microTAStic: Joseph Rufo on acoustofluidic EV isolation; Alice Zhang on self-powered smart systems

My

talk is entitled Floating to New Heights, a Nanoscale Acoustopholitic Technique for Isolating Extracellular Vesicles. This is meant to just be a short talk, but I want to give a brief primer on acoustopholitics in general, and then talk about some of my work on extending acoustopholitics to the nanoscale for isolating EVs.

So, just

a little bit about myself. I'm originally from Philadelphia, Pennsylvania. I'm a huge football fan. Go Eagles! I did my undergrad at Penn State University in central Pennsylvania, and I had to give them a shout out because they're currently playing in the

college

football semifinals, so I'll be watching that. It's actually

going

on right now, so

very

exciting. Then I moved down to North Carolina and did my PhD at Duke University in Dr. Tony Huang's lab, the Duke Acoustopholitics Lab. And

after

graduating from Duke, I have since moved across the country to Santa Barbara,

where, as

was mentioned earlier, I'm currently working with a stem cell biologist, Max Wilson, using some acoustopholitic techniques to study stem cell biology.

But

the focus of my talk is going to be about acoustopholitics in general and the work of the Duke Acoustopholitics Lab. So, for those of you unfamiliar with acoustopholitics,

in

its most simple definition, it can just be described as the integration of acoustics and microfluidics, or devices which manipulate fluids at submillimeter

length

scales. One of the most famous

is

hypercapillary, and it flows across a lateral flow assay and can detect the presence of certain analytes, whether it be antibodies. Pregnancy tests are also another very commonly used lateral flow assay. But I think a lot of people here, microfluidics might not necessarily know how familiar they actually are with them.

But there

are two general approaches for generating acoustics waves to integrate acoustic devices with these microfluidic devices. The first being bulk acoustic waves, which if you've ever bought a piezo transducer, these are the acoustic transducers that people are most familiar with, where you can see it vibrating in thickness mode or shear mode, most commonly in thickness mode. But this is where you can actually see the transducer vibrating, and it can be at audible or ultrasonic frequencies.

And

then on the right, there's another mechanism called surface acoustic waves, where you actually deposit metal electrodes called IDTs, shown here on the right,

onto

the surface of a piezoelectric substrate.

When

you apply an alternating current to these electrodes, the surface of the substrate will deform. And if your kind of input alternating frequency is matched to the wavelength of your electrodes, you can generate these surface acoustic waves, which propagate along the surface of the piezoelectric substrate.

In

the Duke Acoustophluidics Lab, we

work

a lot with both types of devices. And shown here on the left is a typical bulk acoustic wave device, where that bulk transducer is actually bonded to the bottom of a microfluidic chamber.

And on

the right, you see a

surface because for

dealing with nanoscale particles, which I'll get into a little bit later, but you can generate much higher frequency acoustic waves with surface acoustic wave transducers, which are much better for manipulating smaller particles as opposed to bulk acoustic wave transducers.

And

what is the kind of reason for that?

So

the general operating mechanism of most acoustophluidic devices is to generate a wave within a microfluidic channel. And when you have a standing acoustic wave, you have pressure nodes and pressure antinodes.

And

if you look at the kind of force balance on particles within this channel,

they

will be directed by an acoustic radiation force towards the pressure node.

This

acoustic radiation force is governed by this FR equation down here, which is a function of the particle volume and inversely proportional to the wavelength.

So

as we get to smaller and smaller wavelengths or higher and higher frequencies, we actually increase the acoustic radiation force. It's also a function of this acoustic contrast factor, which is shown here, which is a function of the compressibility and density of the fluid and the medium and the difference between the two.

In

general, the

force

opposing the acoustic radiation force is known as a viscous drag force,

which

is proportional with the radius of the particle.

All

this is to say is that, you know, we have this basic mechanism where larger particles experience a larger acoustic radiation force than smaller particles.

So

they'll migrate to the pressure node faster than smaller particles,

which

makes it really easy to perform size-based separation.

One

of the things that the Duke Acoustophluidics Lab has pioneered is also exploiting differences in density and compressibility and also very small differences in the sizes of the particles. And what do they use that for? One of the first applications that it was used for was for isolating circulating biomarkers. So circulating tumor cells are very rare cell types that are released from the primary tumor into circulation, and the thinking is that for kind of non-invasive diagnosis or early diagnosis, if you can isolate these circulating tumor cells with high purity, you can get a really

better

insights into early stage cancer progression.

The

issue with it is they are extremely rare. So there's only 1 to 10 cells per 10 milliliters of blood in most cancer patients.

And this

compares with the 1 million white blood cells per milliliter of blood and 1 billion red blood cells per milliliter.

So

you're really trying to find 1 to 10

per

billion cells, which makes it really difficult to isolate.

In

addition to that, they're fairly heterogeneous. There's a lack of highly specific markers for CTCs. They're generally larger in size and have a lower compressibility and lower density. They also are some membrane proteins that you can use for isolating them. But again, they're not highly specific.

When

initially developed, these acoustophilic devices were mainly used for manipulating cells. But we wanted to shift to the nanoscale to take advantage of these extracellular vesicles that are being investigated for their use as a circulating biomarker. So small extracellular vesicles, SEVs, sometimes called exosomes, are 30 to 150 nanometers in diameter. And they contain different nucleic acids and proteins and biomolecules from their cell of origin.

They

participate in cell-to-cell communication. And they're highly abundant. So there are 10 to the 7th to 10 to the 9th EVs per mil of blood in comparison to other circulating biomarkers, like circulating tumor cells, which only have about 0.1 to 1 circulating tumor cell per mil of blood.

In

addition to that, they're found in all biofluids and are being extensively looked at as biomarkers for neurodegenerative, autoimmune, and a variety of different tissue-specific diseases.

But

in kind of

moving

this from the microscale to the nanoscale, that was kind of my goal when I first joined the lab. And, you know, I obviously know that there are 50 nanometers in diameter, but I didn't have a good intuition of how much smaller that actually is than a cell. It's 400 times smaller, but I always like to come up with analogies to really help myself visualize this. So if you took a basketball, which is about 24 centimeters in diameter, and made it 400 times smaller,

it

would be about the size of a grain of brown sugar. So you can imagine the difficulties that would be involved in trying to develop a technology that can manipulate with high precision a basketball and a grain of brown sugar.

In

kind of coming up with the parameters, you know, under identical conditions by this force balance, a 100 nanometer particle will experience 1 million times less force than a 10 micron particle. So how can we increase the acoustic radiation force? There are two primary approaches that we take to do this.

One

is increasing the pressure amplitude, which is a function of the input voltage that we apply to the electrodes. But there's actually a maximum kind of value that we can achieve here, because our piezoelectric substrate eventually cracks when the amplitude gets too high.

The

other thing that we can do is, through nanofabrication techniques, fabricate very, very thin transducers to decrease the wavelength of the acoustic waves. And

right

around the time I was joining the Duke Acoustophiletics Lab, they had published this very interesting paper about isolating exosomes directly from whole blood using a two-unit kind of sequential filtration strategy, where the first unit, which operates at 20 megahertz, removes all particles larger than one micron, so all red blood cells, white blood cells, and platelets. And then the second unit separates out these larger apoptotic bodies and microvesicles from the smaller exosomes. And this technology was very powerful, very impressive.

We

can do it, you know, compared with, like, the traditional techniques, like ultracentrification. It's much more automated. It's much faster.

But

one of the primary limitations is that when we isolate these exosomes, or EVs, there are still smaller proteins mixed in with that sample, which, depending on the application, may or may not be okay.

So

what I wanted to do was come up with a way to separate the EVs from those smaller contaminating proteins. What I came up with is this technique called flocculation via orbital acoustic trapping, or float,

which

uses a thermoresponsive polymer combined with these high-frequency surface acoustic wave transducers. And our devices shown here, we actually create a mini-droplet centrifuge. So the focal region of each of these transducers are slightly offset, so that when we apply a voltage and turn these transducers on, this droplet in the center will start rotating very fast.

The

specific application that we were looking at is to isolate EVs from urine samples.

And

the reason for that is because of the presence of this THP, or TAM horsefall protein, in urine. It's the most abundant urinary protein. And under standard centrifugation protocols,

what

this protein does is actually forms like a mesh-like network that actually precipitates out of solution and traps a lot of the EVs with it that end up getting discarded as waste in the intermediate centrifugation steps. But, you know, you ultimately want to recover as many of these EVs as possible.

So,

I mentioned that we use a thermoresponsive polymer. We tried to use NIPAM,

which

is a temperature-responsive polymer that's highly biocompatible and used as a flocculant for other applications like mineral recovery.

And

its basic mechanism

is

that under this lower critical solution temperature, or LCST, of 37 degrees Celsius,

it

is very hydrophilic.

And

as you raise the temperature above that point, it becomes hydrophobic and will start to aggregate together.

So,

you can see just NIPAM and water, images of it here, below and above the LCST.

When

it's above and very hydrophobic, it's very opaque.

And

below, you can't even tell that it's dissolved in solution.

So,

what we did was we take a urine droplet,

we

mix it with this NIPAM, which has a cationic functional group on it to allow it to absorb to the surface of the EVs.

As

we start, when we place the droplet with this NIPAM solution on our droplet centrifuge and turn it on, the droplet begins to rotate and controllably heat.

And

what happens is, once we get above 37C, all the polymer on the surface of these EVs want to grab onto each other. And it starts to form these flocks that ultimately aggregate into a single pellet that we can remove from the droplet and analyze downstream for different applications.

So,

in characterizing the droplet centrifuge, we can see here that the kind of ultimate temperature that it reaches is a function of the input voltage. And we wanted to choose an operating temperature that got us over that temperature of 37C without going too high. So, we chose 10VPP. Here is kind of the real-time temperature measurements,

in

addition to corresponding videos where we mixed 50 nanometer fluorescent polystyrene beads with the NIPAM and turned the acoustic waves on. So, at the early stages, points one and two, you can see that you can barely see the fluorescent particles because they're so small.

But

as they start to aggregate and come into contact with one another, the fluorescence intensity is enhanced. And we ultimately are able to get the majority of these 59-meter particles to aggregate all into one central flock. We kind of conducted these across polystyrene particles across the size range of EVs and show that it's, you know, it's capable regardless of the size of the particles.

And

importantly, we wanted to incorporate this into a real-world application aimed at isolating urinary EVs and comparing it with the ultracentrification technique. So, in part F here, I have, this is nanoparticle tracking analysis of fluorescently labeled urinary EVs that we have.

You

can see the size distribution. And then we actually aggregate them together and they form this pellet. We can resuspend this pellet in another solution and measure the size distribution.

And

that also gives us an idea of the yield of the number of particles that we recover from the initial sample. Our yield is around 91% of the recovered particles. We do see that there's a slight increase in size attributed to, in this case, the polymer is still absorbed onto the surface. But most importantly, in eye here, what we see is this ability to selectively isolate. Here is Western blot analysis results for THP and two EV markers, CD63 and TSG101. The initial urine sample contains all three. In our isolated sample, we do get some THP, but we get a high yield recovery of the CD63 and TSG101.

And most importantly, in our kind of waste sample, so after we remove that pellet and run the Western on the remaining fluid, there's nearly no markers for the EVs left behind.

And

then we wanted to show that it was compatible with some downstream functional readout. So we conducted RT-qPCR on two different microRNAs, MIR-21 and MIR-204, and showed that our yield is much higher than that achieved when just using the standard ultracentrification protocol. Because as I mentioned earlier, that THP network actually traps and a lot of the EVs end up getting tossed out in the waste in the intermediate centrifugation step.

So

overall, float is a rapid technique for the enrichment of EVs or other nanoparticles. It enables very high yield recovery and selective recovery of EVs from contaminating proteins.

And

it's also compatible with low-volume samples, making a promising approach for point-of-care applications.

Going

forward, some of the ideas that we had are to incorporate this thermoresponsive polymer, NIPAM, into our continuous flow device

to

enable a kind of all-in-one solution for EV isolation from unprocessed biofluid samples. So you can imagine kind of a third module

that

would then separate the EVs from soluble proteins based on kind of that controlled heating aspect of the acoustic waves.

In

addition to that, we used a very basic version of NIPAM for this that just had a cationic group added to it to allow the molecule to absorb to the surface of the EVs.

But

you can actually do some polymer engineering and have functionalized thermopolymers to only allow the polymer to bind to maybe specific surface markers to enable enrichment of specific subpopulations of EVs or any kind of bio-nanoparticle of interest. Quickly, I'd like to acknowledge the Duke Acoustapolitics Lab and my advisor, Dr. Tony Huang.

And

then I'll open it up to questions and just play really quickly the video

in

real time. Let's see if this will play.

Here

we go. This is actually the fluorescently labeled EVs being concentrated in real time. And you can see the droplet is rotating at extremely high speeds.

We

surpass that kind of LCST. And then

the

particles start colliding in. These urinary EVs start colliding into one another.

And

then we can come in with a glass capillary and actually remove these EVs and transfer them for downstream analysis. So there it is. There's the glass capillary. We can remove it. And

my

talk may be a little bit out of the microfluidic areas. So I will talk about from Alice to later, the adventurer of our self-powered smart system. I'm from Peking University located in Beijing. And let's see. So I was joined Peking University 25 years ago. So it's a pretty long time. So that time everyone was working on math. So MAMS, you know, everyone knows these pictures is sensors and actuators in the, you know, micro-nanoscales. So that time we named this as micro-electromechanical system. So but from my side, you know, I make another kind of type of this MAMS is miniature everything and make it smart.

Because not only, you know, try to make something really smaller, but also, you know, want to make something very smart. So, yeah, since 25 years ago, I was working in this field and I tried to make some fun. So you can see this is my group. So it's Alice Wanderlap. So because my English name is Alice, so I want to make something really joy and fun and interesting. So I made this, you know, my lab was Alice Wanderlap. So with all these young, you know, students and really, really great collaboration with many of my colleagues and friends. So we make something interesting. So today I'm going to, you know, give you some examples as, you know, from Alice to Aleda for the topic is right.

Because the name link and also in the last, you know, five or three years. So everyone wants to talk about AI, AI, AI, you couldn't make these words. So from my side, I'm thinking, you know, what AI meaning is really in my research and what we can do for this AI. Let's look at these two figures. It's myself who was a carbon-based life, right? Yeah, so later in the future, the artificial intelligence will be a silicon-based life. So from this carbon-based intelligence life and to a carbon-based, to a silicon-based, you know, intelligent life. So what we can do and what we can contribute from our own research.

So I think in my research, our three kind of keywords were related, you know, with this future artificial intelligence. First is self-powering, another word is active sensing, and the third is about smart system. Why, you know, I choose these three keywords. Let me explain this first. So let's see that, you know, Aleda, so this is super, you know, artificial intelligence. So it's not in real, but will become in reality in the future. But think about this girl, why is it super-powered? Because it can, you know, generate the power by herself. So that I'm thinking, you know, we need to power, you know, the future, you know, artificial intelligence system.

So what's the best way, you know, to use battery or use the long wear with a long tail? That's not the final solution. So if we can, you know, contribute to some technology from power by themselves, that would make more sense. So 20, you know, 12 years ago, so we started as

triple

generators. So this was a quite interesting story. So because Professor Dong Lin-Wang from Georgia Tech at that time, he was invented in the triple electrical nanogenerators and published the first paper on his own journal, Nano Energy. So I met him in Italian on conference. So he told me this, but anyway, beginning, I really couldn't believe it because, you know, triple generators like, you know, you use the friction, all these can, you know, very normal, very normal motions, all these things. It can generate a high power.

So that time, at the very beginning, I couldn't believe this, so, but as a scientist, I think, you know, if you don't believe it, you need to try by yourself, see if this was the truth. So after the meeting, you know, in Italian in June, you know,

2012,

so I back to PKU and I started to try, you know, see if this can work. This was the first thing,

the

examples we did, these triple generators in my Alice Wonder Lab, because at that time I have a lot of samples on my hands as for this kind of micro-nano durable structures and taxing surface. So I made a lot of these samples, so that time I want to use it, you know, for other applications. But with this, I think, okay, let's try with this micro-nano durable structures and make a simple, you know, sandwich, you know, triple generators, see if it can work. So we make this very simple structure, so you see the double layers with micro-nano durable structure, there in between, it's a couple layers, used as the electrodes, also as a function for this, you know, movement, the frictions.

So this triple generator, so the first generation was work, see the output of this, you can see the current, not only the voltage can be go up to, you know, 400 volts, also the current can be go to 100, you know, micro-ampere. So this is a big step, you know, this can, you know, generate real powers, and this can be used as continuous output, can be converted to more than 10 volts. So that makes a lot of sense, because many of these micro-nano chips or devices or sensor activators can be turned on by this, you know, output or by this simple, you know, triple generators.

So since that time, we started, you know, to focus on this topic to make the nano generators, see if we can make, you know, many kind of helpable or usable generators. So this was the simple fraction surface, you know, triple generators. Why we make this, because one of my students was talking with me, he said that Alice, all these complicated structures with double or triple or even multi-layer surface is, you know, a big volume also, not only the volume, it's also needing more space. But most of the applications will be focused on all these, you know, micro-nano devices, which are based on touch screen, you know, kind of the ends.

So if that's possible, we can use, we can develop something, you know, by this touch screen, you know,

devices.

So that time, we make these micro-nano durable structures with transparent, you know, materials, and put it on the surface of this screen. And then can be generated powers. So this was the first, you know, single friction surface triple generators. This really, really exciting and, you know, push our research moving forward to generate the power in a simple way and also, you know, with high output. So this was for the generators with a triple generator and piezo generator hybrid together and to make a much larger output. So you can see these simple structures with micro-nano durable structure to make R-shaped, you know,

generator.

So this can harvest the, you know, power, the power, the moment, not only from the friction, the touch and the screen, the touch and the departure. Also can generate the power by this moment, by the piezo. So this can enhance the output. This wave shape, you know, the structure with multi-layers also can generate the power from a triple and piezo. So this makes the output more reasonable and for applications. And this was the carburet structure with triple generator with a magnetic, you know, magnetic electrical generators. So everyone knows that a magnetic electric generator can be have a larger, you know, current and the triple one can have, you know, larger and higher voltage.

So this can be, you know, combined together and to apply applied in many, you know, applications for this. So like, you know, the wearable and

the

wearable devices. So all these kind of applications also can be, you know, move forward into this very nice structures with a three-dimensional movement. So this was an example for the solar cell. So we make these mirrors on the surface and this tunable, you know, tunable structures can be, you know, follow the light in the whole angles. So this can be covers the energy from the solar. And also this can be covers the energy from the triple. So this could also can be enhanced application in these outdoor can devices. So my group has many, many of these hybrid energy harvesters and for different applications.

And all these surveys and looks very interesting because all covered by all MEMS fabrications. So it can be mass production and also can be very cheap and very convenient for everyone to follow this structure. And another kind of step forward is for how to use, you know, can these structures can be as sensors. Like we see from Aleta, you can see that Aleta was super because not only Aleta can, you know, can

be

function, but also can be super sensitively, you know, response to the outside and they make decisions. So by his custom and this close, you know, you can see that this was a super sensor network. So with many, many super sensitive sensors,

you

know, on it. So that time we think is that possible use micro maps for sure as for the sensors. But another thing is that possible can we combine these, you know, micro generators with these sensors because this has a good kind of principle. You see that by this, you know, triple generators also can be think about the electrification. When you touch on the surface, you can generate the electrons on the surface and also, you know, these electrons with rich information of the output. So this can be used as a sensor, active sensors, because it's, you know, signals was based on your own, you know, moment and all these, you know, reactions.

So this was the first thing example we use this principle to make analog smart skin. Why a name analog? Because you see the structures is on this, this was original structure. So you can see that it's a full electrode. And on the flowers, this was the demo. You can see that it's totally transparent. And the electrodes was on, you know, the urge of these devices. So on this large areas, just need four electrodes. Then you can get, you know, the precision, a very precision location of the output, the force and the locations. So these meanings, you know, really helpful to use as all these smart skins.

Like on a lid, if you can put this transparent, you know, the smart skins, you know, on the surface, you can easily to get many informations by this. So this was really, really good. So I remember on one of the news for this publication that now the robots confuse outside. So that's exactly what we want to do. Make all these, you know, robots of, you know, artificial intelligence later. It can really fuse outside, fuse the environment. So this was the first sensor we made.

And

then based on this, we're moving forward to, you know, get some more exciting devices or sensors. So this was non-contacted electrical skins. So because just now that one, you need to contact the surface. So many things you don't want to, you know, contact, maybe remote control, remote fillings. So is that possible to make it as remote fillings? Then this was a special design with surface, you know, patterns, different treatment. So we make this, you know, this non-contact electrical skins. So this non-contact electrical skins give us more space, you know, can potentials for application of this, you know, kind of sensors.

And this was a game designed by the students that can be, you know, remote to play this kind of games by themselves.

And

this was a 3D multi-function electric skins. The principle is all the same, but the difference is this was a multi-layers in the 3D dimension. So this 3D dimension can get you more information about, like signatures. Use this as examples will be easily to understand. When you,

you

know, use the signatures, this was not only by the patterns, also have information, you know, by the force. So if you can detect the force in the 3D dimension, this will give you more kind of precision or operation, you know, for this detection. So this is 3D, you know, multi-function sensors was really usable for this kind of applications. And this was another kind of smart, very smart design.

I

like this very much because this was used patterns. Many times you can see that, you know, you want to get more kind of precision information. You need more elect holders. So like in our lab, we not want to, you know, just increasing, you know, the electric, electro, you know, kind of numbers.

So

this will be kind of very complicated. So is that possible to design something, you know, with limited elect holders and you can get more information? So this was by the, you know, design. So just use five or six, you know, elect holders. We can cover more than, you know, 200 angles. And this can be get output information as more can reach and, how to say, multiple, you know, information on this. So this is really, really great kind of design. But this also can, you know, relate to this picture. So because of that comes, you know, multi kind of data. Yeah, really, really big data. So you need the upgrade mode to, you know, enhance your output and to calculate all this information.

So that time I use motion learnings, motion learning technology to enhance our, you know, hardware. So use software and the hardware together, you know, to get more information. So last year, we remember that, you know, the Nobel Prize go to motion learning. So exciting events also can use motion learnings to enhance your research. So it's really kind of exciting and great, you know, integrated of this kind of new tech, not new technology, but this, you know, topics into your own research. So this also help us a lot, you know, for this, you know, systems. So this will come to the last part of all the smart systems.

Actually, we see that you have, you know, you can power yourself and you also can, you know, have a multi function, the sensors, you know, by this simple principle. And they use the soft or flexible materials. And the last part will be, you know, if you really want to make a function, you need to make like smart system. It's not just the sensors or just, you know, actuate or something. It must be a sensor. A smart system can be working together. Let's see, you know, we want to make this system. We need some kind of more exciting technologies because, you know, not only one sensors, not only one, you know, activators, not only one generators can be a system.

You can, you should to integrate them together. If you want to integrate it together, you need a complex, you know, system. But the thing as is that possible, you know, to make the system not that complex and also can be reversible and reconfigurated, you know, by this. So that time we think about we need to use one very powerful materials and magnetic materials. So we synthesize this hard magnetic graphene, you know, nanocomposite in our lab. And this, you know, we shot as HMGN. This was, you know, composite has a very special kind of functions. It can, first, it can be, you know, enhanced the electric, you know, connections.

So you need multi layers, you need multi, you know, sensors or devices on the surface. So, but you can use this magnetic aided electrical, you know, electron, electric, you know, connectings. It can easily to, you know, save a lot of space and also can be solved the complexing line and all these works. So this was the first part we use is magnetic aided, you know, materials. The second is we found that surprisingly this magnetic, you know, materials also can enhance the, you know, the sensing, you know, performance, the sensors of performance. So I will give you some examples here.

First is this was can be really enhance, you know, this self-alignment and this very high reliable connections by this, you know, magnetic materials. You can see that these were different electrodes and with magnetic materials, you can automatically, you know, mounted on the right space and the right location. Also, the electric, electron connections was very stable and very good, you know, kind

of

resolutions of all these were connections.

So,

so this was the first part of this. The second part is you can see these configurations by these materials. It's really flexible and also stretchable. We make it stretchable too. And with these, you know, materials, this substrate, easily you can make different sensors and actuator to mount it on this, you know, substrate. And also you can easily to change, you know, the functions with different sensors, like this, all this, you know, standard design with connections of electrodes. And you can refunder it, reconfigurable the sensor and actuators on this simple structure, on this, you know, substrate.

So we tried many different sensors, more than, you know, five or six, you know, kind of sensors on this. And this was the schematic design. You can see that two electrodes devices and three electrodes devices, you know, the simple design on this, and you can easily mount it on this. And this was different sensors. And this was a real product. And we fabricated this, you know, on the polyamide or polyamide. And this can be transparent and flexible and stretchable, you know, the devices. So this was with two electrodes, three electrodes, and then this was the sweat sensors. You can say with seven, you know, electrodes. All this can be the very standable, you know, models.

Then use this, you can easily change it to others. Let me give you examples, you know, for this application. You see that just now that was examples. And this was real, you know, applications used on the artificial intelligence robots hands. And on these robots hands, they have many different, we have about one, two, three, five, six, seven, you know, sensors. And all these sensors was mounted on different, you know, place. Because it's different place, like, you know, in the center of this, it's easy to harvest the sweat. Other places you need the force sensor or temperature sensor or gas sensors, different other parts of that. The connections of these wires and all on the different space.

Then you can see this was the details of different, you know, sensors. This was the

structure

of the whole system. With this whole system, easily you can get, you know, different sensors and also with the energy harvest inside. So this was really can help you to, you know, self-powering and also kind of different, you know, parameters about outside. And also can get more information on this. And this was a real kind of data or examples, how to, you know, make the detections. Here, you may ask the questions is that, you know, you have so many sensors, you know, on this single, you know, substrate. So maybe there are the sensors and they close talk a lot, you know, not that super, you know, or, you

know,

acturation will be a question of precision, of some questions. So here we have some strategy to solve this, you know, coupling issues. So we design the system always, you know, some thinking about to decoupling of these multiple sensors. so we also use algorithms to solve these problems too. So on this structure, on this robot hands, we have seven sensors and generators on that. So it covers physical sensors, chemical sensors, the biomarkers, and so we're different. So it's decoupling by the seasonal design, also by the algorithm. Afterwards, the data processing to solve this problem, then can get all these different parameters on the same structures. So this was another example.

So we try to move our mass production. So this mass production, you know, the sensors is, you know, by this mass fabrication too. You can see this, you know, this tactile sensors was last year, or makes this collaborate with another professor, is also my PhD student, Meng Di Han, you know, as designed based on this MEMS sensor and also by this simple structure and simple principle. And also, you know, for the, you use different materials to make this tactile sensors, is a mass production. Super smart sensors in this structure is a three dimension. Just now all the sensors on the surface, on the, you know, on the rubber hands is 2D dimension. And this was a 3D dimension.

This can be harvested,

you

know, information from the 3D, you know, for this textiles. And you can see this was, you know, different design of this, but by the simple, you know, fabrication too. This can be, you know, the whole waiver level fabrication of this, you know, really, you know, mass production and a big array, the large array of all these sensors. And this 3D, you know, textiles sensors is very easily, you know, to apply it in many, in different applications. Also, put in different, you know, location of the robots. And this was another high density design for this, you know, sensors, this textiles sensors.

You can see that milling, you know, on a very small kind of surface, on a very small, on a very small space. So, but this can be enhanced robot hands, you know, for the textiles, the sensor address, so very much, so they can get very sensitive information and especially as 3D, you know, information on this. So, this was a larger area application. You can see that this design was also follow, you know, the previous one also uses, you know, multifunction and reconfigural structures, you know, to fund all these, sensors to put all the sensors on the substrate and to easily, you know, to get more high resolution and high performance of the sensors.

So, this was the application of that you want to put it, you know, on different kind of coverage surveys. So, this surveys, they need more kind of flexibility. So, you can use this serpentine design to make it flexible and to make a more kind of precision on different locations. So, another

part

in the future, we want to enhance all these sensors and this kind of substrate and with generators and also actuators on the same system. This will be an all-in-one smart system. This was a long-term, you know, kind of work in our group. And also, I think this will be a long-term, you know, achievement to be a real artificial intelligence, you know, by the small sensors and actuators in the system. So, this was, in my group, we really can focus on these topics. So, in last, you know, a few years, we really published several, not only the papers,

we

also published several books on this for the smart system and also for these nano generators. One was in Chinese, another two was in English. So, if you are interested, you can use it to find it online, you know, all these works was already in these books was well explained and calculated.