Selective Mode Coupling in Microring Resonators
Sep 5, 2023
"Selective Mode Coupling in Microring Resonators"
Amir Arbabi, Electrical and Computer Engineering
CNST Nanotechnology Workshop 2013
Good morning, ladies and gentlemen.
Thank you for inviting me to share some of our research concerning chlorine resonators
today with you.
So today, I will talk about selective multicoupling in micron resonators.
I will explain what that exactly means.
My name is Amir Abubi, and I work with Professor Linford Goddard at Photonics Systems Lab.
So here's the outline of my talk.
I will start by motivating you about what we are doing and why it is important.
Then we'll give an introduction about micron resonators for those of you who are not familiar
with these devices.
In the next section, I will move to explaining simulation design tools that we developed
for simulation and also designing these devices.
Specifically, I will talk about how we can find the resonance modes of these devices,
also how we can use a coupled mode theory formulation to calculate the mode coupling
in a ring resonator.
Then we'll talk about the fabrication of these devices and measurements.
And at the end, I will summarize my talk.
So to motivate you about this research, our main work is on changing to make better laser
diodes.
So laser diodes or semiconductor laser diodes are structures that have an active section,
and usually they have two mirrors.
So similar to any laser structure, you have an active section or a gain medium, and have
two mirrors that create the feedback.
For single-mode laser diodes, you usually have a selective mirror.
In this case, we have a gain medium, which is the active part, and a reflective mirror,
which has a selective reflectivity.
So as a function of wavelength, the reflectivity of this distributed Bragg reflector, which
is just a periodic structure, but changing the index of the material, can make that.
So the reflectivity of this structure looks something like this.
So at some design wavelengths, you have a large reflectivity.
And then there are these side lobes, which cause the side modes in the laser at the end.
And so the laser wavelengths will, the laser will place somewhere around this design wavelengths,
and then we will get some side modes because of those small reflectivity peaks around the
actual design wavelengths.
So such a structure usually has a large footprint.
To get a narrow enough full width half max for this reflection peak, we usually have
to write this for a few millimeters, even a few centimeters.
So it's a nanostructure with a period of about 150 nanometers, but you have to write that
for a few centimeters.
The uniformity usually becomes an issue because you have to keep the period fixed, otherwise
this resonance, this reflection peak becomes wider.
The other problem that I mentioned also, or the noise in the laser output that you will
get because of these side lobes in the reflection, they will cause side modes in the laser.
So what we want to do is we want to replace that reflector, which is a distributed Bragg
reflector, with a smaller device, is a resonance device.
So we have a ring resonator with some grating written on the ring waveguide, and we expect
to get, from this structure, we expect to get such a reflection peak.
And this structure works based on coupling the degenerate modes of a resonator.
And as you can see, you can get narrow reflection peaks.
There's no side modes in this.
So because the whole structure is pretty small, it's in the order of less than 100
micron in size.
And also, when you want to tune the laser, usually you tune the wavelengths of this reflector.
So when the reflector itself is small, we would need smaller powers, or its power to
tune the structure.
And as I mentioned, there's no side modes in this structure.
The VR thing is the full-width half-max of this reflection peak does not depend on the
length of the structure.
For the DVR case, if you want a narrower reflection peak, you have to increase the length of the
device, as I mentioned, to one centimeter.
So here's an introduction to these traveling wave resonators.
That macro resonator is one type of those.
So in these traveling wave resonators, what you have is in the light end, and then the
light is being trapped in a circular path.
So in the ring resonator, you have a light trapped in the ring, so that light propagates
along the ring.
You can have these modes just in a macrosphere.
So if you make a macrosphere, because of the total internal reflection, you can have light
being trapped inside the sphere.
You have also a form of these resonators.
You can make a microdisc, and the light can be trapped along that.
Also, if you reflow this macro ring, you can get microdisc, you can get microtroids.
So there are different designs of these traveling wave resonators.
The main thing similar between these is that light is trapped in a circular path.
So we are mostly interested in macro ring resonator.
The main reason is that they can be fabricated on a chip.
So we can fabricate a lot of these devices very close to each other.
The fabrication can be controlled very well, because it's a top-down fabrication.
So a ring resonator, as I mentioned, you can set a light, and the light is coupled to this
resonator whenever this resonator has a resonance.
And the resonance happens when the path length for the light is a multiple of the wavelengths
of the light.
So a fabricated device looks something like that.
You have a ring, a bus waveguide, and the ring resonator is coupled to that.
Normally, when you measure the transmission of this device, so how much light is transmitted
through this device, we'll see dips in the transmission.
Where the ring goes to the resonance, the whole light gets coupled to this ring resonator
and gets lost inside that.
So we get zeros in the transmission.
The modes are uniformly spaced.
Another important thing is that because this structure has circular symmetry, at each resonance
wavelength there are two possible modes in this structure that can propagate at different
directions.
So you can have a clockwise mode or a counterclockwise mode.
They have the exact same frequency and ideal structure.
So the idea is to couple those two degenerate modes at selective wavelength.
So if we add a reflective element inside the ring, so the ideal ring doesn't have any reflective
element inside it, but there are two modes that can propagate in contra-propagating directions.
If you add a reflective element inside the ring, then we can get those two modes to couple
to each other.
And then from a ring, we can get a reflection.
So without this reflective element, we don't have any reflection.
The light goes in, rotates there.
But when you add this one, this mode that's coupled, and the input one will be coupled
to the other mode, and we can get a reflection from the ring.
So if you plot the reflectivity of this device, so the reflectivity is color-coded as a function
of wavelength, and also the reflectivity of this small reflective element inside the ring.
So the reflectivity goes from zero to one, and this is a function of wavelength normalized
to the free spectral range, which is the distance between two resonance wavelengths.
And this is the amplitude of the reflectivity of this device.
We can see at some, even at very small reflectivity amounts for these reflective elements inside
the ring, we can get almost a reflectivity of one from the overall device.
So the one is the red color there.
So you can have very small reflectivity inside the ring, couple those two modes together,
and get a very large reflectivity from the total device.
More interesting than that, we can have a device here that has a selective reflectivity.
So if we design a device here that has a reflectivity as a function of wavelength looking
like these dashed lines they've drawn here, then we can have reflectivity from the whole
device only happening at one wavelength, and have zeros at other wavelengths, at other
resonance wavelengths of the ring.
So we can get a single reflectivity from the ring.
So rings, as I mentioned, usually have equally spaced modes, but in this case we can select
one of the modes and then get reflectivity at that specific mode.
So for such a device with that selective spectrum, we can get a reflectivity of the total structure
to look something like that.
So we have a transmission, there's a main reflection peak, the blue lines, and at that
reflection peak we have zero transmission, and we can have a very large reflectivity.
So this is just a zoomed in view of that one, you can see that it looks very similar
to what I proposed in the motivation section that we can have a reflectivity peak without
any side modes.
Something more interesting is if you consider a, someone can mark a laser that has a mirror
which is wavelength dependent, so if you have a mirror with reflectivity which is wavelength
dependent or frequency dependent, then you can do a calculation and show that the line
widths of the whole laser will be reduced by a factor of fc squared, which fc depends
on the slope of the phase of the reflectivity, and also the slope of the change in the amplitude
of the reflectivity.
So if instead of having just a normal mirror, like a cleaved facet, if you have a reflective
element which is, its reflectivity is wavelength dependent, then we can get a reduction in
the line widths.
If you calculate that reduction factor for our device, we can see that depending on where
we set the wavelengths of the device, we can see a reduction in the line widths by a factor
of around 150.
So we also developed some design tools for designing these structures.
The structure is very large compared to wavelengths, it's very hard to simulate the whole structure
using full wave methods.
So what we did with, as I mentioned in the FAST method for doing that, what we did, we
developed a method that we can use to calculate resonance modes and use that resonance modes
to calculate the reflectivity of the whole device using a perturbational method, looking
at a couple more theories with it.
So the resonances can be found using 2D simulations because the device has cylindrical symmetry.
The phi dependence is known because of that cylindrical symmetry.
We can solve that two-dimensional problem and find the resonance wavelengths and the
quality factor of that.
Then we can use those modes that we found and expand the modes of a device with some
perturbations added to that in terms of those original modes that we found.
Just the only difference is that in this case the amplitudes are a function of phi.
And we can get a coupled mode formulation for that which shows how those two amplitudes
are getting coupled when there is some perturbation added to the structure.
Using that we can do our design using this method.
We used this method to simulate a 2D structure, just a toy problem, and we found that using
a finite element method, the simulation results using that, and we used also our method to
do that and compare the results.
The results matched very well between our simplified method and a full wave simulation
which is a discretization of the whole structure using a finite element method.
We used that design tools and fabricated these devices.
We used silicon substrate, boron oxide as a lower cladding, silicon nitride, LPC with
the silicon nitride as the core for these devices.
LPC with the silicon nitride is known to have a low absorption loss at telecom wavelength
and that's why we used this material.
So as I mentioned we have a reflective element which is selective.
In this case it's a DVR written on the ring.
So this is a fabricated device, when you zoom at that it looks like something like that.
So we wrote this using immunolithography, we etched it, and then we deposited the top
cladding, cleaved the devices, and deposited AR coating on the facets so that the reflectivity
from the facets doesn't interfere with the reflectivity of the device itself.
We can have also different designs and our design looking like that improves the, increases
the smallest size in the device and makes it possible to fabricate with photography.
So for measuring these devices we couple a tunable laser, we pass it through an optical
isolator and a polarizer, we sample the light that goes to the device and then we send the
remaining light to the device.
The transmitted light is coupled to the device using tapered fibers and we measure the transmitted
light that goes through the device and also the reflected light is measured using an isolator,
using a circulator.
So the light that goes in, the reflected light goes through this detector.
So we can measure the input, the transmitted light, and the reflected light at the same
time and then we scan the wavelengths and we measure the reflectivity of the device
and transmission of that at these single wavelengths.
So we get reflectivity spectrum and transmission spectrum looking something like that.
So this design was, the device was designed to have reflectivity at 1550 nanometer.
We see a single peak with a very large reflectivity.
There are some side peaks, much smaller side peaks.
That's because of the roughness in the file.
So the cycle roughness causes the modes at this wavelength to also couple to each other
but it's much smaller compared to the reflectivity at the design wavelengths.
So we measured that over 110 nanometer range which was set by our laser and over that whole
range we had only one single reflectivity.
If you zoom in at the main peak, the main peaks look something like that.
So for the reflection, it doesn't exactly look as what we expected, although the transmission
looks fine.
We figured out that this is due to the reflectivity that we get still from the facets and there's
interference between the reflectivity from the device and the facets.
We modeled that and so the solid line is the measured one, the model which considers also
the reflectivity from the facet is the red dashed one.
We did the fitting and we extracted the reflectivity of the device.
So if you plot the reflectivity of the device, in this case this is the reflectivity and
the transmission are these dashed red lines.
We can see that we get a single reflection peak, there's no side modes.
The blue line here shows a conventional design of a DBR which is for getting the same full
width half max, should be 4.3 millimeter long.
So in summary, our device is 70 times smaller, has the same reflectivity, and it doesn't
have these side modes, so it also has a faster draw off for the main peak.
So just to summarize, I introduced the device today, these reflective devices, they can
be useful for application in laser diodes.
I presented some of the simulation methods that we developed for designing these devices
and presented and showed some of the characterization that showed that these devices can really
be fabricated with the desired spectrum.
So thanks and I'll be happy to answer any questions.
Any questions?
I have a question.
Is there any advantage to coupling these micro-ring resonators together?
You talked about a single micro-ring.
If you put two or three, is there an advantage to that?
Yeah, there is an advantage and people use them for making filters because a single ring
resonator, as you saw, has periodic transmission depths.
If you couple two different ring resonators with two different free-spectral ranges,
you can get a single transmission depth.
But the main problem is controlling the gap between those two and the amount of coupling.
Have you studied that at all?
No, not yet.
I'm also curious, what sets the scale of the resonators, the diameter of the rings, for example?
Can you scale those down if you want to?
Right, you can scale them down, but there is a minimum radius that the light starts
to radiate out.
So the angle that the light gets reflected at should be larger than the total internal
radiation angle.
If it's smaller than that, it becomes one, you get radiation from the ring.
That one's been a long day so far.
What are the quality factors that you got from fabricating these micro-rings?
The quality factors were about, this one that I showed was 10 to 5, but we have increased
them right now to 10 to 6.
So we're improving the fabrication process.
Have you studied any thermal fluctuations in those devices?
Yes.
Are they sensitive to temperature or not?
They are.
So the resonance wavelength moves by around 17 picometers per degree C.
Also, on high Q devices, you will see thermal visibilities at higher powers, like in milliwatts
of powers.
And you will see thermal non-linearity.

