Hearing departure from the regular

rector of the Center for

your undergraduate university

from Cal Perry also Cal Tech

research materials

because we're going to number research

President Sarkozy engineers from

that afternoon to thank you.

Organizers for inviting you for this
presentation and today I will focus on

some of the work we have been doing in
the last few years in the development up.

So the cañada for tonic of structures for
the lab on chip sensing and

this is the list of people who have
contributed to this to this work.

So I got a quick
introduction to motivation.

Based devices are useful for
lab on chip applications and

then address some of the major challenges.

Is that we have in such
sensing modules and

then focused mainly on two technologies
we have developed to address those

challenges that are high Q silicon
micro this present a Taurus and

Altro very compact on chip or
spectrometers using for tonic crystals.

This cartoon shows the general
platform for a lab on chip sensing

module using optical tools light
from a basic guide interact with

the molecules to be sensed here
in this interaction region and

then the output of this region is
monitored by the onto the spectrometer and

the reason is we usually
use a spectral features for

detection of desired molecules and
also finding the concentrations.

So we need then to look at the spectral
features of the transmitter

light in order to see
the signatures are valid.

We are looking for and then there
is always a delivery region which

in most cases is a microfluidic
Channel and of course after

the spectroscopy we can have detector
array and electronics on chip so

that would be a system that does
the detection analysis and processing on

a single separate cunt chip which might be
in the order of a few millimeters in size.

So there are a lot of applications for
such devices bias and think and

Health Sciences is one of the major ones
and of course environmental fencing and

bio and chemical detection
Homeland Security that type of thing.

If you look at the challenges that exist
in making very many H. or devices.

Well we need certainly
a high sensitivity when we

shrink the size of the device
in the hope of reducing.

Amount of a specimen needed.

We need to have a very sensitive light
matter interaction to be able to detect

the presence of the molecule while the
concentration is very low and therefore

this can be transcribed into a strong
light matter interaction which could be

enabled by having a very high intensity is
of light interacting with some molecules

the system has to have various
small size in general and

also we need to have a non cheaper
spectrometer in order to reduce

the size of the system and I won't have
him back here spectrometers next to

the very thing girl chip that
defeats the purpose then we need to

have a coating a structure that kind of
attaches to the molecules of interest.

So we have kind of true domains
Wunderman is the optical domain but

the device is designed and
fabricated on solar car and

then the second the main is
like the approach we choose for

attaching the molecules to the surface
deliver is usually easy using

microfluidic channel but then we need
to develop surface coatings that

allow us to kind of trap
the molecule of interest.

Well other molecules that exist in
that fluidic channel are not attached

to the system and also the sample very
very Macon is them having it compact and

reliable would would be
another challenge to solve.

These challenges from
the level of difficulty

one of the thing is this
high sensitivity systems

that we are developing cultural high I.Q.
resonators to address and

the girl is by high I.Q.
resonators we will enhance the field into.

And city and as a result we will
increase the concentration without

increasing the size and then on cheaper
spectrometers there was another challenge.

There's extensive knowledge for
the coatings but still there is need for

that and also the sample delivery
man mechanism is probably the least

problematic area because there's a lot of
effort in the area of microfluidics using.

Some of the fog of fear and
the need here is probably minimal

considering to the amount of
work that has done in this area.

So.

In the rest of the tug hour we'll talk
about these two important challenges and

the solution to address those challenges.

If you look at optical resonate horse and
Chip.

We usually have a structure that is
in the form of a desk or a ring and

then the electromagnetic energy
kind of circlets out that in the so

called Whispering Gallery mouth.

This is very useful because usually even
you have a long propagation length and

in this length of propagation you
have interaction with the molecules

you kind of accumulate the signature
of the molecule over that length.

So if you can kind of wrap the length of
propagation around the same pair if you're

really then you will have had effectively
large propagation lengths without

having a large size so that's the main
advantage of despair in gallery modes and

these resonators at high key
you have two major property.

One is they have a long photon lifetime
meaning that a photon can travel

propagate many times around this
this before being dissipated.

Or transmitted into
a waveguide next to that and

the second one is the field in hand
spend the higher the quality factor.

The higher is the field intensity and

therefore we are kind of using two
things to increase the sensitivity.

One is using the huge intensity
of the field that increases

light matter interaction and
to having an effect of little log length.

By wrapping around the same volume.

So there are lots of applications for
high I.Q.

resonate towards including the optical
signal processing sense in which is

the concept of the star and
also a non-linear optics and

cavity Q.E.D. Which one way or
the other use the high field intensity.

And also the long term lifetime.

So in this way you if I have
a resume ter coupled to a wave guide

then the mode of the resonate or
will talk to the wave cried mode and

if my light has a frequency at the mode of

the resonator that frequency is dropped
from the wave guide into the resonator.

If something attaches to
the surface of the resonate or

the index of refraction changes
the resonance frequency changes and

then by monitoring
the output the spectrum.

I see a shift in the in the resonance and

therefore I know that something has
attached to the surface after resonator so

that allows us to have a very compact and
high sensitivity structure.

So let me give you an overview of the type
of high I.Q. resonators that have been

a study in silicon and related to
silicon Metairie all of their work or

to use are still up ten in silly
car micro toroid devices at Celtic.

These resonate tours are flooded
in silicon dioxide and.

Therefore they can not be use
in silicon based platforms.

They cannot also be connected to
the waveguide they usually bring

a tape fiber next to this resonate or
talk to that.

So it has its own application but
not suitable for

the lab and chip or structures.

Then there has been a lot of recent work
in photonic Crystal resonators where

head for structure over the period of
time to Crystal is changing from one

side to the other has been designed to get
to cues in the order of a few millions.

In principle this is appropriate for
lab on chip the main problem is if you run

a couple this resonator into a waveguide
then you have to remove one row of these

holes and therefore you cannot put your
waveguide at any arbitrary distance

to this resonator you have to bury
the pair you do city of the structure.

Therefore you cannot have
the optimal coupling of the two and

that usually cause problems when
you want to have a strong coupling.

Then going back to discuss
structures in solar current this is

the architecture of an undercut this.

This is Silicon desk then it
used to be silicon dioxide here.

That is under cart and then everything
is on top of silicon substrate

and the reason this undercutting was
done because it was believed that

the silicon dioxide here reduces the Q.

and by undercutting cues in the order
of a few million where at Cheve.

Again the problem here is when you
undercut your a structure you're going to

undercut the surrounding
then you can not have

the waveguide easily coupled into this.

So you have to use outside fibers to
couple light into and out of these

a structure and for lab and ship our
play cation it is not appropriate.

But the still it has its own
applications in the others.

I mean it's like quantum computing and
quantum processing

their idea of our structure is this where
a waveguide and a resonator are chip.

Well talk to each other and then that wave
grad will take the light into the next

stage but then the problem
is in a couple of years ago

the best use reported there in the order
of ten to the five hundred thousand and

there was a big belief that well
the reason that you get such a low Q.

is because silicon dioxide causes
linkage of light from silicon

into the bottom substrate and
therefore you have to live with that.

So we did a lot a lot of investigation
on the role of the silicon dioxide so

this is a typical silicon on
insulator platform or S.O.I.

substrate that those of you
who work in commune room

are probably aware of
the idea here is if you

leave this oxide underneath solar
car it has lots of advantages.

First of all it allows you to
have a lab on chip platform.

Secondly it improves
the heat sink properties.

So if you have a high intensity
light you generate heat in here and

then if it is silicon dioxide
it's much easier to dip dissipate

compared to the case when you
undercut it and there is air in here.

It's also possible to integrate with other
elements up on chip and then you can also

add active elements to that like a mass
transit store by using so-called

pedestal best where you are allow a very
thin layer of silicon to remain for

just don't ping to make a P.
and junction and it's also robust.

So we did a lot of investigation
to see what this does

to the Q Of the resonator
kind of to work against.

A common belief in the field and
then we notice that if you remove

all the imperfections assume everything
else is perfect perfect fabrication and

no surface a story perfect
substrate the only reason for

reducing Q. is the existence of
silicon dioxide then Q.'s should be in

the order of ten to twelve which
is not the case certainly so

we can clue that that despite
all the belief in the field

the existence of silicon dioxide
was not the main reason for

low kill and the low key class primarily
by the fabrication imperfection.

So we spent a lot of time to clean up
optimize a fabrication process that in

essence is just a routine fabrication
processes so you start with

the silicon on insulator platform you add
a layer of thin thermal silicon dioxide.

Then you add an electron beam resists
you do electron beam lithography and

then you transfer despair into silicon
dioxide using plants Maya chain and

remove that electron beam resist
kind of making the heart mass for

etching solar car and then you add
silicon to make your structure.

So this is a typical baby guide coupled
to a disgrace on a tour we develop and

we got very good surface quality and
air traffic this by

working a lot on optimizing
the properties of the fabrication.

So fabricated the structures that
characterize in a very simple system

again we have a swept laser system that
covers a wide range of wavelengths in

the infrared and then we have light
coming into the waveguide and

then going out into a detector and
then we added all kind of

the Martin techniques in order
to minimize the noise effect.

So this is a typical output
you get transmission.

From this vague guide as
a function of wavelength and

each one of these Depp's
represent a cavity mountain.

And then if you investigate some of these
modes by looking at the band width and

then dividing it to the center
frequency be very able to obtain cues

in the order of a few millions something
that proves our assumption was

correct that you could still
keep the silicon dioxide and

then be guarded very repeatable
even many a structure is different.

Fabrication trials different structures
in the same chip and also for

the first few radial
modes of the structure so

that was a big kind of
change in the pace of.

Efforts for high I.Q. resume two hours.

Then we worked up to my eyes in the couple
and after the light from way got

into the resonator because for sensing
application the existence of a high I.Q.

resonator by itself is not enough we
should be able to send light from

waveguide Our it to propagate a lot
of round trips and then come back so

that in this process we monitor what was
on the surface after after resonator.

So if we look at the depth
of these develops.

We will see different transmission
reduction from a few D.B. reduction

all the way to twenty five D.B. reduction
and the higher the higher the reduction.

I mean the better the coupling is or
the more light getting into the resonator.

In theory of a guy cavity
coupling there are two

quality factors defined Ron is
the intrinsic quality factor or

cues zero after resonator this
is the quality factor defined

by the in clinical losses losses due to
surface a state due to a scattering and

this and that and then the other one is.

Q.C. order the coupling key to the losses
that are the useful losses which

is losses because of coupling
into the waveguide and

the optimal coupling occurs when
the loss after after resonator

is equal to the rate of signal coming
from a right into the resonator So

if these two rates are equal then you
have one hundred percent transmission of

power from waveguide into the resonator
It's called Critical coupling.

Well here we are able to
get critical coupling but

not refer are the lowest order mode
of the resonator that is of interest.

So we have to investigate in detail
what was the reason for that.

And by looking at the formal
offer couple inquire fish and.

Which are you don't like to get into
too much detail of its merits but

at least giving you the idea that couple
in question depends on the field intensity

is after two a structure waveguide and
resonator and then the phase mismatch

between the two restructures if you look
at the most of these are structures.

Each one of them has a propagation
constant or beta for

the wave guide and for the resonator
if these two betas are the same because

if not then the phase mismatch will result
into a washout of the copulating So

in order to monitor the phase match.

We looked at the effect to be in
their switches this beta divided by

zero propagation can stand as
a function of They've length for

different resonator modes and
compared it with the waveguide.

So as you see for
this disk of twenty Micron diameter.

Twenty.
My current radius the first and

second most of the rest.

They try to have a considerably
higher value of propagation constant

than the waveguide known and
as a result the phase mismatch is high.

The couple in between these two modes
are not good and therefore Q. are coupling

as a function of different distance
between the two is always very high.

It's like ten millions forty millions and
if the intrinsic use to millions.

It's killing her that you cannot
couple all of the energy here.

But if you look at the third you'll
see that the Q. is the lowest which

means here because of the phase
matching you get better couple and

better coupling means lower Q. and
then you can get critical coupling here of

end intrinsic Q. is about two million
as well so to solve this problem.

We need to work with the phasing One
idea is to change the disk radius ten

micron same waveguide couples to the
second order not much better shown here.

And therefore there is a systematic
way of designing business

structures by playing with two
paramilitaries disk diameter and

also the distance between waveguide and
the resonator So

now we are able to achieve critical
couple into any mode of interest.

There was another resonator structure we
also develop where we keep a pedestal

of very thin sixteen nanometer solar car
next to the resonator and the reason

we do that is in first of all it gives
us a better heat sinking property and

secondly it allows us to
kind of doping with Peor and

materials to have active
control of the resonator.

So again for this a structure we have
a lot of modes as observed in here and

then these modes have choose again in
the order of a few a few millions.

In designing these are structures high
intensity applications like sensing in

a linear quantum one issue that always
exists is we work with high intensity

is we want to have high sensitivity and
that usually causes thermal issues.

So in principle if you look at
resonator and then you solve the heat

transfer problem you will see that when
you increase the optical power coming

into the resonator the heat generation
will go up and that starts to change

the residence property of the resonator
because the temperature change will

change the index of refraction of solar
can through the thermal optic effect and

therefore that changes
the resonance of the structure.

So it's just taking it aphorisms and then
it cools down and it takes it back into

the resonance and eventually if you don't
do anything to solve the terminal issues.

You will go into instability region.

Meaning that temperature
changes changes index enough

to brings you off resonance the light
doesn't go into the waveguide and

then it cools doll it gets back to
the resonance and this repeats.

So in order to solve this problem.

Meaning that allowing us to operate at
higher and higher intensities we need

to look at the temperature distribution as
shown all are here and the main issue is

the temperature is the highest where
the intensity of the field is the highest.

And this curve shows part of or
inside the resonator

divided by the thermal conductivity
of the structure and this is the cue

up the resonator that as you see we
can only operate in this region.

Otherwise our resonance across
start to deviate from their

original resonance because of
the change in temperature.

So the first message here is if
you want to get to higher and.

Partner role keeping everything else.

You need to increase the turmoil
conductivity of the structure as much

as you ten.

And if you look at
the kind of the properties

of the undercut structure as
the existing the structure before

we start this research and the discussion
substrate to there the format with

the pedestal Venti is nonzero or
the conventional one mentee is.

We will see that this is a structure has

considerable advantage over
the undercut the structure.

It is shown over here.

So what we did here is we monitored the
temperature inside the structure at each

different intensity and the normalized
everything to the temperature raise here.

So this is the normalize
Delta T.-Max which means the maximum

change in temperature inside the structure
divided by the maximum change in

temperature in that undercut structure and
as you observed.

When T. is equal to zero which
means conventional their skin.

We will get depending on whether we put
our on this top or oxide to increase Q.

we will get at least a factor two and a
half improvement in the thermal attack and

if we allow a Peristyle of fifty nanometer
exist then we can get a factor for

improvement in the intensity
of the structure.

Now a factor of four in intensity
means huge factor in the linear

effect because now linear effect very
good some power off the intensity.

So this this is one of
the major areas that these

structures are better than the existing
ones and now we are working

on reducing the thickness of silicon
dioxide from the convent Chanel.

And all the way to the limit
that you will just barely be

limited by silicon dioxide and
then by reducing this thickness.

You will increase
the thermal conductivity so

you still can use it at
higher values of power.

So I think that summarizes the fact
that now we can repeatedly design

a structure this Cuse in the order of
a few million with high temperature

stability allowing us to have
huge light intensity is and

that is mostly enough for our right
mater interaction the other challenging

thing is our outward compact on
cheap spectrometers using for

tonic crystals as
a introduction into photonic

crystals they are periodic structures in
the dielectric material as shown here.

This is a C.M. of a silicon based for
tonic Crystal the holes represent

a periodic area of air in cellar Conn and
there are two lines up

in Crescent applications are photon
a crystal one line is for

tonic bandgap which is a range of
frequencies Verno photon is our loud and

rich people use this structures to
make cavities waveguide lasers and

into a greater optical structure.

The secular line of application
that is the concern here is

unique dispersive properties
of these are structures.

So here I have shown the dispersion
in the form of constant

frequency contours in the propagation
domain K. X. and K. right.

Are the X. and Y. components of
the propagation vector in two

directions X. and Y.
and A is the lattice constant here.

So each care of shows
the constant frequency contour

at different values of K. just to.

Remind you in a box of material
a mega frequency is C.

time scale speed of light times
propagation constant magnitude

of propagation constant so
the current to Ruby a circle because K.

axis crowd plus K. Y.
Square is Omega square overseer Square.

It's a circle.

That's what we see here but
these are structures are capable

of a range of this person
properties from bark.

To very flat.

To very sharp variation.

And also the curvature could be positive
and negative opposite sign of curvature.

So it allows you to develop
dispersive devices that

are not possible to be developed by
any kind of partner Terry are and

each one of these regions that I
talked about cons like flat region

sharp region positive or negative region
have their own application in the field

of optics that I will talk
a little about some of them but

our focus is on wavelength separation
in a very compact structure.

So a device that does the veiling
separation either in the form of a very

lengthy multiplexer or a spectrometer
in an ideal form would be a device

that separates the wavelengths in a space
and then the separated wave lengths

propagate a minimum distance to
be completely independent and

then they don't deflect or increase
their size inside their structure and

they are also separated from
the unwanted contributions a straight

right unwanted paralyzation and
things like that.

And then in addition to that they
are compact and so on and so

forth and of course in the back materials
you can to do that like in in a prism.

You can separate colors but.

You have to propagate a large distance and

then the variable length distance
between two edges and collars is a lot.

So the fraternity crystals
allow us to get very

close to this ideal a structure that
no balk at that area all can not.

So you know to show that I'd
like to introduce three.

Effects in photonic crystals

run is the super prism effect the sharp
dependence on the wavelength.

So in super prism effect valence
are separated in a photon a crystal

in a very dramatic fashion so
this could be all the tens of microns or

so and these wavelengths could
be a few nanometer apart.

Then the second one is the so-called
negative diffraction where

the diverging beam of light is brought
into focus by a photon a crystal

by having a negative diffraction
compared to what about the material has

a negative refraction is another property
that they claim is available for

photonic crystals that light that
comes into the structure or a fact

in the direction that is usually opposite
from what is coming mentioned all.

Well all these three things have been done

in separate what we did in this
research is we were able to design

a structures that combine
simultaneously all these three effects.

So in that sense right.

Comes into the structure with multiple
wavelengths then different wavelengths

are separated by the by the super
prism effect after separation they get

focused by the negative diffraction
effect into a small channels and

by negative refraction they are separated
from a stray light so that combines all

the requirements that an ideal separator
needs and then at the output of this.

Structure we usually use an area of
They've got to sample the signal and

monitor that and
we usually put two channels per each

way guys per each channel just to get
a better idea about the crosstalk.

Well we were able to prove that there
are a structures that I want to

have these three effects together so.

The regions that are at the corner like
that are the high super prism regions.

Then we have the negative refraction
region represented here and

also negative diffraction region
added to that and this for

tonic Crystal was fabricated
a structure like this.

So we have an area of waveguides that
we could use to extract a structure at

different angles of incidence then
we have a region of solar car

that is just allowing right to
propagate before getting in and

this is where our photonic
Cristela structure is.

And we were able to demonstrate
in a structure that has a hole

sizes of two hundred
nanometers in diameter with

a period of three hundred seventy
nanometer for channels eight nanometer

spacing are separated with
better than six point five D.B.

true crosstalk isolation and
about one point five D.B.

non-uniformity Well this was the first
ever demonstration of four channels

with this much wavelength separation and
the size of

the structure compared to all existing
run was two orders of magnitude smaller.

This is structure is about seventy Micron
by seventy Micron this is the photonic

Crystal region and
each one of these waveguide is six

micron and is allowed to sample
the signal in the output.

So this is the actual experimental
result of what we are observed

at the output of these
waveguides monitored by.

And infrared camera as you see for
each wavelength.

That is designed in this plex air.

We have the output of
out of a rat's monitor.

So for this wavelength we have
to wave guys that have most

intensity as designed for
to correspond to each wavelength.

Then the second wave length has another
two wave guys and other to one another to

the main thing is the two major wave
grads are independent of each other or

separate which means you
have minimal cross start.

You have separation and of course there
is some cross stock as I mentioned it's

the cross start production is
better than six point five D.B..

Now to prove that the three effects that
I talked about family tameness Li exist.

We also monitored the output for
the unwanted polarization for

which none of these three
effects were designed and

the first thing we observed is for
the same three for Wave variable length.

First of all we see no suppression
they co-exist in different bad guys so

super prison effect exists here and here.

Secondly the sizes of the output
beams are large when they cover ten

waveguide sixty Micra here we have only
two wave guys so this means negative

diffraction really works where and
third and most importantly and

these are run to contributions or
occur at a set of waveguides here.

That is totally different
from the setup waveguides for

which to run to a contribution of care
is lab run going this way the other one

is going that way that is because
of the negative refraction.

So this really shows that the idea of
having the simultaneous effects works.

Not only that but also we can have
more degree of control by exciting

the input from a different waveguide
therefore having an effective angle

of incidence on the a structure
that is different from one another.

And what happens is well wavelength and
output response.

The same for
channel the multiplexing exist but

by monitoring it at thirteen fifteen or

seventeen degrees of incidence we shift
the wavelength cringe of operation so

we have that if we have
control also present for us.

Now our main goal in lab on chip
is to make spectrometers not

the multiplexers and
it happens that for designing

a spectrometer you have much more
flexibility than we do multiple X. or.

Well the reason is if you come
with multiple Waveland channels

in the Mounted plex or you want them to
be completely separated you would like to

allow no cross star because these
are the information bearing channels and

you don't run to any subscriber
get information of the other one.

But when you are talking
about a spectrometer

you really don't want to
separate the spectral channels

you want to estimate the strength
the relative strength of each channel.

So you can have some cross
stock in the output.

So for example here I have shown
a device that does a spatial

spectral mapping meaning
that different lengths

are mapped into a spectrum
a spatial patterns like this and

the only thing that I want is these
patterns be diverse and nuff that I could

use this as a linear system and do inverse
filtering in order to get my spectrum.

So all I do is I come out run wave length
at the time and train my system measure my

transfer function then I come with
the unborn the desired unknown SEC

now monitor what is in the output but
since I know my transfer function.

I do inverse filtering in order to get my
spectrum and universal turned is usually

estimation techniques like least minutes
in a school and things like that.

So here we have much more
flexibility in designing these or

spectrometers and
in addition to that elaborate chip sensing

I don't want a spectrometer that monitors
two hundred nanometer constantly.

What we want is the really kind
of adaptive system that looks for

a special feature.

So for example in if I have
a resonator coupled to a wave.

I have a dip in the transmission and

I want to see what happens
to distributor it shifts or

not the shift is what I want because
the shift will tell me that resonance

has changed something has attached
to the surface of the resonator So

I'm looking going to a special few special
features after the after spectra and

for that for tonic Cristela
spectrometers are excellent candidates.

So the idea here is very simple.

I want to design a pattern detector
a spectral parent detector that monitors

the shift after people are in ROM on
sensing monitors the presence of four or

five rather than peaks that's all I want I
don't want to monitor the spectrum of my

signal over five hundred nanometer that's
not the goal in a lab on chip sensing and

therefore we may need to reconsider
the design of for a structures.

So this is the result of
the same structure designed for

a very blank do multiple lexing as
an a spectrum better than the monitor this

what we do is we come at each
wavelength at a time monitor the output

then change the wavelength Bronnie told
the output in order to train our system or

in order to calibrate its transfer
function then we come with

a single peak and
try to monitor the output and

then use our system to estimate
the location of the pic and this is

what we get estimated wavelengths Russ's
the input wavelengths blue shows why.

We estimated using our spectrometer and
red shows the actual wave length and

as you see they are very
very close using the system.

This is their standard
deviation of our error and

as you see we have about point
three in detection accuracy or

detection error after
location after peak for

a peak that is one point five five
microns so it's about a factor of

ten to the floor in that now we went ahead

and designed the structure to
be an optimal a spectrometer and

not of a lengthy multiplex there and
it happens that for

that we don't need to have complete
separation of the channels so

as you see if you look at the same
area of waveguides the up to mom

has some overlap between different
wavelength channels and as a result

we were able to get a noise a relation
that was much better than that it is

a very lengthy estimation error as
a function of noise to signal ratio and

as you see for this region where
the noise is thirteen below the signal.

We will have ten P. commuter
wavelength accuracy and fifteen and

a meter band with now twenty could
meter wavelength accuracy is

ten to the minus eleventh and center
wavelength is about ten to the minus six.

So if my she might pick shifts
by one in hundred thousand.

I'm able to detected using the system

that brings us the capability
of single molecule detection.

Now this is the actual
performance of this structure.

Again we designed it to be an optimal
spectrometer not available and

the multiplexer and

this is what happens is that in this fifty
nanometer operation range as you see.

This is each line corresponds
to one waveguide and

each column correspond to one wavelength.

The key here thing is
that the variable and

channels are not completely separated
from each other they have overlap but

if you look from here to there you
will see a diverse distribution

in the waveguide for the power of these
waves that's all we need for the for

the spectroscopy we need a diverse
a special a spectral mapping

this will be enough for
us to measure the location of our peaks.

Now the more recent get
on published work we

have done is to look at the higher
bands of photonic crystals and

this shows that constant frequency
contours each number is wavelength in

nanometer and this is the prohibited
range just because in this range right.

Escapes from the top of the structure but

by looking at the incidence
at thirty degrees.

We were able to get a very good response
this is group for our city angle

as a function of wave length the slope
of this tells me that then I change my

wavelength by one nanometer how much
how much angle difference I have in

the in the direction of propagation so I
have a lambda in nanometers and the land.

Plus one changes its angle and this is ten
point five degrees per nanometer compared

to what we had in our first demonstration
that is point seven degree in a meter.

That is a factor fifteen yet
improvement in the performance

that when you have this kind of
improvement that you could divide it as

you wish to the size of the structure
to the resolution of the structure.

It depends on what your goal is if
you are if you have enough wavelength

a curacy you can make your structure
smaller by a factor fifteen.

If the size is and a half.

You can make your accuracy.

Right.
Better by a factor of fifteen.

So this is the structure that is
fabricated we need to add a buffering or

stage to minimize the reflection because
the group philosophy of the second band

is low for a structure of the first
twenty by fifteen Micron we were

able to get three nanometer resolution
rather than eight that we had in the past.

Remember this for seventy by seventy
microns so if you look at the kind of

reduction in the size and reduction in
wave length that shows the factor fifteen

and again we are able to separate
the Bay of legs successfully.

And this is the spectrometer
performance again for

a good range of signal to noise ratio and

we are again able to get ten P. commuter
at thirteen D.B. signal to noise ratio.

Let me just finalized by saying that
now that I have these desks and

I have these spectrometers then we
are introducing a new concept in sense

in which we call it multiplex sensing the
idea is we use the same real estate and

infrastructure by having a spectrometer
to detect multiple species.

So here I have each resonator designed at

a different wavelength and
coated with a different material that

can attach to a different virus or
molecule and

then this spectrometer separates
the signature of these resonator So

let's say it's resonator has frequency or
wavelength that is three nanometer apart.

So I'm dividing my fifteen meter
operation into three nanometer

windows in which one of these
resonators will operate.

So all of them.

Sure the same as spectrometer but
in the output I'm able to say OK this

is three nanometer for a species one
species two and so on and so forth.

The reason that marks and their
interferometers like that are Perth for

about because you could make them in
the balanced form that you have to or

as an a tourist.

For a species.

If you are fabrication there
are terminal drift and

things like that that will be compensated
and one of them is counted one of them.

So the phase difference of the two is
monitored in this interferometer by

the a spectrometer so all we need is
in set up a laser we need an L E V And

then we can have this wavelength
division multiplexing

sensing rather than optical
transmission that you around.

There are and as a result these
two technologies are able to after

a structures that would be at most
including all the analysis in the range

of a millimeter on a side that are able
to detect a hundred different species

with the accuracy of
a single molecule detection.

So that's a very attractive solution for

lots of applications hand health and
stores.

Probably Top care detection
of infectious diseases

detection of explosive and
so on and so forth.

So in conclusion.

Silicon photonics has a lot to
offer because of the unique

properties it has for designing different
up to call devices that could be

called Naked in the firm off and
on chips or stem with the addition of

advantages of available fabrication
tools because of CMOS technology and

relatively low cost of the substrate
we could get very high Q.

resonators in silicon that facilitates
the light matter interaction for

all to our sensitive sensing and
then by using photon a crystal we get very

accurate detection spectrometers
that could operate

in the right range of a plank with a very
good accuracy and then by combining.

And adding the coding and delivery their
invasion in having very compact lab on

chip sensors that could
hopefully revolutionize the.

The kind of the sensing market.

Thank you very much for the attention.

I am.

So the first question is right.

Sensing mechanisms have been in
the region we ourselves are looking into

something called multi-modal sensing.

So we are looking into three modalities
of sensing implemented on the same chip.

I explained in this talk
is in that sense think so.

The sensing is done by monitoring
the index of refraction on top of

the after resonator when you
change the index of refraction

the resonance frequency of
the resonator changes since

it's a very sensitive resonator the shift
of one pair of hundred thousand

is enough to be detected because
it's a very sharp sharp peak.

You could envision things
like fluorescent sensing than

the molecule that attaches to the surface
is pumped up the crowd beam and

then it's fluoresce and then you have
mechanisms to grab a good portion of

fluorescent into the waveguide and
monitor that that would be useful for

biomolecules and then the third one
would be surface enhanced ROM and

sensing where you pump
your signal on top and

I'm sorry the molecule on top
of the resonator using the rest.

And then you get the ram on signal
back into the resonator and

into the waveguide for processing
the second thing that you ask is

What are the mechanisms that can change
the resonance frequency of the structure.

Well of course the first thing
is the radius of this desk

which is done during the design.

Number two is the index of refraction
after the center after this.

And it's Perry ferry within a few

tens of nanometer because the mode
usually dies out very quickly.

So if something attaches to the surface in
that Perry ferry then you're able to get

the change in the effect of index
by changing the center of the.

The residence the next effect
is the thermal effect so

if you change the temperature that
changes the index of refraction again and

that change will change as
there wasn't a big change or

resonance frequency in the next
one is the free carriers.

So if you inject carriers electrons or
holes into a cell or can that kind of

acts as a conduct levity and that changes
the index of refraction again and

that changes the resonance
frequency of the structure.

So all these mechanisms
that have been studied.

Well one way or the other change
the index of refraction and

people used for example injection
of carrier to modulator signal.

So you are passing light from a wave guide
next to a cavity and then you change

the resonance frequency to be on resonance
and off resonance and therefore light

propagates but doesn't propagate so
you have bits of ones and zeros and

they have demonstrated sixteen gigabit
per second modulator using that effect.

But these are the main mechanism that
change in the Excel for a fraction.

Yes well.

If you would like to censor
our food is there yes it does

not too big a deal but
it's a relatively big deal because suppose

you have a carrier fluid
that incorporates a very

low concentration of the molecules that
you have so it's going to be a rush for

you through it is usually used to deliver
the molecules and then it is washed and

removed so at the time of sensor we
don't have anything in the free channel.

So basically fluid brings the molecules
attached to the surface when it's rushed

away the molecules are still attached to
the surface because it's a bonding and

then and then you have no
fluid at the time of sensing.

Now having said that there
are ideas that people

up to a fluidics are investigating
to bring the fluid

to module like the property after
structure so for those ideas.

You need to have the fluid in there.

Order to make sure that you monitor it the
way you run a modified their way around.

And then they envision all these
kinds of plants and combine and so

that you have different fluids with
different in this is combine at the right

proportion to kind of marginalized
the performance the way you want and

sensing is just the carrier to bring
the molecules for them to attach that's it

it's removed at the time of since you're

me remember the limit

of detection after a light where the for

sure we have not demonstrated
the entire system yet the rocking.

OK you can detect for a fraction in
the order of ten to the minus seven so

Delta and in the order of ten to
the minus seven can be achieved but

design is appropriate in principle there
is hope for single molecule detection.

But there is hope for
single molecule detection

experimentally we have not demonstrated
ten to the minus seven get

our exact resonance frequency.

For modulator that's the problem.

First of all these are disks.

So these are much harder than rain and
the reason is the range has the cues

in the order of a few tens of thousands
two hundred thousand million.

So it's just a very sharp pick if you have
a little arrow you're going to miss it.

For communication application
that's a big deal for

sensing it's not a big deal
because what you're looking for

is not an exact frequency
looking at the frequency shift.

No matter what your fabric you
get close to what you want but

the thing that defines the sensing is just
a shift in the location of the pic but

the exact location of the pic and
the exact

location is monitored at the time of
kind of training the system I tell.

Rating the system but then when you look
at the shift if you choose the chip.

That that is not a big issue and
sends it back for

optical communication and
signal processing.

That's a major issue and in order to
fix the center of your resonance.

Then you need to implement some
feedback mechanism like a thermal

terminal signal just apply
some temperature change in

order to change the index to just bring
it exactly to the place that you want.

Yes.

Critically on a meter the best that
you have demonstrated is fifteen and

a meter and the reason is you need
to have super prism effect and

other two updates in order
to have that and fifteen and

a meter is the best that we have achieved
to date that we have only three effects.

So as a result.

These are structures are not good for
wide purposes spectrometers

incensing Usually the amount of
shift that you're looking for

is not even run nanometer
is an order of meter.

So for sensing there's no problem you
can have multiple channels mandatory but

if you're looking for

just looking at their spectrum of a
fluorescent object or something like that.

This is not the right tool to use

and exactly.

Photonic crystals.

If you kind of double the size of
the structure it does the same job

at double wavelength.

So you could kind of bring the operation
wavelength to any range you run but

just scaling the structure.

Terry out is you should
not go to live lengths for

with silicon has absorption so
if you try to come back.

Roughly nine hundred.

Fifty nanometer or so
when you start getting a lot

of absorption from silicon and then you
get structure would be ultra Lassie.

But in the infrared it's possible.

And just one comment I'm
glad you brought the issue.

You may think that this is
limited to infrared but

it's not because we are demonstrating
the same concept invisible and

near infrared using silica nitrite
as the substrate which has much

much smaller loss visible and are so it's

possible to just choose the wavelength
of operation so this is my range.

Let's do it you

know the thing is you have this
photon a Christan let's say.

Comparable to this area.

I believe that's right.

So if you change the entire size of
everything by ten percent just operates

at the same the same performance
at the change by ten percent.

So this is a land that zero.

This is which is a feature size and

again the main issue is since
you just monitoring the index

of a fraction the very violent up
operation is defined by the structure and

availability of sources and detectors not
by the molecules because the molecules

index doesn't change really two hundred
three hundred nanometer window so

that gives you the flexibility to choose
the wavelength up operation in there and

you want if you're working for example
that fluorescent turned out to turn

a different story because the fluorescent
of your molecule is being monitored and

if you lock in with surface
enhanced ROM and you put Mattel.

Like a structure and then you need to work
in the wavelength that you get the surface

enhanced meant for that specific metal
that is mostly in the range of visible and

real I.R. like eight hundred nanometers so
that for index.

You don't worry about the Troy stuff
they've length related to the material

that you want to sense.

Thank you.