SMART: Shrink Manufacturing Advanced Research Tools (Michelle Khine)
Sep 5, 2023
MichelleKhine 08 06 14 youtube
Thank you, Andrew, for a really nice introduction. I am impressed you remembered all of that and it's always nice to be back at UIUC.
So I've been here quite a few times and I always love coming to visit.
And today I thought I would share some of the work we've been doing on a platform that we've developed. I call it SMART.
It's a SMART platform and it stands for Shrink Manufacturing Advanced Research Technologies.
Before I do that though, I wanted to talk a little bit about
kind of the context of, as technologists, what we should be designing for and what we should keep in mind.
And so this kind of got me started. I gave a talk in Tel Aviv a
little bit ago and the idea was on the future of medicine.
And so we had to kind of come up with our,
you know, our big ideas for how we saw the future of medicine and what tools we would need to develop to get us there.
And it got me thinking, and I'm starting a new course at
UC Irvine this winter,
as part of, we have an NSF IGERT for biophotonics, and I'm going to be teaching a class on
opportunity recognition.
So the idea of, you know, if you're starting off your careers, if you're trying to find a research project, or if you're, you know, going to
jump into a startup company, or you're going to start a company by yourself,
what are the interesting areas to tack on? What are the interesting?
And there are a lot of things that you can look at in history to kind of, and see what people have done right, and
see what were kind of the different factors that played a role in the success of certain fields versus others.
And so it got me thinking about what I think would be, and these are just my opinions, about the future of medicine and what
it will look like by 2025. And so I think
there are four main areas that we're going to see huge differences in. One is point-of-care diagnostics is going to be more pervasive.
So instead of going to hospitals, or clinics, or even your physician's office,
you're going to be able to just do diagnoses at your bedside.
So you can go to Walmart, or Kmart, and pick up, you know,
what is the equivalent of a pregnancy strip, and be able to tell if you have, you know,
the flu, or if you have something worse. And
you know, I think that there's an important need for that, as well as pervasive health monitoring,
because hospitals in the U.S. are actually the third leading cause of death in the U.S.
So you want to get people out of hospitals, and you want to keep them out of hospitals.
So I think in previous generations, when I was a student,
we talked about curing diseases and coming up with ways to diagnose diseases. Now
we're talking about monitoring health instead, and I think that that's an important paradigm shift.
Personalized medicine, so that means a lot of different things to a lot of different people. In the context of what I'm doing in my
research, it means
trying to create medicines that are appropriate to certain subpopulations, so that you don't have
toxic side effects. And so I'm going to talk a little bit about that. And finally, the pet project
I'm really excited about is, I think that instead of having, we see all of these, you know, a
democratization in education. I think we're going to be seeing that the next step to that is a democratization in
discoveries. So we see MIT has these large online classes now, so you can be, you know, a kid in Africa,
and you can take classes at MIT.
I think that we will soon be seeing that inventors will not just be coming out of the Ivy Towers and and the big,
you know, research labs, but that people will have access to be able to create and to work with other people
via social media to really, and to harness that power. So
my research, you know, these are areas of interest to me,
I think, in terms of how I'm seeing the future, but also trying to poise my research to address these areas.
And how we're doing that in my lab is, oh, this was interesting.
This was the cover of The Scientist just two days ago, three days ago.
And it says, Think Big, Go Small. And that is really the focus of my lab.
So you guys are all in a really great place here with, you know, doing bio nanotechnology is definitely a great
opportunity, and trying to figure out, you know, where you fit into this and how you can make an impact.
I think you guys are really well poised, and you guys have picked a
phenomenal summer course. So congrats to you for coming here, and you're already ahead of the game.
So I don't think I really need to do an intro to lab on a chip in this, to this group.
I know you guys went into a clean room earlier today, and you guys have heard lectures and done some microfluidics already.
So, you know, the idea of scaling everything down to the small scale is, is really, I think you guys are well,
very knowledgeable with that. The approach that we're taking in my lab that's different from the way other people are doing micro fabrication is
we're patterning at the large scale, and then shrinking everything down to the scales that we want to achieve.
And so if you guys went into the clean room earlier today,
I know you guys did some photolithography and traditional silicon wafers, and so that's, you know, really kind of the top-down gold standard of how to do
micro fabrication, right? And this is all piggybacked on the semiconductor industry.
You know, lab on a chip as a field is really piggybacked on a semiconductor industry to make these wafers, to make these
devices, and then with soft lithography with, you know, George Whitesides,
you know, PDMS, you can now
do micro molding and make many of the same copies with one silicon wafer.
So that really transformed the way lab on a chip has been done and microfluidics, and it really catalyzed the number of publications that were coming out.
But there are still some limitations to traditional top-down
manufacturing, right?
You eventually hit the limit of resolution defined by Rayleigh's criteria, right?
So approximately lambda to the wavelength of light over two, you're going to hit that limit, and to get the smaller and smaller features,
you have to go to shorter and shorter wavelengths, right?
So then you have to do E-beam lithography. You have to go to smaller and smaller features.
And that gets prohibitively expensive.
Also, PDMS as a material is not the best material to use when you're trying to mass
manufacture your point-of-care microfluidic devices. PDMS is not mass manufacturable.
There's a lot of inherent problems with
things sticking into the sticky matrix and then leaching out at later times. And when I was a graduate student, we started a company
at Berkeley
called Plexium Biosciences, and we were making these microfluidic chips.
And we were making these chips for single-cell electrophysiology,
for drug screening, and everyone liked the chips we were making, but they told us we couldn't use PDMS.
We had to move over to a hard plastic like a polyolefin or polystyrene.
And so that kind of got me thinking, and that stayed in the back of my mind.
And so what we've developed in my lab over the years is taking
shape memory polymers, so they're just sheets of
plastic. They're basically, it's shrink-wrapped film, and
depositing different materials on it. So you could do standard photolithography.
So you can put things like metals down onto the material. You can do different types of surface modification.
You can do etches into the material.
Mechanically, or with chemicals, or with plasma.
And then you shrink down to get the features and a resolution that you want.
So to demonstrate this, I thought it would be fun to watch the word shrink shrink.
So you can see that the features keep their fidelity relatively well. And so with this approach, we've really
developed, you know, things for point-of-care diagnostics, pervasive health monitoring, personalized medicine, and finally distributed
democratized discoveries. I'm going to talk about each of those today, and that's kind of how I'm setting up
my talk today. So to kind of just jump right in there, for a point-of-care diagnostics,
we want to integrate nanocomponents into microfluidics. So if you think about this, so this is a shape memory polymer.
You just saw it away, the polymer shrinks, right? And if you have, we were doing microfluidics with this, and it carries whatever is on
the plastic. Whatever you print on the plastic, it'll carry it closer together.
So you can imagine that if you have some sort of ink and the ink reflows, you can get
microfluidic channels out of this. If you have a stiff layer, a stiff material down,
right, the stiff material can't shrink. So it buckles. And if you can control the buckling well enough,
you can get very interesting features, and you can use them for a variety of different applications.
And so we looked at the scaling of the different
materials, depending on the stiffness of the thin films of metals that we're putting down. So we're talking about gold or silver.
And you can basically predict, so we took
SEMs of these, and then we took the fast Fourier transforms to figure out the size distribution
of these features in 1D and in 2D. So for 1D wrinkles,
we just clip them with binder clips so they can only shrink in one dimension. And you get these kind of really nice
uniaxial wrinkles. And these are our predicted values with the dashed lines.
And you can see that, you know, we can predict relatively well the distribution of sizes of our wrinkles.
And we can tune these
by putting down different thicknesses of metals and different stiffnesses of metals.
So you're like, well, why would you want to go through and do this?
So one application is if you want to integrate
nickel, if you want to integrate ferromagnetic materials into your microfluidic channel. So instead of, who's done a FACS here?
Fluorescence Activated Cell Sorting?
How about MAX? Has anybody done Magnetic Activated Cell Sorting? Okay, a few people.
So with MAX, right, so the idea behind this is that say you want to separate out different
DNA.
You want to separate out your
sample DNA from the rest of the milieu, or you want to be able to separate out a certain population of cells.
So you would tag your cells or your DNA with magnetic beads,
and then you would try to pull them down with an external magnet.
External magnets are not very strong, as you can imagine, so the throughput and the selectivity of
MAX generally is very low.
So people have been working on trying to integrate these into microfluidic devices and adding ferromagnetic materials to the
microfluidic chips so that you can get an enhanced
magnetic gradient
right at these sharp, small structures.
And so we put a very thin layer of
nickel down onto the structure, onto the plastic, and when we shrunk them, you can see that we created this really high surface area
nickel structures. If you look at the cross-section of it, you can see that they integrate tightly into the plastic.
So they're robustly integrated into the microfluidic channel, so they don't wash off. You can sonicate them,
they don't wash off. We did some squid analysis as well as XPS to show that we do have the
the nickel on the structures, and we looked at the
magnetization. We also did some magnetic force microscopy to show the force at the
at the tips, and we showed that we had an enhanced magnetic field gradient at the tips of these sharp structures.
So we wanted to see how well we could do in terms of making a little microfluidic device to pull down
beads, magnetic beads, mixed in with a mixture of magnetic and polystyrene beads down to a concentration of
0.001%
magnetic beads to the rest polystyrene beads, same size, one micron. And we showed that, so we tried different flow rates
because it's really a balancing act, right, between trying to pull down your beads as they float across with
sort of force of being pulled down with the shearing force of how quickly you can flow these
these beads across. And we basically showed that pretty much at all flow rates
we were getting a very nice purity and a nice enrichment. And so at the lowest percentages, we were getting a
20,000 fold enrichment, which was larger than any anybody else had had shown at the time yet.
So it's a really simple device. You basically have
plastic channels, the middle layer is is just cut out entirely, you align them,
you stick a permanent magnet on the back side of it. And when you have the permanent magnet on,
these are just little refrigerator type magnets,
you can see the fluorescence intensity increasing over time, the beads are fluorescently tagged.
And as soon as you take off the magnet, you can see that they go away. So we showed that we can
do DNA
sample, we can purify DNA, and we compared this with the gold standard,
which is the QIAGEN kit for DNA, and then we did two PCR, and we showed that we were getting a better DNA extraction
with our
nanoscale magnetic traps versus just the QIAGEN kit alone.
So
that's one application for sample prep that you would want for point-of-care diagnostics. The larger
question that we've really been tackling over the last several years is when you go out to the field,
how do we bring the,
you know,
the detection capabilities that we have in the laboratory
to the field? You know, this chip, this field of lab on a chip. My chair of my department,
Professor Abley, who's also a lab on a chip person, says, you know, we call it lab on a chip,
but right now it's more like chip in a lab. And it's really true, right?
You have this little microfluidic chip, and then if you ever go into any of our labs,
you have all these big pieces of equipment, you have, you know, $50,000 laser shining at it,
you have all these pneumatic pumps, you have all this stuff to try to
integrate with this little chip that's supposed to cost about 50 cents.
So we wanted to be able to
increase the signal to noise of
fluorescent markers so that when you go out into the field, you don't have to have a big excitation source.
You don't have to have a big laser source. You don't have to have a big detector source. So how do you do this, right?
So similar to a pregnancy test or, you know, these point-of-care diagnostic devices,
we wanted to be able to increase the
capabilities of fluorescence so that we can
be able to detect infectious diseases
at lower and lower concentration. So if you look at a typical array, right?
If you're diluting the fluorescence detection antibody on one axis, and you're diluting the
primary antibody on the other, there's always going to be a quadrant there that's going to be
and so people have been trying to improve this signal to noise, if you will, and one way to do that is
to leverage
surface plasmons. So
propagating surface plasmons are just electromagnetic waves that
are at the interface between a metal dielectric, right? So if you shine an incident
light on this, you have these waves that propagate, which aren't terribly useful.
But
if you localize them, so if you have nanostructures that are smaller
than the wavelength of light that you're in, that then there's incident on it, what you can do is enhance your
electromagnetic field around those points and you can
interact with the dipole in a fluorescent molecule so that you can get really, really enlarged
signals in your fluorophore. And you can do this with other types of molecules. There are various types of
detection modalities. You've heard of CIRS, surface enhanced Raman spectroscopy.
So all this is based on leveraging these surface plasmons. There's also localized
LSPR. There's a variety of different ways to leverage these surface plasmons.
So what we wanted to do was just to increase the intensity of these fluorophores by
a phenomenon known as metal enhanced fluorescence.
And so the idea behind this is that you would increase the fluorescence intensity of the die,
but something has to give, right? So you get this increase in the fluorescence intensity,
but you also get a huge decrease in the lifetime of the fluorophore. And that's actually a good thing because the fluorophore
will go into its excited state and drop back down more quickly. So you actually have a more photo-stable molecule.
So it doesn't photobleach over time as quickly.
And so we didn't invent this. This is something that's been around for a while. People have been doing this with
roughened metal surfaces. So this is
basically just silver islands.
So you can sputter coat or sandpaper down silver. And people see very modest increases in the fluorescence intensity on an order of
5 to 50 fold. You can also create very precise little
antennas. So
bowtie antennas. And right in between these gaps, you can see huge increases in fluorescence intensity.
So you can think of it sort of like a lightning rod effect, right?
So you have these enhanced electromagnetic fields there. And when you put a molecule in there,
that molecule is interacting with that enhanced electromagnetic field.
So what we wanted to do is we wanted to bridge
this gap between having something that was extremely sensitive, and these features are very expensive and very difficult to manufacture.
Versus something that is very cost effective, but still will give you these very large enhancements.
And so leveraging what we've been doing with the metals on these shrink wrap films, we
used these wrinkles. So this was
we put down a combination
of nickel and gold. And we looked at we looked in a far infrared. So this is two photon excitation.
And we got down to single molecule detection of
IgG tagged with a fluorophore with FITC. So what you can see is that this is
fluorescence correlation spectroscopy. And you can see so this is going down in time and this is spatially. You can see that
every so often you see this big blip
in the green. And so the green is on our structures. And you can see that that's a molecule, a single molecule actually going by.
And we've analyzed this and we looked in the time domain as well as in the space domain to see that these are actually
individual molecules floating by. So we can pick up individual fluorescently tagged
IgG molecules. And you can see that the pixel count for the fluorescence intensity is much larger for our structures versus on the controls
which is on the glass slide.
And so we wanted to make sure that this was indeed a metal enhanced fluorescence effect.
So I said that there were two kind of parameters for that. One is an increased fluorescence intensity. The other is a decrease in a lifetime.
And so we did lifetime measurements as well.
And you can see that this is what the fluorophore would look like on just a glass slide typical about
a 2.5 nanosecond lifetime. And then on our structures you see this really decreased lifetime.
And so we recently published this
in Optics Materials Express and it kind of had
the press had a field day with it using
So we got a lot of write-ups in different magazines about shrink wrap use using shrink wrap to detect
single molecules.
This obviously has its limitations, right?
I'm talking about being able to pick up single molecules and what we're doing is we're letting that molecule diffuse across
these structures. And so every so often you would see a blip, right?
And so this isn't and this is two photon excitation, right?
So the setup is still very expensive and it's a near field effect.
So it's not a homogenous. It's not that everything that comes near the surface gets lit up.
The metal enhanced fluorescence effect is a near surface effect.
So it drops off after about 40 nanometers from the surface. It has to be really really close to those nanostructures.
So this wasn't really quite what we were looking for.
We wanted something that was more robust and that would give us a far field effect.
And so we started playing around with this idea that, you know,
we kept looking at this film and we kept thinking, well, you have this thing and you saw how much it shrunk.
It shrunk by 95% in surface area, right?
So that means that whatever you put down on the surface would concentrate by 20 times, right?
So you put something down, you'd bring everything together and you'd have basically in this little footprint what you had all over the surface.
So we thought, well, a 20-fold increase in the intensity wouldn't be so bad. But how can we improve that even more?
So we started playing around with silica, which is SiO2. People typically use it as a dielectric. It's a glassy material.
And we evaporated it down and now we've come up with different ways to put down large areas of silica.
And then we,
optimists treated it and we did a biotin-striptavitin linkage to the surface and we fluorescently tagged it.
And this is what the structures look like.
Depending on how long we evaporated the silica for, you can see that it changes the architecture.
An interesting thing with this was that if you just look at the glass, a glass slide
with the molecule, the fluorescently tagged biotin-striptavitin, you see that you can barely see one spot on there, right?
After you shrink it, if there's no
silica on the plastic, so this is just the plastic, you can see that you get
an increase in the total number of spots on there. So in the same field of view, instead of one spot,
you can see all 16 spots.
And you can get approximately a 20-fold increase in the fluorescent signal.
But on our silica structures, what we noticed is that we were getting over a hundredfold increase.
And so this was very interesting because then we can get down to lower limits of detection than you otherwise could
without
using these structures. And so Sophia, my grad student Sophia, just got the cover of Advanced Optical Materials with us.
And we wanted to see, well, what can we do with this now?
So then we started playing around with DNA, and we wanted to see how low a concentration of DNA,
so we want to do DNA hybridization on this, and we wanted to vary the concentration of DNA
to see how low we can go, all the way down to one percent DNA.
And then we put a fluorescently tagged complementary strand of DNA in.
And we compared this against
a glass slide. So here is the same, we actually,
here is the same, you can see six spots here, and you can see
a lot more spots on a shrunk piece for the same field of view. And you can see that
with,
so the red line here and the blue line here are the limits of detection, which is
three times the background plus the standard deviation.
And you can see that we can get to a much lower limit of detection on our silica structures than we can on
a glass slide. So we can get down to
0.5 nanomolars.
So
that we thought would be pretty interesting for point of care, because one, it's robust, right?
So you're not waiting to try to hit a hot spot. The whole
area is enhanced. And two, this is a far field effect.
So you don't need any specialized equipment to be able to do the excitation or the detection.
Jumping gears a little bit, wake you guys up. This project is a little bit more fun, I think.
What else can you do with these structures?
So these are these kind of really rough structures. And one of the problems that my grad student noticed
was he's like, I could get much better results if I could only wet the surface.
I'm like, well, what do you mean by that? So every time I try to pipette down my solution,
it won't drop onto the surface because it's so hydrophobic. I said, well,
you know, I think that's not necessarily a bad thing, right? If we can make really nice, robust, super hydrophobic surfaces,
we probably would come up with a ton of applications for that.
So with these surfaces, what we did was he was making all of these super hydrophobic surfaces in his metals.
And I said, well, you know, we don't really need them in metals. What we need them is to be able to
replicate them into different materials. If we can make different materials super hydrophobic, then I think that would be interesting.
So my grad student, Jolie, started imprinting these structures into basically every type of plastic she could find in the lab.
So if you left your lunch bag around, it got imprinted.
And she started, so then we took some, so this is what the super hydrophobic surface looks like.
So this is us trying to wet down the surface.
Do you see it isn't really easy taking those contact angle measurements because
it relies on a premise that you can actually wet the surface before you can take the contact angle measurements.
And so this goes on and we can't really wet it.
But we can do this in a type of, in a variety of different types of plastics.
We can do this in pretty much every plastic that we tried.
And so super hydrophobicity is defined as having a contact angle of greater than 150 degrees and a sliding angle.
So this is the angle, so the contact angle is the angle that it makes with the surface.
So you can see it stays kind of as a ball.
And then the sliding angle is if you tilt it more than a certain number of degrees,
how quickly will it start rolling off your surface? So it tilts, as soon as you tilt it more than 10 degrees,
it's considered super hydrophobic. If you can tilt it more than 10 degrees and it starts rolling off less than 10 degrees, sorry.
And so we did this and then we wanted to come up with an application for it.
We said, well, if it's super hydrophobic and it doesn't wet, then things shouldn't grow on it, right?
So we tried to, we drew E. coli on it and then we cultured the E. coli and we saw that on the super hydrophobic surfaces,
these were just, with the structures in polystyrene, polycarbonate and polyethylene,
nothing stuck to it and nothing grew in it. But on the flat equivalent plastics, we got E. coli colonies growing.
So we thought that this was interesting for a couple of different reasons. One,
we weren't doing any chemical modification to the surface. A lot of people have created super hydrophobic surfaces.
You probably have seen the Never Wet commercials.
But it relies on the fact that this is, there's a chemical on it, right?
So if you don't want to have to go through FDA approvals, if you want just a material that can go,
you know, that can circumvent all that, you use existing materials,
we can do that with our structures. This is just a structural modification to existing plastic.
And so we thought that this would have applications for doing, making
surfaces hydrophobic, super hydrophobic to saliva and blood and it's shown to have been super hydrophobic to these body fluids.
So we thought that there would be huge interest in that. It turns out the biggest interest so far has come from the food industry.
So we've been contacted by a ton of food companies because they want this for
wrappings for their foods.
Um,
so, so that's kind of fun. So it took us in a completely different
arena than we were planning on going in and now we're doing some food science in my lab.
But
we're also doing some diagnostics with this. So we thought a fun thing to do would be is
if we tried evaporating on these super hydrophobic surfaces, because we noticed that these surfaces don't get dirty, right?
So the things don't stick to the surface on a truly super hydrophobic surface.
So what happens if you just let your solution evaporate on the surface over time, right?
The droplet actually, if you watched a droplet over time,
it gets smaller and smaller and the contact angle that it, the contact length that it makes with the surface
actually comes in also. So you can imagine that you have solutes in your droplet.
It'll be concentrating it down to a small little size.
And so we did this and we were getting concentrations over 160x. So we looked at
BSA in urine for this, which is a prognosis, a diagnosis for various types of
diseases. We were looking at preeclampsia in pregnant women and we could detect preeclampsia levels of BSA in
in urine samples with this approach.
Oops.
So in terms of point of care, we also have been using these surfaces.
So we talked about nickel, we talked about nickel gold for the two-photon excitation.
We talked about these structures purely in plastic for
for super hydrophobic surfaces, but we can also, these also make great electrodes, right?
You have these really high surface area electrodes that can be put into plastic and you saw that they were
robustly integrated into the plastic.
So you can make really low cost, high surface area electrodes for point of care applications that integrate into your microfluidic channels.
And so we demonstrated that we can make these electrodes
with
this was gold and we did a cyclic voltometry. You can see this was
electrochemiluminescence and we looked at basically its response
by applying different voltages.
The neat thing about these metals, so I didn't really talk about characterizing what these electrodes look like. So if you look at the
plastic, the metal on the unshrunk plastic, it looks like this.
And if you've ever just tried to sputter coat or deposit metal down onto plastic, it wipes right off, right?
There's very bad adhesion and people have tried different ways to adhere to metal better to plastics. But after you shrink it,
it's actually nicely robustly integrated and the resistivity actually goes up.
Sorry, the resistivity goes down and it matches that of bulk gold after it's shrunk.
So these are four point probe measurements at different distances between these electrode pads and you can see that
the resistivity of bulk gold is here and
on the
on the unshrunk it's up here, but after it's shrunk it actually matches that of bulk gold quite nicely. So you have these really nice
electrically conducting thin films of metal integrated into your plastic.
So what can you do with this? Well, we thought well this might make some interesting strain gauges.
And so we started looking at these as strain gauges. We figured out how to transfer these
these metal, these electrodes into other types of more flexible materials.
And so you can see, let's see.
I hope this video plays. So you can see how stretchy these electrodes are.
So
and then you can see that we can take
so we put my student John put this on his finger and you can just
watch him as he bends his finger and you can see the the change in resistance.
So
we have about 20 minutes, 25 minutes left.
Okay, so I can talk a little bit about what we're also doing in terms of personalized medicine.
We have a stem cell project going on using our structures, but I see a lot of people getting sleepy. I'm falling asleep.
So I can just
jump to the last part that I'm interested in talking to you guys about it and open this up for question and answers and
make it a little bit more interactive or who wants to hear the stem cell stuff?
I guess if you guys really want, oh, you do want to hear the stem cell stuff. All right.
All right. All right. Keep going. Forging ahead then. All right.
So in terms of personalized medicine, what we're working on in terms of personalized medicine
it's usually when I teach class and people fall asleep. I throw candy at their head.
I don't have any candy with me. So we've been trying to develop a biomedic
approaches to improve
functional heart tissue. So we're taking
pluripotent stem cells. So these are human embryonic derived as well as induced pluripotent stem cells.
And we're trying to turn them into heart cells and people have figured out how to do this quite efficiently now.
So you can get relatively high yields of cardiomyocytes of heart cells. The problem with cardiomyocytes though
is that these all the ones divide from pluripotent stem cells whether it's induced from
what it's from induced pluripotent stem cells or from embryonic stem cells.
You guys all know what iPS cells are? Induced pluripotent stem cells? Yes. Okay.
So the fundamental
blockage point right now is the fact that these cells
have an immature phenotype. So they have a fetal-like phenotype, which means that they're smaller in size,
they lack mature structures and properties, and they don't have the contraction force order action potentials that
their equivalent mature cardiomyocytes have. So people have been trying to figure out ways to maturate these cardiomyocytes. People have put, you know,
have actually put in different ion channels into the cells,
but you don't really want, if you want to be using these for
to put into humans, you don't want to be transfecting these cells with
with ion channels. And so there have been some papers showing that nanoscale cues regulate the structure and function
of macroscopic cardiac tissue. So this was the approach that my lab took. You saw these wrinkles before.
And so we were using these wrinkles and we figured out that we can just plasma treat the plastic.
We don't have to put down a metal. You just need a stiffer film.
So we can oxidize the surface to create that stiff film and shrink it and you can create these nano grooves
and we can put down our cells, our cardiomyocytes onto these nano grooves,
and we can watch them and we can see if they mature over time to be more mature cardiomyocytes.
So the idea is to treat them like they're mature heart tissue and see if they actually act more like mature heart tissue.
And so you can see them beading here.
And we were growing them and we were watching them bead and we were taking different measurements, optical flow measurements, and then we
we realized that we can actually get a lot of information out of these optical flow measurements and I
So I don't know if these, my computer crashed right before we started and these two videos are not going to apply.
But we basically did some optical flow analysis. So basically we assigned vectors
to the displacement of each of these cells as they moved and from that we backed out interesting characteristics of these
cells as they were beading. And so we looked at the duration of their beading, their frequency,
and we got back out a lot of information.
And so what we started doing was we started doing drug studies for looking at cardiotoxicity.
So we applied different drug compounds to them and we looked to see if we can actually pick up
physiologically relevant information about these heart tissues that would predict
you know, the cells actually responding in adverse ways to these different drugs.
And so we did this with a variety of different drugs. I think this was varapamil isopraniline.
Varapamil isopraniline for our first paper.
We just we published that in biomaterials.
We also compared this against different things that we were doing in terms of actually labeling the cells doing action potential
propagation measurements.
And we realized that we were actually getting quite a bit of information by not fluorescently tagging the cells at all.
By just doing optical flow measurements.
These are action potential measurements. So
so what we're doing now is instead of
pre-assigning. So what we did in this first paper was we decided we're going to look at a couple of different things.
We were going to look at action potential and with the optical flow actually look to see how the action potential changes the sorry the
the duration of the beating changes over time. And now we're incorporating some machine learning algorithms.
So we're just running the optical flow. We're taking videos of these things.
Then we're feeding them into our machine learning algorithm and we're picking out differences
from the drugged versus the undrugged cells. And we're seeing that we can actually do as well as we have
a GFP tagged reporter cell line and we're showing so far that we have the equivalent
sensitivity of not even a little bit improved sensitivity to the fluorescently tagged lines.
So in the very last part of my talk, I want to talk a little bit about what I mean by democratized distributed discoveries.
So
when I first started my academic career, I started at UC Merced. I know some of you guys were at UC Merced.
I started there before a lot of you guys. I was there in
2006. And you know, I when I was being recruited there,
I was told that we were going to get a clean room and
and you know, everything that I did was based off of having a clean room.
And obviously we didn't have a clean room. This is what campus looked like when I got there.
And
how to keep my research alive, right? As a junior faculty, it's publish or perish.
And so after banging my head and staring at the cameras for a few months,
I went back to
my favorite children's story. I went back to the basics. You guys know what trinketings are?
So those of you who don't, I don't know if this video is going to play. This video is not going to play.
But they're basically pre-stressed polystyrene sheets.
And you heat them up and you stick them in the oven for a few minutes and they shrink down to a
rectangular size. And people, kids make little decorations out of this.
And so this was about 2007, 2008 when I
figured this out. And I didn't have any grad students at the time. My undergrad was the first author on the paper.
And my colleagues warned me that this could be career suicide if I tried to publish this because I wanted to call it
Trinketink Microfluidics. So I submitted it to the best journal in my field.
And the paper had gone, went viral within days.
And I got called from the editor of
Lab on a Chip.
He was wondering, we had more downloads than all the editorial society of chemistry journals.
And the
president of Trinketinks called me. She was wondering why labs were buying boxes and boxes of Trinketinks.
So that really kind of got me started on doing this. And you know, fast forward. So this was
I think we published that first paper in 2008.
And this year my lab decided we were going to do a review paper to see how many people have actually
picked up on this technology. And we had over 150 references.
So it's really exciting to see that this little thing that we started, you know, people are using.
This is actually from Professor Nam's group here at UCI. And he figured that he can densify
silicon nanowires with this shrink approach, right? And so people are using this. This one is for to improve
solar cell concentration. So there's a lot of different, you know, areas that people are going into with this approach of thermally
induced miniaturization using shrink film.
And so
that was really exciting to me. And I think that as an academic, as a researcher,
that's part of the joy of doing what we do, right? You come up with an idea and you can disseminate that.
And being in nanotechnology, you guys all know who this is, right?
Nobody?
A little history lesson. So that's Richard Feynman. In 1959, he gave a talk at Caltech called There's Plenty of Room at the Bottom.
And he talked about the coming of nanotechnology.
He said, what if you can make something that can make something smaller that can make something smaller? How small could you go?
And so I love, he's like, imagine the power of 100 tiny hands, right?
And I love that quote. And I tried, I did exactly that. We made a Shrinky Dinks mask
and we shined UV light through it to a smaller sheet and we made micro lenses out of this. And we reduced,
we got down to a 99% reduction in size using the sequential shrink approach.
And so I submitted this paper to Applied Physics Letter. I was very proud of this. I got sequential shrink colon
100 tiny, 100 tiny hands colon sequential shrink for micro lens
fabrication. And this was the only time it ever happened to me, but I got exactly one required modification to paper.
We had to change the title. I couldn't use 100 tiny hands.
But I really love that quote. And so it stuck in my head and
I decided that I was going to do something with this title.
And
the idea of 100 tiny hands working together was really powerful to me.
And so naturally, because all of my research is based off of a children's toy,
I get asked to give a lot of outreach talks to little kids, right? So
we decided to start
a cooperative in my lab called 100 tiny hands. And the idea behind us was to democratize science a bit and to disseminate it
so that kids at any age can become inventors. So we wanted to invent inventors.
And the idea behind this was we were going to start with some of the technologies we had developed in my lab, including
developing lenses that kids can, so each of the inventors were my students.
So I asked my students who were doing well and already published in my lab who wanted to work on this.
And so this is Jolie, and this is cartoon Jolie.
And she is working with a high school student to make these
microlenses, to make these lenses that strap onto your cell phone or iPad. And this is before the lens.
This is, sorry, this is no lens.
This is without the lens as well. And this is with the lens. So kids make their own lenses and they can image things
up close. This is a fun little ladybug video that one of the kids took.
And we wanted to create this community.
So these kids would, you know, they can get boxes with these different kits.
But we wanted to encourage them to create their own kits. And we wanted to create a social media platform
so kids can upload their own ideas. And we would help them manufacture and sell their best kits, the best ideas.
And the kids would become the inventors and would get the revenues from the profits from selling these kits.
So we launched our very, UC Irvine's very first Kickstarter campaign in March.
And we raised more than our goal. So we had some seed money to do this.
And so with the super hydrotension kit, so this is the super hydrophobic surfaces. This is another kit.
This is the kit that Sophia created.
The idea behind this is that we would give kids these surfaces to build things with so that they can make mazes and games.
And discover things like, you know, about surface tension and surface science.
And
so each one of these kits, we just, I'm very happy with this. This is our very first comic book.
We have little storybook comics
that teaches them about the different sciences.
And it comes with the kits. So
pass that around.
So these are little cartoon characters that come with the boxes. And what we're trying to do is to not only increase,
you know, kids enjoyment and engagement with STEM, but we really want to increase the the pipeline of
PhDs in engineering. So two-thirds of all PhDs graduating in engineering
in the U.S. are not U.S. citizens. Only 18 percent of PhDs in engineering are women.
These are some pretty alarming numbers to me when I started discovering this.
So we wanted to figure out how we can get more kids
interested and retained in STEM. And so we're trying to make, you know, education fun and interactive.
And more like the way we do science in the lab in graduate school than the way they teach science at the elementary school level.
So,
so this is my team and all of our wonderful collaborators and some of the funding we've had over the years.
And I know a bunch of you guys are interested in the Hawaii conference that I'm organizing with
Professor Rashid Bashir and
and
Ali Khadem-Husseini from Harvard. So the three of us organize this conference every other year and we're doing that this December.
It's going to be in Oahu on the North Shore at the Turtle Bay Resort. And
we are still, we've got a fantastic lineup. Go to the website. We've got a fantastic lineup of speakers.
It's a really fun week-long conference. It's five days.
But it's kind of Gordon style. So everyone sticks around and just, you know, has a really good time.
So we are still accepting abstracts. So please consider coming and that's it.
So we, you know those structures with the metals, when the metal shrinks it creates all those wrinkles.
So we basically just emboss those structures into plastics. Yeah, so it's um,
you know, you guys have heard of the lotus leaf effect.
So this is the idea that it's the Cassie-Baxter model of superhydrophobicity.
So the idea is that you have a rough enough surface so that you always have
air pockets trapped in between. So when you drop a water down by surface tension, it rather sticks to itself than to wet out the surface.
Yeah, so what we would like to do is, you know,
we're kind of open as to how we do this. We're trying to get them into schools. We're working with Girl Scouts now.
We're trying to get more kids involved. So, you know, outreach programs. We do a ton of outreach programs in my lab.
But yeah, definitely.
So what sets the limit of how much something can be shrunk? It's how much the film is extruded in the first place.
So these are
polyolefins are blown out. So they're pulled, like, stretched out into, it's like a rubber band basically.
You stretch it out and then freeze it in that configuration. So it'll shrink back to its relaxed state.
So however far you can extrude it. And it's also a function of the molecular length of the polymer also. So kind of the limits of that.
Yeah, so we can, I don't think we've hit the limit on how small we can go.
We've done some nanosphere lithography and then we've etched in using that as a mask. And we've gotten down to sub
50 nanometers in terms of holes that we can make. So I think we can push smaller. I think that,
yeah, I don't know what, I think what's interesting to me is integrating with other types of interesting materials.
I know Professor Nam is, you know, putting his nanowires down and graphene.
And I think that there's a lot of interesting
applications in terms of putting dissimilar materials onto the shrink film and using that as a process to combine different materials.
So as a process step. And that's kind of what we're doing with our sensors now.
So we can put dissimilar types of materials into the films.
No, once you shrink it, yeah, it's a one-way street, unfortunately.
Yeah.
Yeah, so basically we're, I mean, it's a very simple
regression model that we're doing right now, but it seems to be working relatively well.
So basically we're taking the videos and we know what is what we're calling the healthy undrug
patterns of contractility look like. So we map those out and then we have different drugs, drug studies.
So we do dilutions of various types of drugs that would affect different
different parts of the heart cells. So one looks at contractibility, another one looks at calcium channels,
calcium transport. So
we, and then we look at the differences in the contraction and then we basically bend them to see if
it's being affected or not. And it's
Yes. Yeah, exactly.
So it's used to, I mean, it's based on the experimental result and extracting the features from the image and recognizing the pattern you are targeting.
Yeah, yeah, exactly. And it seems relatively good in being able to predict. And we give it an unknown
response, right, but we know, and to test it. And it seems like it's pretty good in terms of its prediction capability so far.
So you said the material is almost as straight because it's stretched from the original. So the original materials just have wrinkles, just like a flat piece of plastic and you have to stretch it out.
So it's shrink-wrapped film, if you guys have ever played with. So we actually back this onto a thicker plastic.
I don't know if I can get it off. It's a very thin, it's one mil, so it's 24 microns thick.
And it's
backed onto a thicker film. And when you shrink it, it just comes right off. So we actually, I don't know if you remember going to like
like Costco and your vegetables are wrapped in, you know, the cellophane-like thing. So that's basically what the film is. So you, the manufacturers actually make them in bulk and they have very poor tolerances over the shrinkage.
But we're working with a couple of different converters to tighten up the tolerances on how much to extrude them.
Because these are very inexpensive roll-to-roll mass manufacturing plastics. And basically you can see how thin it is.
So this is the shrink film. And that's just the adhesive that, that we back it onto so we can handle it more easily.
Yeah, yeah, they do. So yeah, so the polystyrene typically the maximum you can stretch it by is, it shrinks by about 60% because that's pretty much the limit of how much you can extrude it out by. So the polyolefins tend to shrink a lot more.
So that's why we've converted over and started moving to some polyolefins.
Yeah, I believe so.
It gets taller, which is very interesting. So, so that's a great point actually that I didn't mention, but it's interesting for fabrication because when it shrinks in the X, it grows by X squared, right?
Yeah, yeah.
I have a question about your sensors that you talked about at the beginning of the talk. So they're mostly based on fluorescence. Is that the case?
So the first, the metal has fluorescence and the far-field silica stuff is, and then the string gauges are, yeah, string gauges, so motion sensors.
So you made a good point that it's, you're using these devices in a lab that have really expensive equipment. How do you envision using fluorescence, like, as a point of care device?
Yeah, so we're moving over. So we're trying to do the same enhancements with chemiluminescence now, and then that's why we moved to electrochemiluminescence.
I think just even being able to do all this with a lower-powered laser and a less sensitive detection setup would already help a bit, but I do definitely want to move away from fluorescence eventually.

