so you can group the terms and then essentially
obtain gamma in terms of kappa f.
So, now you can substitute for gamma in this expression, in this expression you substitute
for gamma, kappa f anyway is kappa f, on the left hand side you have kappa f h by 2,
just to consider the solutions you might want to scale up the right hand side as well by
multiplying by h by 2 on both sides.
So, on the right hand side you have gamma kappa f h by 2 divided by kappa f h by 2.
This is just to you know kind of replace all of those variables by kappa f h by 2, that
is all I am trying to do here.
You can even do that by writing this h by 2 into the square root operation for gamma,
because gamma is as we have seen from the above equation will be k0 square n1 square
minus n2 square minus kappa f square.
So, when you take h by 2 inside the variable out there or inside the square root, this
becomes h by 2 square, this will be h by 2 square.
And then denominator is essentially kappa f h by 2, so what you have is a solution in
which I can treat kappa f h by 2 as the independent variable, because all the equation out there,
I mean the equation that we have written essentially depends on kappa f h by 2.
So, what is the method of solution?
Well, you can do a numerical solution.
So, how do I numerically solve this equation?
First I will figure out what is the minimum and maximum value of kappa f, what would be
the minimum value of kappa f?
Well, the minimum value of kappa f will be when, remember this is a triangle, so there
is a beta here, sorry that is the transverse thing, so this is kappa f, this is beta and
together is the incident wave vector or the wave vector inside the film which is kappa
0 times n1.
So, you can actually have the length of kappa f become 0, at which point k0 n1 will be exactly
equal to the propagation constant beta.
So, if you start shrinking this height, then k0 n1 actually falls on to beta 1, the projection
of k0 n1 will then be exactly equal to beta, when kappa f which is the height of this triangle
will be equal to 0, so that is the minimum value that one can actually have.
So, the minimum value of kappa f is actually equal to 0.
What about the maximum value of kappa f?
Well, we know that beta has to be there within say k0 n2, I mean it has to be greater than
or at most equal to k0 n2, when this happens, then you are essentially into cut off of that
particular mode.
Therefore, the minimum value of beta will be k0 n2, at which point kappa f max will
actually be equal to k0 square n1 square minus k0 square n2 square under root.
You can rewrite this one by taking k0 out and then you have square root of n1 square
minus n2 square.
So, we have determined what is the minimum value of kappa f, we have determined what
is the maximum value of kappa f and within this region or within this range is what we
are actually looking to plot this equation, so or numerically solve this equation.
So, all that you have to do is you find out given n1, n2, lambda, you can calculate from
lambda k0 value because k0 is basically 2 pi by lambda and lambda will be usually specified
in the free space region.
So, given these parameters n1, n2, lambda, I think these are the parameters that you
need.
Oh, you also need the height of the waveguide, let us say that is h, then what you have to
do is to first determine the minimum and maximum values of kappa f.
Once you have determined that, you write a code.
So, let us say you have kappa f being some sort of an array, which you take some kind
of a linear spaced array from 0 to kappa f max, I am using 0 because that is a minimum
value and you create some 100 points or maybe 1000 points depending on what points you want
to create.
So, all that this has done is that it has taken this range from 0 to kappa f max, this
is kappa f max, what you want to do is to actually multiply this one by h by 2 because
that is your independent variable.
So, you have this range of 0 to h by 2 kappa f max and in this range, you basically create
some 100 points.
So, this can be done numerically and then numerically obtain LHS, which is basically
tan of kappa f h by 2.
So, you can of course, give a certain variable say z equals kappa f h by 2.
In that case, you are simply calling this routine in whatever the language that you
have as tan z.
Similarly, what would be the right hand side?
The right hand side is essentially what we have seen, square root of, so this is the
right hand side here and kappa f h by 2 is what we have called as z, therefore denominator
is z here, numerator is z by 2 whole square.
Now, if you directly put z equal to 0, numerically you will be having a problem, therefore, you
do not normally start your vector at 0, but you actually start your vector at a small
value, let us call that small value as epsilon, so epsilon, epsilon may be at this particular
point, it could be just a nonzero value, it would not be very large value because you
will then be moving away from the minimum value, but it will not be also so small that
you will end up with numerical errors.
So, you choose your value depending on, you actually look at what you are getting from
your MATLAB code or whatever the code that you are using to find out the appropriate
value of the small nonzero value, epsilon s.
And once you have found that one, you can write down all whatever that you had, for
example, I mean, write down the right hand side thing, so k0 square, you had I think
n1 square minus n2 square minus z square, then there was h by 2 square divided by z,
so you find out, and z is the array that contains these values, so these are the different values
of kappa f h by 2, which is your independent variable, so you create the left hand side
array, the right hand side array.
And then what you can do is to find the points where LHS is equal to RHS, now equals is a
very interesting condition, therefore what you normally do is, you subtract LHS from
RHS, and then say if LHS and RHS, the difference between these two values is less than some
predefined factor, so I will call this as P factor, if P factor, then note corresponding
kappa f h by 2, that would be the value of kappa f h by 2, from which you can then calculate
beta, you can calculate gamma.
What is the P factor?
Let us say, you are looking at, in numerical solutions, you are looking at the difference
between two vectors x minus y, and you want the error to be in the order of say something
like 1 or 10 to the power minus 6, so this is of course just my condition, this P factor
could be any value that is dependent on the application that you have, but in general
looking at the range of kappa f, an error difference of about 10 to the power minus
6 or 10 to the power minus 3, in whatever that range would be sufficient.
So once you have figured out that P factor, then you can stop the program, and note down
the corresponding value of kappa f h by 2, and that would give you beta versus gamma.
So this is a numerical way of solving for the characteristic equation, that will give
you the number, without any doubt it will give you the number, but it does not give
you the, you know what is that, the physical intuition as to what is happening, to get
the physical intuition, we go back to the equation here, again I am going to write down
everything in terms of kappa f, but please remember that you know what is gamma times
h by 2, so it is a lengthy expression, so I do not want to write that lengthy expression,
but anyway I have written that one, that is why I have written down this as gamma h by
2. So substitute for gamma here, this is gamma,
so you can substitute it here, and then get that lengthy expression that you have. Now
the graphical approach is actually very interesting, what it does is that you plot the values of
you know the left hand side and the right hand side, so I define z as kappa f h by 2,
I have defined this one, and then I am now going to plot this tan kappa f h by 2, so
how is the function be a function of z like this, where z is kappa f h by 2, well at 0,
at z equal to 0, tan function will be 0, but then it increases steadily, and of course
goes off to infinity at pi by 2. So when kappa f h by 2 equals or nears pi by 2, the corresponding
value of tan function has actually gone up to infinity, it kind of asymptotically goes
up to infinity. Then of course as you increase this value
of kappa f h by 2 further, it starts at pi by 2, and then goes through 0 at pi, and then
goes again back to infinity at 3 pi by 2, this is pi, so this is pi, and then you can
keep moving. So how far do I need to move this one, well I know what is kappa f max,
correct, I know this kappa f max beforehand, so all I have to do is to take the argument
of z from 0 to kappa f max times h by 2. So let us say kappa f max h by 2 happens to
be at this point, I am just basically taking some random numbers, but in the exercises
you will actually be getting the correct values, so you can go back and put the exact value
of kappa f max h by 2 here, and then complete this problem, or complete the understanding
of this one. So let us say this kappa f max h by 2 happen
to be here, so all this region beyond this is really not important for us, so you can
scratch this out, or you can remove that one. Now let us look at the right hand side,
what is the right hand side look like, when kappa f h by 2 or z is 0, the right hand side
actually shoots up, because numerator will be finite, but the denominator will be 0,
so the right hand side shoots up, so as you move from kappa f h by 2, you will actually
see that it starts from infinity, and then starts to go to 0 like this.
Of course the rate at which this falls depends on what values of n1, n2 and lambda that you
have taken, but no matter what it is, it starts here at near 0, and then as it now goes down
to 0, it is clear that the right hand side and the left hand side have to cross at a
particular point, and this crossing point is the solution, the corresponding value of
kappa f h by 2, from which you can easily find out what is the propagation constant
beta. So this is what we would call as the fundamental mode, and this mode happens to
be the Te0 mode. What would be the situation for the TmE1 modes? Remember for the TmE1
modes, the right hand side is basically n1 square by n2 square, and then the corresponding
gamma h by 2 by kappa h by 2 is essentially the same thing.
So all you have to do is to now take the right hand side, and then draw another curve, which
would be slightly away from this one, so it would actually, you know, depending on
how far n1 and n2 are, but it would actually be slightly away like this, and it would also
go down to 0. It starts at infinity and goes down to 0, and you can see that the mode is
just around the Te mode, so this would be the Tm mode, and this would be the Te mode.
These are the even modes. How about the odd modes? Well for the odd modes, you will have
to start with minus kappa f h by 2 by gamma h by 2, correct? So because there is a minus,
the odd modes actually occur in the fourth quadrant, and because the numerator is kappa
f h by 2, they would actually go something like this, right? So it would not touch the
first branch here, do not worry about the points being touched here, that is not of
consequence. So the first intersection happens in the region when kappa f h by 2 is greater
than pi by 2. So this would be the first Te odd mode.
Similarly, for the Tm even mode, you will have to start somewhere here and probably
meet or maybe you have to start somewhere here and then probably meet here, so this
corresponds to the Tm odd mode. Again, please note how closely this Te1 and Tme1 are situated
depends on the index values of n1 and n2. In fiber optics, n1 or in optical fibers,
n1 is very close to n2, therefore their ratio is almost 1, which means Te and Tm modes are
kind of degenerate, that is they have the same propagation constant, whereas in an integrated
waveguide of which slab waveguide is an example, n1 and n2 contrast is usually larger, so these
two modes are not the same, but they are slightly different. So they actually are two different
modes usually in the case of an integrated optical circuit.
So we have seen how to go about solving this characteristic equation for in the graphical
method or in the numerical method. There is one additional method, which is again a graphical
method, which in which we instead of considering kappa f and the corresponding values of beta,
what we do is we normalize these values by defining certain values called as b, which
is essentially defined as n effective square minus n2 square divided by n1 square minus
n2 square, and then we define v as k0h square root of n1 square minus n2 square, sorry b
is basically square root of this one, I forgot the square root, and then from the transverse
resonance condition, you express the transverse resonance condition in terms of v and b now.
So the transverse resonance condition, if you remember was 2 kappa f h minus 4 phi Te
for the Te modes being equal to 2 nu h, where nu was the integer as we have seen earlier.
And phi Te was essentially tan inverse of gamma by kappa f. So you can actually express
with the help of this v and b, you can transform the TRC equation, which would have dependent
on gamma and kappa f into an equation which is dependent only on v and b, and the universal
graphs that you can actually plot means that you do not need to worry about which waveguide
that you are actually calculating, you can calculate this normalized way, and then apply
it for your waveguide, you can actually scale these values.
So this b versus v curve, which we will see for fiber, optical fibers as well, would look
something like this. So you would have a Te modes, and then you have the next group of
modes like this, and then the next group of modes. So given the parameters, you can actually
first find out what is the corresponding v number for your waveguide, and from there
find out what are the different propagation constant for different modes. We will discuss
this approach when we talk about optical fibers in more detail.
Thank you very much.
