OK closer to

your in department that's got his
undergraduate degree the University

of Pennsylvania very sharp
got his master's there.

He worked for a postdoc to Paris and
he was for two years.

George French culture there for
a while during MIT in two thousand and one

interesting note is that when he started
teaching there at MIT first actually

being very successful of the MIT one
number as awards I'll just mentioned and

this is a real award of Asian award and
a Guggenheim fellowship

to talk about the house also be part

of the Endeavor Firefly via works if
you're going to talk at all about.

The Stars.

Thanks Victor.

So first I'd like to thank you.

Actually all the faculty for inviting me
and hosting me today with a great number

of conversations today I want to tell you
about some work I've been doing in my lab.

It's actually relatively new and it's
only over the last four to five years and

it's not really something I had on my
radar screen when I started at MIT

you give these seminars about
what you're going to do for

your next ten to fifteen years and
this was definitely not innit.

It was from a discovery that we made and

a number of clever students contributed
to it and so it basically is

going to tie together themes of
microfluidics micro particles and

making them and thinking about
something interesting to do with them.

So let me just start by actually
thinking people so I don't forget.

So these are among my
first two students and

they've really pioneered the development
and innovation in this in this project and

what's interesting is they independently
have founded two different companies

based on the technologies.

Dan and I have a company in around the MIT
area and then actually has a point to

carry diagnostics company in India another
number of very talented students and

post-docs have continued on
the project but I'll talk about today

we've been funded by various sources
specially N.S.F. and then I have been

very interested in my colleagues at
MIT I should even put more people here.

It's a very collaborative atmosphere and
we've been collaborating

since the inception with
Professor Hatton and more recently with.

Professor Gleason and strength.

So I'll talk about particles but
I did want to say something about what

the majority of the rest of my lab
does this is one slide snapshot.

So I'm actually trained
more in Palmer physics and

we'd love to look at single D.N.A.
molecules and do experiments

those devices and manipulate them and
we love to stick D.N.A.

into small confinement in their stand how
their conformation and their exchange.

Are we like to build devices in which
we can take a single molecule and

we can stretch it out and linearize it.

And that's very interesting for both
understanding of fundamental dynamics but

also doing a single
molecule genomic assays.

And then often in the course of doing
this we discover a new phenomenon.

For example just stretching out
a molecule watching how it relaxes

we discover things about how.

First relaxation or even just this month
we just published an article in P.S.

that showed if we jostle an electric field
we could actually not a single molecule

and look at single molecule knots and how
a knot relaxes at a single molecule level.

But I'm not going to have time
to say anything about that.

I'm going to talk instead today about
micro particles in complex micro particles

then we got into this merely by thinking
that we wanted to synthesize particles

that are more complex than just
the standard polystyrene fear.

And so we wanted to make things
that had interesting shapes and

virtually any shape that we could draw and
think about.

Think especially about putting chemistry
in select regions and I'll show you later

how we do all this but what you're seeing
is actually a collection of particles.

So this was around Christmas time.

So my student put a little Christmas tree.

Not very functional but they showed
exquisite control every color is

a different chemistry and
we make millions of these at a time.

You know high in the West the ratios
features bio compatibility is very key.

So a lot of the things I'll show
you when we go to applications.

The fact that we're actually
making these out of hydrogels and

some of these applications this is
a particle which was made around cells and

you're seeing all the green
which is the fact that most of

the cells are still alive after
I make a particle around them.

We make them feel responsive
these are magnetic particles

actually responding to a field in this
direction they're not forming in shape.

This is actually their shape and we're
trying to understand how shape combined

with fields can give rise to new
structures we can make them very soft so

traditionally this fear you buy out
of a catalogue is very rigid and

you don't think of it to forming under.

Small.

All pressures these are things which
are small as a red blood cell and

the form they can fold over on themselves.

So actually this half donut is this
particle which is folded on it so.

And lastly.

When we make these out of hydrogels
they're actually great to load with

a number of materials.

This is a collaboration we did
years ago with Jennifer Lewis

actually making ceramic materials but we
can make them anywhere from you know very

rigid to very soft loaded with cells
loaded with magnetic materials etc.

So before I tell you how we do this and
we were motivated in part by the world and

so there's a very nice review
that came out by Mike Solomon and

sharing plot sort of so
there at the University of Michigan.

And they really brought
together a number of ideas and

added to them that were already
existing in the close to

literature which was in thinking about
self-assembly and maybe emergent behavior.

You'd really like to have more
complex building blocks and

so they tried to summarize or sort of come
up with a periodic table of complexity of

the shapes and chemical patchiness is what
they called it so this is a good example

of a chemically patchy particle where
you selectively maybe put a hydrophobic

region in yellow in the center which
is flanked by hydrophilic regions and

you want to be able to say exactly where
it is you know is it in the center or

maybe over on this quarter or
this quarter.

Except for a going to make a diversity of
shapes you know linear shapes shapes which

are branched these are sort of evoking
motifs from polymers you can think about

cross-linking assays that might emerge
from this aspect ratio making things which

are faceted or just the changing convex
concave interfaces Excedrin and early on.

Sharon had done a lot of these simulations
sharing plots are showing you know putting

maybe chemically patchy regions
on a particle in silica.

Could arise very interesting structures
and sort of you know the idea of thinking.

About your Lego kid you know as Lego
has got more advanced over the years

you know when I was a kid you just
had the very simple Lego as my kids

get very advanced ones and they build.

Things out of a Star Trek and
things like that.

I built little square houses.

So maybe if I have more
sophisticated Lego's I can build

you know the ship or something.

And then I'm a reality
just by training and

so this is a real image of triangles and
as you put these suspensions the higher

volume fractions actually go ordering and
packing of these is going to change.

And so the flow ability of
reality will change drastically

these particles are pretty rigid This
is a different soft particle and

so you know if I can make things that
different shapes that are very soft maybe

how they flow in things which are trying
to mimic our body may also change.

So just from these fundamental studies we
wanted to have a tool box to do this and

then maybe we could do
something applied this useful.

So that probably the genesis
of the this sort

of advanced functional Collatz I'd say
are probably in eat paper where you have.

Which might have black and white on
different sides you apply field and

you can make pixilated structures which
have micron in texture and length scales.

Actually Julie did
magnificent work in her Ph D.

in looking at shape and how cells
recognize the shape of particles and

the the associated just what she
does lots of other cool things here.

But these showed you know you
have the same chemistry and

you create different shapes and
cells consensus.

I'll tell you about these particles
we're doing in diagnostics but

some of my colleagues also like to
build little building blocks and

trying create emergent
tissues out of these OK So

we've got to create things that
are on the micron this length scale.

We've got to create diverse shapes.

We like to pattern chemistry and

we'd like to make them bio
compatible with the end of the day.

So we sort of stumbled on.

This.

So we sort of rediscovered some
things that were already known and

added to it and made a process.

So this is Dan Dan here is
about two hundred microns.

So he was one of my early students he was
actually trying to make hydrogen pads and

stick them on to surfaces and we wanted
to do some so your ass is with them.

But what he was doing was we were too
cheap to go in the clean rooms and

do all this so we were actually
doing projection lithography

basically taking our microscope
that sits in our lab

taking a very cheap transparency
putting it in the right.

Optical plane using the objectives which
are typically the most expensive part

of your microscope and using that to D.
magnify the image down on the surface.

OK So the thing that went
horribly wrong in his experiment.

Well he thought was horribly wrong.

Well he was trying to do one hundred
microns which was actually pretty good but

it was horribly wrong that these
things could floating away.

And so you know he wanted to create
structures that sat there and

didn't go away and then he wanted to
put a fluidic channel on top of it so

he made thousands of little
bands that were floating around.

OK.

And it just so happened we were making
these out of hydrogels I'll tell you more

about the materials.

But it was very repeatable.

I mean it's a little graphic process.

So you know his face was
basically ingrained on.

Actually the mass that was in situ here so

we got a replicates of then
you know at this length scale.

So we immediately saw
this as a way that we

could potentially print particles but we
had to think about why it was happening.

You know we didn't know why
he was sort of floating away.

And you know while I'm the chemical
engineer my background in

chemistry is pretty shaky.

So we had to go back and
think about you know basic concepts and

photo chemistry and things that.

People who actively use photochemistry and
free radical Plimer zation know well and

so we were using a certain monomer late so

that the actively groups on the end
are actually going to be doing

the cross-linking one third
initiated by a photo initiator OK so

you hit the U.V. light that creates
a pattern kicks off your reaction.

And you propagate forward in time.

To create a crossing network and
generically

you know if I don't put any chemistry in
this you know I've got an initiator a hit

light I create free radicals can kick
off a series of different reactions but

there's always a competing reaction.

And so you typically try and
the oxygenate as well as possible.

You're simple because oxygen
is very reactive with these

free radicals it's a quencher.

So if you're in the lab you
trying to oxygenate things.

If you're in industry and
you lemonade things you just say that

there's a so-called skin layer or
layer that doesn't Plimer eyes and

you deal with it and you wash it away and
you have your lemonade underneath and

the skin layers turned out to be
typically on the order of a micron.

And everything that I'm saying here
this is very general to just free

radical plumbers ation we just happen to
use a lot of accolades in our group and

particularly like payback relates.

So what we realised was actually this was
happening because of the fact that then

I was trying to do this and.

The bottom surface was
actually a slab of P.M.S.

which is actually highly permeable
the oxygen so he essentially had a very

oxygen rich region and
then to show you more about that modeling.

But essentially provided an oxygen
source that was quenching reactions

just at the bottom interface so

that they could never have been
happening near that interface.

And we sort of turn this on itself and
said Well.

You know let's combine this idea of doing
lithography but instead of just having an.

Of a fluid.

You know let's explode everything that
our lab is doing in microfluidics and

that others do in terms of using small
scale flows to pattern liquids by flow.

So the rainbow you see here is just
because of the channel topology and

the relative pressures you're
putting at the various ports and

maybe if we can combine these two and

the concepts I just describe we
could then print out particles.

So we're going to use light to
define shapes and patterns.

We're going to use flow and the fluid
channels to dictate where chemistry is

we can use light to join them together and

we call this process stop flow lithography
So essentially flow and lithography but

it's better to stop your flow because
otherwise you use your everything else.

So.

Going back to the basic process
is something like this.

So here's a microscope.

All down here.

This is supposed to be a mask.

So this is a basically a transparency a
high resolution transparency here it just

has off of a clear region and so light is
going to go through the clear region or

use optics to D. magnify this and
my device is made out of P.D. a mess.

So anybody who does microfluidics this
is the workhorse of microfluidics if

you don't go home in your shower and
look at the caulking it's P.M.S.

which has a little bit of
a filler to make it look white.

OK so it's a soft elastomer
it's very easy to work with.

And it sets pretty readily and
it's very permeable to oxygen so

we basically have a device which
allows a lot of oxygen to go through.

I'm going to then introduce my
monomers which will be some sort

of typically are trying whatever something
which can cross-link a photo initiator So

there's a very fast competing reaction
a high Dom core number that actually

says that I won't jell actually have
a liquid here but actually form and I.

Particularly to particle
a gel particle in the center.

So with that layer that doesn't arise.

I can easily flush things away so
I can apply pressure poles and

I don't create one particle at a time.

I have an a hole that I typically create.

So then we automate this and we just
cycle through plume arising in flushing

out particles so we have if you want
a particle printing press at this point.

So we've done a lot of modeling and

I want to show this is just sort of for
a talk a simplistic version but

this doesn't have all the reactions but
it said the state.

You know from you know even
undergrad transport that if you look

at a first order reaction it's
sort of a steady state and

you make it dimensionless you have
a number that pops out and so

if there's Dom core numbers
large you have a boundary layer.

And you can sort of turn around not
knowing what the length pro-create length

scale is you can pull it out of here and
get an estimate actually of how

big this this boundary layer is and if you

plug in typical numbers you get something
like you know order of a micron or so.

So what we have essentially is we have
a good reaction all throughout but

we have these little boundary layers
that are happening at the surfaces and

if you put in a full bottle basically for
conversion.

So if this value is one that
means that I have not anything.

And this is a dimensionless
coordinate zero is the bottom

of the channel one this the top.

This is a dimension less time over
time I convert just in the center.

But I have this nice sharp transition I
have a boundary layer just at the edge.

So really I create these two distinct
regions in my channel where I make

a particle and other regions where I
don't and I have this recording layer.

So we can bring it together.

And we spend a lot of time figuring
out how to do this quickly so

we can actually make enough
particles that we can either

do reality whatever we want with or
send the collaborators.

But the key was starting and
stopping flows quickly and

so classical syringe problems which
is sort of work microfluidics

grew around are not fast enough.

There's too much compliance and
we build homebrew pressure systems and

there's a very nice actually
lubrication in the hydrogen I

make sense problem that you can do
in thinking about the response time

of this device because every pressure
pulse I do in my microchannel if

I have high pressure here to flow
in monomer lower pressure here.

I sort of in a maze that generated
way deflect my device and

once I quit liberate these two sides and
pressure.

Well the height changes
because this is a membrane and

it wants to squeeze down to here.

So this actually sets
the compliance in the system and

so you can actually exactly solve this.

This will squeeze flow problem and
say that this drainage time depends upon

a number of parameters of your device and
the material properties.

So you might expect that more viscous
fluid drained slower was viscosity you

know scales quadratically with the length
of your device linearly with the with this

is the modulus of the material you make in
stiffer material it responds faster and

it's very sensitive to the height
to the third power and

we shown all this in modeling and

also confirmed in experiments and what
that translates to is we can then design.

A priori devices that have very fast
response time so there's a time stamp that

you cannot make out but
basically we do a cycle per second.

So we can flush an Plimer eyes and flush
it out and be ready again within a second

and a lot of it was dictated by actually
making very fast responsive pressure.

Systems.

So the last twist on this IS THAT WAS SO
now I can flow in you know I don't

have to worry about the expendability
of my material or the things like that.

I just have to make sure it can flow when
and the fluids I work with are pretty much

Newtonian but
I can cope flow multiple fluids or

multiple fluids with embedded
different chemicals.

So this is a schematic of a device
of fluidic device which has

four entries so I can imagine the
reservoirs which have monomers in them.

So maybe this one has a dye that is going
to be conjugated green to my particle

just for the sake of an example and
this one's clear green clear.

So if I superimpose.

The light over top of all these streams.

I can actually conjugate
them all together.

So I dictate where the chemistry
is this is going to turn

into a particle now by the way
I introduce the streams and you

can do this in a very predictable fashion
at the at the microfluidic length scale so

then I'll translate this in
the whole series of particles.

If I vary this pressure make it
bigger than the stream gets wider and

I can very straight size on the fly.

So this is a cartoon this is real
particles an example these actually had

three streams they were doing some
genomic assays with some collaborators.

So they actually had in the center
you may be able to make it out but

they actually had different letters.

So I think there's like a K.

in the a few other letters to actually
give identity to the particles.

OK so now you can use flow to
the pattern straightness for

different chemical
regions in the particle.

So taken together now it's a pretty
nice way of getting what I might.

I think about as chemical
complexity year and I saw trape.

So different regions radio specific
chemistry at the micron like scale and

fairly complex shapes and we think
you know you want to be out here and

so you know some things can get you
slightly more complex shapes like a lot of

template techniques but they're not very
good at getting you very complex up here.

So there's a sort of sweet spot up here
where our technologies are pretty good.

It's a continuous automated process so
we can set it up in the morning.

You know we have many of these things
going and they generate particles just

an automated fashion you leave it and
go back at the end of the day very mana

disperse so all the particles we create by
NIST standards are monitored dispersed.

This idea of patterning
discrete regions for

having different chemistries
is really important.

I'll show you how we exploit that.

And you can put a bunch of stuff in
there and not be harmful to it and

part of the reason why is that
the wavelength of usually U.V. light that

we're using is not the wavelength which
is most damaging to D.N.A. or R.N.A..

OK so it's actually shifted
to a higher wavelength and

where most of the absorption and
actually destruction of D.N.A.

happens because the U.V. when we tried to
actually get this funded originally by

the in age that's that's why I got
rejected was because they said on are you

going to fry all your D.N.A.
And you know it's not going to work and

that doesn't happen because actually
there's very little energy of his or

by D.N.A. and we've also shown that
to be true for proteins and you can

make a lot of responsive materials just
by embedding or actually growing within.

Materials in these very open hydrogels.

So I want to segue from that this
is a very basic technique and

just tell you about two sub projects
from that this is one that Victor.

To that we started a company
based upon just a year ago but

this is sort of the Genesis and then
the next generation of what we're doing is

thinking about in coding these and
we have physical bar codes

making them in the hydrogels and
doing some sort of bio sensing assaye.

So this is one example of a particle.

So you see it has a bitmap so
the dark regions are actually.

In the particle this other region
I'll show you is a sensing region

was actually doing some D.N.A. sensing.

And what we wanted to do is to think about

you know what are some of the needs
in this realm and do hydrogels and

the way that we can photo pattern
them have any advantages.

And this is a very simplistic overview.

You know when you think about
different ways of using them.

You have to drill down the details but
I think for a talk like this will suffice.

Is that we realize that you
know we can mend markets and

people were people are interested
into sort of discovery diagnostics so

biomarker discovery knowing what
you want to look for when you make

a diagnostic doing that diagnostic
what may be doing drug discovery.

And you can imagine
companies like Novartis and

said if the event us are very interested
in all of these sorts of things and

it depends you know typically you
might be looking for D.N.A. R.N.A.

proteins or
other small molecules it depends upon you

know really what you're trying to do
just to give you an idea of the scope.

Novartis when they're
doing drug discovery.

They actually have their in-house you
know proprietary compounds which are one

point five million different compounds so
that's the Novartis library and.

We work with the head of high
throughput screening and so

they'll want to actually go through and
screening through all of those.

Compounds.

So the magnitude of complexity is
enormous and some of these problems.

So you know we didn't want
to solve all of them but

we thought about you know what is needed.

Generally so especially in
thinking about clinical settings

you still need high throughput maybe not
to the extent that I've talked about but

still sensitive in accuracy
pretty flexible design and

we can't make something that is just so
outrageously expensive.

You know that you just know from a gut
feeling and what you're doing in your lab

that will never translate you know
that is just too cumbersome and

you know there are compelling economic
reasons to think about these problems.

I think.

So we're going to develop
a multiplexing assaye So

basically the idea is you can envision
one scenario in which you have

a small vial of fluid and
that fluid could be a drop of blood.

It could be you took cells and
laced your cells and you're looking for

what's in the cells but it's typically
a small volume and you want to look for

a number of molecules in there so
you might want to look for

a number of different D.N.A. sequences
different R.N.A. maybe different kinds.

So today you can usually
do wanted two things.

These are the traditional ways so you can.

I think most people have
heard of micro rays.

So there are typically planar substrates
and basically say you're looking for

D.N.A. So D.N.A.
will hybridise to its compliment.

You'll spot in different
regions different sequences.

When you get a chip from a vendor you know
what sequences where because there's some

sort of orientation indicator and
you have to look up map.

OK so you have positional coding.

If you want on the chip.

So you can actually put a lot of dots.

It's fairly sensitive but
it's very low throughput

the assays are rather inflexible
because essentially this chip it's.

In its design once you get it if you say
well more interested in these sequences

where you sort of stuck
with the whole chip and

in general the cost of
the SES are rather large.

And much of the world has
thought of going to particles.

So there are inherent mass transfer.

Advantages when you start thinking of
having particles now in this pollution

that you can mix and agitate while
you're trying to get the D.N.A. to find

the different particles The problem is
these particles are moving all around.

So let's say that I made five different
particles with different D.N.A.

sequences and I mix them together.

When I fish them out how do I know that
this particles different from the other

particle and
right now a company sells us and

basically put different
levels of floor force.

The problem is that you can't
distinguish many different in a robust

way levels of intensity and then you
usually trying to use fluorescents

to sense what bound to the particle so
your assets aren't very sensitive.

But it is nice because
it turns out that is

the way you handle these it's
higher throughput you can mix and

match or you can mix and
match different types of beads.

So people really like this
in terms of flexibility but

they turn out to be polystyrene
it is a pretty failing surface so

you go to a lot of arduous work to make
it non fouling and there's a I said

a limited number of codes distinguishable
particles that you can make.

So we thought of coming in and
making something that might fill the sort

of need in terms of being relatively high
throughput relatively high sensitivity

sensitivity and sort of filling
in this part of the matrix.

So gels it turns out to be or

a great thing to work with there's
a lot of noise for more work.

Actually a group in Russia which
we're looking at gel pads and

they looked at things for
example in terms of comparison of

a gel pad versus a glass slide and

affinity of things like nucleic acids and
nonspecific binding and

basically what they showed is that
you basically got close to a sloot.

Solution Association Constans
when you use gels

because Djoser often about
eighty percent water and

so this was a very logical and
advantageous thing you're still stuck with

a lot of transport limitations in
dealing just with the substrate though.

And as I showed you previously particles
are a lot nicer to think about than planar

rays because of the mix and match easy
way to skin them out using basically

flowing them and then interrogating
them with a laser and so we thought of

with our technologies sort of combining
these two together and making particles.

So what we do is this is a typical motif.

So this is a single micro particle and
we made it by flowing for streams.

CO flowing them together superimposing
masks over top of the stream and

projecting this shape.

So the shape is dictated by the mask

the different chemistry
is dictated by the flow.

So this is giving it's a bitmap so I have

with this over a million different permit
to a million permutations that I can make.

So I could in principle decipher
different million different particles.

OK so I compare that right now already.

What I told you is the state of the art of
particles and that's about one hundred so

we can really likes fairly high and

then I can put in different
sensing regions in this gel.

So this is actually after a D.N.A.
assaye and I put one sequence here.

Or actually have a building control
region which is really nice.

So every particle actually
has a building background and

then this is a different D.N.A. sequence.

And the nice thing is that you can
actually commercially by D.N.A.

So you just type whatever sequence
that you want and you can get them

with an ng group that we can actually
pull them rise into our matrix.

OK so the sensing molecules are covalent
Li bound where I want them in my particle.

The other thing is that where I do.

Sensing and read out my code or
in two different regions.

So there's no cross talk I.E.
I don't convolute the signal from

my code with the very faint signal
sometimes of what I want to sense.

So it works in principle these were very

easy things to do in terms of
the ass Ses because we bought.

D.N.A. is that we're clean.

You know we got them synthetically from
a vendor just spike them into buffer

solutions and
then we're trying to make particles that

would sense three different
sequences of D.N.A. We knew

that the particles were different because
they had different bitmaps on the side.

So we this was a control this
had one sequence of a second.

These are with the particles
looked like and

then basically we challenge this
cocktail to complimentary D.N.A.

complimentary D.N.A. all had a floor for
so if they bound to the particle

they would light up this other side of the
particle and so this matrix to shows you.

These are different
particle types laid out and

this is what I added to the solutions so
if I don't know anything.

They're all black over here.

If I add the complement to this particle

you know this guy lights
up these ones don't.

Vice versa or if I complement
to both they both light up and

this is showing qualitatively
that we you know we did produce.

Calibration.

Six cetera so
we can make this quantitative.

And if you want you can do
this on the same particle.

So you know I can instead of having three
separate particles have three different

regions and challenge again basically
all this information is now

compressed into a single particle.

So from a flexibility standpoint

if I thought about the research
setting this is very nice

because this allows flexibility I may
not always be interested in this D.N.A.

sequence versus this one but if I build
a diagnostic that's sort of set in stone.

The F.D.A. is said I should look
at these certain markers and

so why not build it all
into a single particle so

that the they have different
advantages in different settings.

But it's a little bit of you
know overselling because

I've got these really complex codes and

if I do some math and I say well
how do I read them out really fast.

You know if I got a million codes and
I got to sit there and

look at a million things you know I'm not
high throughput and that's actually you

know probably the biggest criticism I had
of my own work but also of the field.

There were a lot of people making
really like fantastic crazy looking

particles that had in principle a lot
of code information on them but

they couldn't read them out quickly.

So you need to be able to decode.

So we build the do this.

So basically we take our particles after
we've done the sensing assaye we put

them into a suck them into a microfluidic
channel basically put them in here and

the side stream is just sort of gently
focussed the particle down the street.

And so it's sort of like a tablet so

it's very easy to get it to sit flat and
then we create a laser sheet.

Much like in your supermarket.

We do a line scan of our particle but

here instead of the laser moving the
particle moves in this reference frame.

So the particle zips by and
we've now reduced actually our codes down

the one information so
we lose a little bit on the coding song.

I'd but we gain a tremendous advantage in
terms of the speed that actually we can

do this and
I'll show you that in a second.

So this is a high speed video.

You see a few particles coming down.

We put some food coloring so
you can see the side streams.

So actually the terminal velocity down
here is about half a meter per second.

So they're about you know
the width of your hair and

at the end of this channel they go about
you know nearly a meter per second which

is quite fast as comparable
to commercial cytometry.

And we can get them very
precisely aligned just using flow.

OK And the great thing is
that they're hydrogels So

the other thing we got done
on this grant was that they

said the devices were going to clog but
they don't because if a particle comes in

the wrong way it actually bends folds
up in the pops out on the other end.

OK So we actually never called these
devices because of the particles.

And so
what we do is you know if I took a camera.

I see bursts of light going by the cameras
are not very sensitive and slow so

I put basically was called
a photo multiplier too so

these are very sensitive and
fast in terms of readout So

this is a second data or less every spike
is a particle and if I blow that up.

This series of hills and
valleys give me a signature about both

the identity of the particle and
then how much is bound to it.

OK so
I basically decode this information and

we've developed software we
don't do this manually so

we decode it in two different ways and we
look for consistency and then we read off.

We have calibration curves for
these regions.

So we can actually run a whole
ass a through our scanner and

decode it almost in real time it takes
about ten extra seconds to decode.

So we know we typically we make particles

we might do some sort of essay on them and
then we scan them so it's a a.

Fairly complete solution.

But you know another criticism
is that I showed you all things

from molecules that are trivial to detect
and I was buying some synthetic D.N.A.

spiking them into you know
perfectly pure solution.

You know that is an artificial
construct for an academic but

if I gave this technology to a biologist
they'd be lacing cells you know they

have all sorts of protein and other stuff
in there take trying to use it with blood.

Except for A and then they probably
wouldn't be interested in the small D.N.A.

I was looking at but other molecules and

one molecule would become interested
in this so called micro R.N.A. So

these are nothing more as the name
implies very small and dodginess R.N.A.

He's typically between twenty one and
twenty three bases long and

they're known to be implicated in terms of
how they are expressed in our bodies and

correlated to a number of disease states.

So everything from cancer to all
timers heart disease diabetes.

You know people are starting
were working with people and

HIV many many more the beauty
about these is that they're much

much more stable than other
protein markers and the reason

is actually just recently discovered in
the last couple years is that they're

actually typically bound to another
complex in blood that actually protects

them from a lot of enzymes I would want
to see them up and break them down.

But they're known they can be
found in a basically a pretty wide

range of copy number and if you translate
this to a drop of typical fluid.

It's something like one to one
hundred atoms are the sort of numbers

you need to look at and
they turn out to be really hard to detect.

So the end of last year there
was this review article and

sort of the you know this is a great
you know technical term that is really.

It's a murky world into tech thing
these basically some people like

to do so-called northern blots on.

People like you P.C.R.
everybody has their nice technique and

that's because there's not consensus
about actually the techniques and part of

it is having to do with how you actually
do the sensing and the fact that R.N.A.

is like to form secondary structures
which can lead to all sorts of bias.

So one thing we did is we actually
then when a step forward or

with our ass A's and we built a new
construct on to the gel a way

of sensing R.N.A. where we don't amplify
the aura name we're trying to sense but

we have our hydrogen I can put any D.N.A.
I want covalent on it and

we have a region that
senses the micro R.N.A. and

then after We've fished out
the micro in a of interest.

We bring in another nucleic
acid this region is generic

on every single particle.

Regardless of the sequence up here.

So it's a way of now bringing a floor
forward to this molecule without trying

to put it physically on this molecule so
voids bias and

this region is the same for
every single micro in a sequence.

So there's no bias introduce It's also
flexible sure you modifications of it but

essentially we hybridised them join
these together so-called like gating and

then we do a detection and so
typically Here's a particle.

This is before target and then you know
we're lighting up and seeing our target.

So we actually get very high.

You know efficiency actually with
this technique and really it works.

I think very well again because
we're doing things in a hydrogen.

And we've demonstrated it on a twelve X.

for looking at twelve different
micro R.N.A. signatures and

this is just some simple data
you can buy purified R.N.A.

which is typically what everybody
uses you can get it from patients

which are known to have certain cancer and
cancer histories and.

Basically looking at different
my Carney sequences and

trying to correlate whether they have
more or less than a healthy person.

So this is sort of a fingerprinting and

the idea is that as you get more
quantitative and more fingerprints.

You can then use this as a diagnostic for

example in cancer because we're
very interested in lung cancer.

So we actually showed that we could get
the rate correlation or basically up or

down regulation on actually clinically
relevant sizes but these are still samples

which in the clinic people have to spend a
lot of tasks they basically take tissue or

blood and they have to purify it
down that just get the R.N.A. out.

OK that's not great.

So what we did is we were
little step further and

said Well you know let me think about not
doing all that R.N.A. purification but

let me also think about amplifying
what I see on a particle and so

we just build a little
extra loop on our detection

which is this we can have a Instead
of bringing in a floor for

here we can bring in something which
allows us to do a form of amplification

the so-called rolling circle amplification
it's a isothermal technique and

it's very easy to essentially replicate
this region of D.N.A. and into the night

on or in practice to add instead of one
floor for ten to the fourth floor fours

and so you know actually when we do men
on the particle we see spots now and

every spot is a single molecule it's
a single molecule of micro R.N.A.

that has been amplified so much that
we can we can count spots if we want.

And.

We can quote to take this and you know
it actually gets us into being very

sensitive so I had said sort
of one to one hundred out of.

Our interesting numbers.

You know so
we're getting very nicely in these

in these areas of diagnostics and
the real proof in the putting for us just.

Recently you know skipped
this technical is

in looking at just roll samples so
just getting serum.

Doing our athame and comparing healthy and

disease patients so we're looking for
a certain micro in a sequence.

That's known when you have prostate
cancer to be very abundant and so

this is our control patient and
this is actually or

a patient that has prostate cancer and
we can quantify this in terms of

number of spots or quantity and
you can look at it different ways.

If you want but
basically you know we get a very nice.

Differentiation between healthy and
prostate.

OK so it seems exciting that it actually
works with even really messy samples.

OK so you can envision now in a doctor's
office just needing a pinprick and

doing a diagnostic assaye for
something like cancer.

And we're trying to build all
sorts of novel assays around this.

So we can do this with the kinds
other small molecules we can make

particles around cells and
selectively lace them open and

sort of think about containers
at the research scale.

This can be interesting.

You get a high concentration
of locally confined molecules

in certain advantages and sensing.

So I think I'm just going to say

five minutes on another topic I'm
not going to so much detail but

to show you completely you know pie in the
sky ideas of what you might also do with

these technologies so that the former
stuff I talked about we have a company and

you know the you can buy
particles from them.

Excedrin disease assays we thought well
you know these things are very small and

we make them a little bit smaller and
let me make them soft.

And the idea was actually in the beginning
just to say well you know what's a pretty

small soft thing at this
like scale in his blood and

these are my my my students particles
they're fluorescent and this is his blood.

And probably if I didn't make them
fluorescent you wouldn't know

the difference.

It's interesting because
they're made out of paid.

So actually we've done.

Inflammations studies so

it doesn't seem to actually certain
cell lines we haven't been exhaustive

the blood doesn't care much about
the particles at this size and chemistry.

OK And
that doesn't cause any aggregation either.

And sort of thinking about it from just a
mechanical standpoint we thought well you

know one function of blood is to be able
to transport three small properties and so

let me just think fundamentally about you
know making things of different shapes and

sizes and having different loadings in
them delivering different things and

building microfluidic vascular networks so
create shapes have different loadings and

chemistry and understand how they
pass through very small junctions and

these junctions are not going to be for
microns.

So that's pretty much the smallest
passage properties that have to see.

So we do a lot of experiments like this.

This is an outline of the fluidic device

this particle had a outer
nominal size of eight microns.

So the only way it could get through
in this pressure driven flow

is by pretty extreme strain going on
different mission of the particle.

And we looked at basically
making a library and

thinking about different morphologies
if you want to call oids

with the design that they
had to fit in a disco.

Which was eight microns by two microns so
that's about the size of a red blood cell.

So you know.

Voiding the fact that they're
not by concave this is a mimic

with this mass this creates a disk

which looks very similar to a blood cell
this is a donut punch out the middle.

Here's a start.

Here's a ness.

And we wanted to understand you
know how how did these things flow

differently in these
artificial vascular networks.

They're all nominally the same size.

Now.

You could also vary the chemistry or in
a simplistic way we just change the amount

of pagan D.N.A. So basically we're
changing the modulus if you want.

So the more peg D.N.A.

for a given shape it's more rigid and
then every color is a different shape.

And as an engineer.

This is a compelling plot I say how much
pressure that it take for breakthrough.

So increase pressure until particle
just can make it through and

it goes through by definition and
this is a log.

Right.
So I have roughly four orders of

dynamic range with these two parameters.

And it's very sensitive to shape so
especially on this low and

you know you can see that there's
over an order of magnitude

difference between the disk and
how an Ask goes through and

the first order though it doesn't collapse
the data perfectly you can think about

the axis over which bending is
occurring and the energy you toss.

It's a lastic energy and so
if I take a disk it passes and

sort of like a pancake that folds up to
go through this small or if this up here.

If I cut out the center.

Well this center region there is no
longer a particle that has to be bent.

So it should cost less energy and
indeed it does and then this is how the X.

prefers to go in this turns out to be even
smaller amount of material bending cos

than here which is less and here and
then the S. actually twists a little bit.

That wasn't intuitive.

We still have to study those more.

So we were going on with these things
making them stealth like making them

functionalized you know we can use them
to sweep up cancer cells we do this with

toner the ideas may be useful to have
circulating sponges that will accumulate

cancer cells many things that I
think are really high in the sky at

this point and you can go back and
load them with a lot of things.

So you can go back if you want to bigger.

These are the cells all sorts and
then a part.

And with my collaborators.

Bernhard trout and other colleagues
Novartis is very interested in

putting drugs in these particles and
so we have a number of interesting

articles that are coming out one is
out now in which we've either done.

Growth of drugs via precipitation and
control polymorph inside these small

hundred yards or actually use hydrophobic
so we make an emotion with an emotion

double motion and the natural motion
contains a small molecule of interest.

So hopefully questions are things
that might not be clear but

hopefully I've shown you that you
know microfluidics with you can make

functional materials but it also can make
things which I think for just a scientist.

I'm a graduate student can be interesting
fodder for just letting your mind go

beyond thinking about a sphere which
most of the quote a world thinks about.

So I'd like to thank you for

your attention I'll be happy
to answer any questions.

Thank you.

Thank you.

Yes

So I mean it depends upon
the particular assay that we're doing.

But the sickly if you have very
fat is the kinetic right so

it's diffusion versus reaction.

So if you have a very fast.

Reaction or binding then you're going
to bind where you first see a site and

that's going to be on
the edges of your particle.

So the particles are the in plane
dimension is wider than the out

of plane dimension but what we've shown
that that region is not because of the.

It's transport.

So you can do other experiments where
you know there should be no binding and

you can quickly.

You know even very large proteins
can diffuse through the particle and

less than a second.

We've done these sort of experiments.

So it's purely because
it's a fast reaction.

So it's another boundary
layer issue right.

Well.

Yes.

I think a couple things.

So because of their flexibility.

Right.

I mean I think you can drastically change.

The both because of the shape and
the flexibility actually.

The time that they actually
stay in circulation and

actually where they where they are in
larger capillaries where they act

which actually gets to because of
interacting with the blood itself.

Right.

So it's not in the real world you know
it's not going to be just the single

particles but it's going to be in
a very heterogeneous mixture of other

like size Cluedo elements
namely the cells and

so if you wanted to interact
a lot with the vests outside.

They also were not you might think
about using the shape to tune that in

the other thing is since it's a gel.

You know I think we should novel ways
of actually loading in having you know

hydrophobic compartments.

Embedded which actually the exterior
doesn't see right so we can have very

inert peg this you know the most
unearthing you can think about and

embed it hydrophobic regions inside but

I mean even there's a really
beautiful work by out of Hopkins.

I think it's basically for pay.

I'm sorry Denish fishers work looking
at you know worms basically they create

die block.

Polymers and get them to self
assemble into like ten micron ish

structures they're very heterogeneous
it's hard to control morphology but

they have really enhanced
an interesting circulation and

they speculate it's
because of their size and

actually the way that a cell it confronts
us cell because it's flexible so.

We'll see.

Thanks.

Sure.

Yeah yeah.

Yeah so basically another way.

It's like an extra shape right.

Most of what I showed.

So we do have around that.

So we've done the interference
lithography So that's a nice way.

Thomas is an expert and
we have a collaboration with him and

then you can use all their
fluids which are inert.

OK And so that by CO flowing those and

using surface tension Excedrin we
have made concave incoming barracks.

Interfaces in the other plane.

It's more I admittedly
it's much more tedious.

So we prefer to think of
things to do with these.

So

the.

Changing away from P.M.S. will only
change our throughput by a factor of two.

OK Because basically we have
we flush a new monomer and

we have a time for the P.M.S. to relax.

And that's about like a third to
half of the total process time.

So on.

It's really we're actually designing
a very large illumination and we do so.

It's simply actually creating
the illumination spot size and

going away from conventional lamps which
are very stable and going to things like.

I think that's the one that is our path
forward for increasing throughput.

We've

only done things with like flexible rods.

And they will actually undergo some
weird dynamics where they don't stay.

It's an instability they can actually
keep going from one side to the other and

sort of like a fly in the wind.

So I don't quite understand this yet
we're trying to do some simulations.

But you know at this.

But that's for blood we have to simulate a
complex mixture and look at what happens.

But at this Ingle particle level even
a flexible filament is unstable and

channel flows in more because of the
coupling between elasticity in the hundred

interactions.

And we have done things other things.

So you can think about.

When I create the particles
It's like an initial condition.

Right.

If I was a simulator.

I might think of let me think
about a small cluster and

put them in different positions and
set them off and see what happens.

And so we do play that game too.

So we make different masks and
create a small array of particles and

we change their lattice
spacing teakwood zero.

And then turn on flow and
watch how they evolve.

So we were looking at
a couple dynamics and

there's a very interesting strong hundred
in them a couple and you get all sorts

of very vigorous It looks almost
like they're brown in and diffusing.

They've moved so much but
it's because one hundred I know mixed.

But I very interested in that the question
you're asking about where the margin and.

Yeah.

Sure.

It's all.

In principle so the real you know we've
made things down to a micron with this

you know so it right now it's limited
just by the reaction kinetic So

basically you know how sharp Can we get
between the particular digit number to

kill a good region and
these are not traditional photo versus So

you add all sorts of things to
increase that resolution so

that that's a first thing to do and
then wavelength of light because

we're not doing anything fancy here you
know using three sixty five now to meters.

The question although you know advanced
techniques even like to photon techniques.

You know should work is just they're slow.

Right.
So you know I think it's suffer

a lot on throughput you know so I haven't
invested too much energy because when I

think about it and
I think about it as an engineer.

It doesn't scale well.

And so
that's what worries me a little bit.

And so
I comfort myself that in the micron world.

There's an interesting.

Things to do and there are competing
technologies at that scale if you want

a homogeneous particle So Joe this
unknown has beautiful technologies which

kill us in that size range but I think
we kill him in the other size regions.

You know.
So we play separately in different domains

and try to exploit the advantages where
it makes sense but that's a good question

of course is the first one is

just like a lot are developed.

Just as well.

Yes So the.

So well I made a scanner in my lab we
have a different version of a particle

that works on all of
the four major brands of.

So the idea is actually that.

You just buy particles from us and use
them on whatever cytometry you want and

we've developed actually
already software patches.

So that you can read off the particles.

You know I can't name friends but
major brands of.

You can guess probably who they are and
and

use them as you want and
probably in the far future.

You know I mean I think that's where
you get traction because everybody

has a say Tom at or around.

Most people do certainly all be
companies safe homes of cytometry.

But I think for
really point if you care things

you don't want to customize
it make it cheaper scanner.

So that it will physically you know
a nurse will have it it'll be like.

You know something that plugs
into your laptop or something.

You know very small integrated
device that's on the room with

you from your own heart.

How expensive.

Is there comes over and over.

So from a cost standpoint
there's not a concern because

if we price it out you know per how much
the regions cost and what we can sell.

We use enough because we
use like fifty percent.

But if we started using it so.

But there's a lot of waste and
that's never good.

Right.

But then also we you know if we're doing.

Like if we're working with antibodies.

You know everything.

I talked about you know that's minuscule
it's point zero one percent of

the cost compared to an antibody So we
just do post conjugation of antibodies and

we can still do ways of addressing
them on specific regions.

So then if we have a really
precious molecule.

We use it very efficiently.

Right.

So it's the pounds.

Right.
But I do think from a green sense you know

we are concerned about waste and

thinking about recycling though I'm
not sure how efficient that will be.

Or worse yet again.