Walter Voit: Nerve Tags - A Route Towards Big Data of the Nervous System

By: Talks at Google

CHRIS WALLEN: Hey, everyone. Thanks for coming. My name is Chris Wallen. I'm an engineer here at Google. I work on the Maps team. I'm hosting today Walter Voit. We went to the University of Texas at Dallas together, got our undergraduate degree in computer science and studied machine learning.

Then Walter went off to Georgia Tech to get his PhD in material science and engineering. And he's going to tell us a little bit about this nerve tag technology that he's been working on. So I present Walter Voit. WALTER VOIT: It's a pleasure to be here at Google this afternoon, or at Alphabet. I don't know what to say these days. But if it is Alphabet, maybe this will be an alphabet that you guys can make in the future. But what I'd like to talk about is a lot of the collected work that we've been doing over the last half decade or so in the area of implantable bioelectronic medicines. But it's this idea that we can build computer chips onto plastics, we can put them in the nervous system, and we can record, block, or stimulate arbitrary nerves in the peripheral or in the central nervous system and build closed loop feedback systems with smart computers, with smart beds, with smart doctors to help understand the connectome, understand the body's nervous system, and ultimately treat debilitating neurological conditions.

A lot of the work that I present here has been funded by some generous grants from DARPA, the Defense Advanced Research Projects Agency, by Glaxo SmithKline, the British pharmaceutical giant, and by the NIH and the NSF and a host of other companies. I'm also going to spend a few minutes at the end of the talk and mention some of the different companies that have spun out of my research lab over the last half decade or so. And so we'll get to those a little bit later. There's a picture of me with some of my DARPA colleagues at West Point sitting on an Abrams tank. And here's a shot of some of our group. And so I don't want to run out of time at the end, a quick shout out to my lab.

I've got about 12 postdocs now and 12 graduate students and about 50 undergraduates at the university that do a lot of the heavy lifting day to day, which ranges from synthetic organic chemistry to mechanical testing to physics and interfaces to electrical engineering to applied neuroscience. And so these students from many different disciplines are all working together to solve some of these grand challenges in synthetic biology and neuroscience, but from a strong materials background. So I wanted to start with some slides from my colleague Kris Famm, who's the Vice President of Bioelectronics Research at Glaxo SmithKline. GSK, over the last couple years, has made a giant bet in the field of bioelectronic medicines. It's also called electroceuticals or Pharma 2.0. But it's this idea that we can use electrical signals to modulate the body's behavior. So today, if you look at how a lot of drugs and pharmaceuticals work, you're using small molecules to really affect what the body is doing. But the idea here is that we can use closed loop feedback systems with miniaturized electronic or magnetic or photonic devices to begin interacting with the body.

Walter Voit: Nerve Tags - A Route Towards Big Data of the Nervous System

And so it's a very different paradigm from using small molecules. Small molecule drugs are sort of like giving someone a fish. You're giving the body something that it needs. For a short period of time it reacts to that drug, and in a lot of cases, then, that drug leaves the system, and people need more drugs. Well, if we can get in and control what the nervous system is doing, we can teach the body to overcome some of these debilitating deficits and prevent the need for long term drugs or supplement drugs with being able to build these closed loop feedback systems with the nervous system. This ties into a lot of really interesting research areas today in neuroscience, including the study of plasticity, or the brain's ability to rewire itself due to certain stimuli.

And we are trying to build these devices that can help affect how the brain is rewiring itself in real time. The nervous system is highly complex, and so I'm not going to belabor you with really any of the details. But we put up this slide just to show that there are a tremendous number of peripheral targets that have very large end effects on especially what your gut, what your viscera is doing, and all these different touch points. We'd like to really change the paradigm for how we're interacting with the nervous system. Today, devices like cochlear implants and pacemakers give you one to 22 channels of information exchange with the nervous system. And I think the future, the future that Google will likely be a very big part of, is big data analytics of the nervous system. It's these miniaturized injectable devices that can wrap around arbitrary peripheral nerves.

And instead of having one touch point of what the nervous system is doing, you'll have these distributed touch points that really tell you a lot about the real time status of a patient. When you go into a hospital today, you get your heart rate measured. You're measuring all these physiological symptoms. But the nervous system is largely untouched.

If you had little devices that were inexpensive to implant that could give you real time performance metrics for the nervous system, it would really change some of the fundamental assumptions and paradigms in the sick care world in which we live today. And the goal of this talk today was to separate out some of this vision and some of the science fiction from the reality of what some of the leading scientists across the country are doing today, what some of the leading medical providers across the country are doing today, and try to paint a picture for you guys about when the appropriate time to put your might on the software and on the big data side into this area might be appropriate. And so if I succeed in the talk, maybe you guys can tell me when the right time to get involved in this is, and so that's the goal.

So today, we've got a host of individual systems that can stimulate the vagus nerve, that can stimulate the cochlea, that can stimulate the spine, that can solve sort of one-off biomedical problems. But the goal is to have these miniaturized devices that can do that across the body. So our solution from UT Dallas are these devices called nerve tags. But we envision these to be these needle-injectable microelectronics that will be smaller than a grain of rice that will be able to come and wrap around arbitrary peripheral nerves and block, record, and stimulate on those nerves. And so our objective is to make these very inexpensive so that a surgery is not costing you $25,000 to put in a device, but that it would be feasible in a hospital environment to put in a large number of these devices. What becomes really interesting is it begins to sync up the business model of big pharma with the semiconductor companies that will likely make this possible. If you look at a company like Texas Instruments, they're looking at products that are largely nine months to four years down the pipeline, whereas GSK spends 25 years developing a drug, only to recoup costs in the last few years when these big blockbuster drugs are on patent.

And moving to these small, inexpensive devices allows us to get high volumes, to get semiconductor companies really interested, to reduce the non-recoverable engineering costs that go into designing these devices. And it also allows us to leverage tremendous resources that have been put into these devices already that companies like TI and Intel have spent toward the Internet of Things. So what we can do right now-- and then I'll move to some data and some pictures-- so this is the kind of device that we can build right now. This is a little PCB we've designed, where we've been able to sort of flip chip or solder on maybe 11 different Texas Instruments components. I keep mentioning TI. We're in the heart of Dallas, and so they are the Google of Dallas, if that makes sense, I guess, that you guys are to Mountain View. And we're increasingly miniaturizing these devices to the point where we hope these will be needle-injectable and easy to put into the body.

So my background comes in this class of materials called shape memory polymers. And these polymers are plastics that can change shape and stiffness at different temperatures. So these are plastics. You can see a video of one of these.

This is being inserted into a warm environment through a 100-micron slit. When it heats up, these materials can change shape, can change stiffness. So in this case, we're self-coiling a matrix address transistor array around a two-millimeter cable.

Here is one of these materials sort of in the flesh in real life. So it's a material that behaves like Plexiglas at room temperature. If I just put it in my hands and heat it up to body temperature, it gets 10,000 times less stiff, and I can bend it or manipulate it into some sort of metastable shape.

And then as soon as it cools off, it gets very hard again in this shape. So our big idea was, could we take these plastics while they're really hard, insert them into the body, and position microelectrodes right up against and around nerves, but then have these guys soften and get 10,000 times less stiff chronically to avoid long-term inflammation and scarring responses in the body? And I'm happy to say that a lot of the preliminary work that we've done with some of this Google funding-- we just got a big Army grant as well in this area to really study the extent to which softening affects the neurological responss-- shows that these soft materials have a lot of promise. We've built devices to go around the spinal cord into the cochlea on the DRG, and then a host of different nerves, mainly in the viscera and in the four limbs. So broadly, what we do is we design these materials called shape memory polymers. I'm a polymer chemist from my PhD training at Georgia Tech.

But then we use photo orthography to build flexible electronics onto these materials. We do things like control the charge injection capacity so we can get a lot of current out of very small areas by nanostructuring materials like titanium nitride. We build devices with low noise so we can actually understand what nerves are doing. And the goal is to really enable this hypothesis-driven neuroscience research. So right now we're providing research tools for collaborators back in Dallas and across the country to help map the connectome and understand debilitating neurological conditions. In the future, we hope to translate this to be a therapy in and of itself to treat patients.

So how these plastics work-- I won't get too deep into the chemistry-- but a lot of materials, if you look at them as a function of temperature, are fairly stable in terms of their modulus or their stiffness. But if you've ever taken a racquetball and dunked it in liquid nitrogen and thrown it against a wall, or taken a banana and gotten it really cold, you can hit a nail into a board with this banana. A lot of polymers, when they get really cold, get really hard and stiff. When they get really hot they melt or they get very soft. We can control that modulus change within a five-degree temperature window.

So we have these materials that are right at a transition that get soft very, very quickly. What we've also found out how to do is, we can have that softening be triggered by the absorption of small amounts of fluid, not enough to disrupt our electronics, but enough to make the whole substrate network very soft. So what you see here is a device that's, when it's dry in the black line, it has this little transition between 40 and 50 degrees. And when it gets wet, that transition moves to between 20 and 40 degrees. Now that doesn't look like a lot, just a little blip on some random curve from an instrument in the lab called a differential scanning calorimeter. But what this allows us to do is give surgeons five or six hours to implant these devices while they're as stiff as PEK, poly ether ketone, or a polyamide, or parylene c. And then once they get to physiological conditions, they'll get a soft as silicone rubber, or in a lot of cases, softer than silicone rubber. And we've found ways, now, to build microelectronics onto these plastics.

So here's another little video of one of these polymers that's been trained to coil around a nerve. So we're using a hair dryer here to show the effect pretty quickly. But you can imagine a surgeon putting this in and getting it to self-wrap around a nerve. In fact, in the bottom right of the screen there, or bottom left on your side, you can see one of our thin film electronic devices that's wrapping tightly around the vagus nerve. That's a two-channel electrode. We spend a lot of time studying the failure mechanics at interfaces to make sure these devices are going to withstand the aggressive, high deformation environments in the body. So this is a tool that we built that allows us to compress these devices and measure the radius of curvature of the electrodes. In this case, we built transistors both parallel and perpendicular to that bending stress.

And we did that on the inside and the outside of devices. We can measure electrical performance, but we can also do microscopy and understand what's happening to, in this case, the gold electrode, or in this case, the DNTT, the [INAUDIBLE] which is an organic semiconductor, what the deformation mechanics looked like, how to design devices that aren't going to fail when they're subjected to a lot of these kinds of conditions. And at first, we do that dry, and then we do it in moist, aggressive biological environments. So this is a typical pattern of what one of our photo lithographic processes would look like. We've came up with some clever technology to sort of begin building devices upside down, and then we can transfer those devices off onto a polymer substrate. So it's sort of like if you were to take super glue and scotch tape. If you imagine taking a piece of super glue that's already hardened and sticking scotch tape onto it, you could pull that tape off pretty easily.

But if you started with scotch tape and actually cured the super glue under the scotch tape, it sticks a whole lot more. You have to scrape off that scotch tape with a razor blade. Well, in the same way, if we can use our polymers not just as a material that we're building electronics on, but we can integrate that polymerization into the gate stack processing of the device, we can get a lot of the polar groups in the polymer to stick to thin film metals, to stick to insulators, and really give us great coupling through these thin film layers. So we'll build between 20- and 200-nanometer layers of gold. We'll sputter on nanostructured titanium nitride, and then use this transfer process to basically get really good electronics built at very small size scales.

And if there are questions later, we can go into that process in more detail. We've hit some of the high points. Each of these steps involves several different photo lithographic steps in the clean room. So some of the early animal work we've done lends a lot of credibility to this softening hypothesis.

In fact, the Chair of Bioengineering at George Mason University, we were able to convince him to move to UT Dallas about two months ago to continue to work with us to really prove out this hypothesis. But Joe and I got this big Army grant just a month ago or so. What you see here is an immunohistochemistry plot of a cortical probe that was put into a rat brain. We were able to record 350-microvolt signals after 77 days, so almost three months in that rat brain, with these PEDOT electrodes. A PEDOT is a conductive polymer that we put, in this case, on top of our gold electrodes. Then what we did is we stained, we sectioned the rat brain and stained it for different cell types. And what you'll notice is that there's very little scarring right around the device, and we can clearly see different cortical regions, the different color stripes that are unperturbed with these devices.

So we put these devices in while they were really stiff, and then after a couple hours, they got really soft and didn't lead to that same kind of scarring response. We've had a number of generations of these nerve tag cuff electrodes. We've worked very closely with the Romero Lab at UT Dallas. Mario Romero is a close colleague who does a lot of the animal work with us.

And he's put our devices onto a host of different nerves and measured physiological responses. This is a study of baseline noise and spiking activity in a hypoxic rat. So we're able to reduce the amount of oxygen that the rat is breathing at a certain time, and cause the rat to then-- certain nerves to fire a little bit more. We've done work with Rob Butera's lab at Georgia Tech. And so here's a picture of one of our shape memory polymer devices as an upstream stimulating electrode, and then as a downstream recording electrode. And so this is the stimulus artifact that we get on our recording electrode. And then this is that sort of ionic signal that's moving through the nerve a little bit later, something called the compound action potential.

We can also measure just random noise or random movement as the animal's breathing, and then we can use spike sorting and filtering to understand what nerves are doing in their ambient environments. This is a lot of work that was done on the sciatic, the cervical vagus, and then we've done a lot of work on the splanchnic nerve as well. Working with Ken Yoshida's lab at Indiana University and Purdue University in Indianapolis, we're doing some neat work on depositing our electrodes onto tungsten microneedles that we can thread interfascicularly through very, very small nerves, and we can use that, then, to position electronics very carefully. I haven't talked a lot about how we get the signal out. The next part of the talk will be here. The beginning is just sort of how we get devices in and how we're interfacing photolithographically defined structures with the nervous system. I'll also come pass around a couple little samples while I'm talking here.

Maybe Rommel can come show these around. But these are just some of the device configurations that we've put together. One of the things that's coming around is some of these plexus blanket electrodes that we're working on with Jay Pasricha's lab. Jay is one of the leading gastroenterologists in the world working at Johns Hopkins University. And he's done a lot studying the guts and the stomach. And we're working on devices for him that can help manipulate signals moving to and from these nerve plexi and understand those. So these self-wrapping coiling devices that can fit around the stomach.

We're doing some really neat work in cochlear implants. So this is a project that we're working on with colleagues at UT Southwestern, specifically a fellow named Ken Lee. And so what you're looking at here, in the video that's playing in the top left, is a cochlear implant from a company like Cochlear. As that's inserted into a cochlea, you can see how this hugs the outer wall of the cochlea. Often, that causes a traumatic insertion, and you're scraping cells, and doing a lot of damage as that cochlear implant is going in. Well, with the self-coiling shape memory polymer-based electrode, what we can do is we can time the recovery force of that material. What you'll notice is that we're never really touching either of those walls as we're inserting this device, and so we can have these modiolar-hugging self-wrapping electrodes that can position our high charge injection capacity electrodes in the scala tympani, this region in the cochlear, up against the organ of corti neurons that allows us to do better stimulation. What we can also do is build, for instance, robotic insertion devices that will help surgeons time this implant correctly so that we can coil several turns into the cochlea.

So why is this a problem? Cochlear implants are perhaps the most successful medical device in the history of the world. There are 200,000 working cochlear implants in patients today. It's a $2 billion a year industry that's been around for almost 30 years. So what I'm going to play for you is some different sounds of what sounds might sound like in a cochlear implant.

So if you only had one electrode working [CRACKLING SOUND] that's what a sound would sound like. If you had two electrodes, [AUDIO PLAYBACK] [STATIC] [END PLAYBACK] WALTER VOIT: You start to get a little bit of definition in the sound, but it's still hard to pick out what's what. At four different discrete frequencies, [AUDIO PLAYBACK] -A boy fell from the window. [END PLAYBACK] WALTER VOIT: And then at eight. [AUDIO PLAYBACK] -A boy fell from the window.

[END PLAYBACK] WALTER VOIT: And then at 16. So this is where modern cochlear implants sort of are. [AUDIO PLAYBACK] -A boy fell from the window.

[END PLAYBACK] WALTER VOIT: So you can probably hear the sentence. It's a weird sentence, I apologize. It has the right balance of consonants and vowels, apparently, to be used as a sentence.

And then at 32 channels. [AUDIO PLAYBACK] -A boy fell from the window. [END PLAYBACK] WALTER VOIT: And then the unfiltered signal. [AUDIO PLAYBACK] -A boy fell from the window.

[END PLAYBACK] WALTER VOIT: So you can see, even between-- [AUDIO PLAYBACK] -A boy fell from the window. [END PLAYBACK] WALTER VOIT: That sort of tinny, very processed sound-- [AUDIO PLAYBACK] -A boy fell from the window. [END PLAYBACK] WALTER VOIT: --and natural audio, there's a huge difference. And we'd love cochlear implant patients to be able to really appreciate sound, appreciate music. My wife, who's here in the back, Felicity, she's an ear, nose and throat head and neck surgeon at UT Southwestern. And so we're working with her and some of her bosses-- Ken Lee is one of them-- to really try to understand these self-wrapping atraumatic insertion cochlear implants. But the idea is, if we can position these electrodes closer to the nerves we're trying to stimulate, we've got higher charge injection capacity, we can get more current out of a smaller area, we can get more specificity, we've got a lot better way to interact with this complex part of the body. So the way we can do a lot of this in a university setting is because we're, in some sense, it seems, some days, not in a real university setting.

Texas Instruments has helped leverage close to a billion dollars into UT Dallas over the last 15 years through some state projects and grants and funding. This building that we work in is a $100 million facility that's built around a $50 million clean room facility. And so even though that's only seven or eight years old, these eight guys that run our clean room have a combined 200 years' experience running clean rooms at TI. And so I can send students like Rommel down there to get trained by experts who've been doing this all their lives, and really get all of the details right. We can focus our understanding on polymers, on interfaces, and really use this as a core facility to process complex electronics onto shape-changing plastics. So our first generation of these nerve tags, this path towards being able to get big data from the nervous system, looks a little bit like a mini squid. We've got a-- and that's not a quantum interference device.

That's like a biological squid. We've got our antenna. This is a little copper coil in the back.

This whole thing is encapsulated in a thin layer of silicone. We've got a lithium polymer 10-milliamp power battery on top, so we're able to inductively power this device at 13 megahertz from a cage that a rat would sit in. And then we use that battery to stimulate and to block nerves in the rat. And then we've got a channel for recording and a channel for ECG. And so that's sort of what the first generation of device looked like. But the whole thing was made very inexpensively for $100.

And then we interface it with our polymer electrodes. So that's maybe arguably the more expensive part here at the ends, where we've been able to photolithographically define complex geometries and then connect them back to this device. In the future, we hope that a lot of this will be miniaturized, maybe into a single die, even, that will be much, much smaller. So here you see a sample of one of these devices. We're able to wirelessly stimulate the sciatic nerve of a rat.

In the top you have the device that's just sitting sort of on top of the rat, and on the bottom you can see that it's fully implanted. But if you look at the relative size, it's still too large to have hundreds or thousands of these positioned in across the body. And this is sort of the reality of where we are right now. Where we want to go is to have this be more like sophisticated acupuncture. You would have these smart, intelligent needles that can put these grain of rice sized devices, maybe made out of our materials, maybe not, but made out of materials that can intimately interface with nerves and serve as this chronically viable, abiotic biotic interface to get information in and out of the body.

And so some of the large challenges come in encapsulation, and so we play with a host of different accelerated aging conditions, both materials and electrical, in PBS, phosphate buffered saline, in bovine serum albumin, in water, also then in animals, to really understand how these materials behave acutely, subchronically, and chronically. Some of the related technologies that have spun out of the lab have to do with building other thin film sensors with these materials. This is a sample of one of the temperature sensors that we've built in the lab. We spun in a company this last year called Pascalor, Pascal for a unit of force, and calore for calorimetry, or heat, or heat flow. And we've built some of the world's most sensitive temperature sensors and pressure sensors with some caveats. The temperature sensors can measure a thousandth of a degree change in temperature, but over a four-degree temperature window.

Now we can change the polymer. And in each four degrees, we can measure something with a thousandth of a degree accuracy. But in areas where thermal conditions are fairly well-known, like in and around the body, this gives us a lot of sensitivity. So when a lot of temperature sensors would have, maybe, a 25% change in properties, we can get a 10 billion percent change in that same property over that four-degree temperature window.

So what you see here is a temperature sensor that's put onto the brachial artery. And we can know that the hand is going to move before the hand moves based on the heated blood flow that's traveling through that brachial artery. And we can get that with extremely high signal-to-noise ratios with some of these temperature sensors.

If you look on the left there, we're tracking finger height in that plot as a distance off of a surface. And so what we can do is sort of wave our fingers in front of a screen and measure the relative distance of that finger to the screen based on the thermal gradient between your finger and that temperature sensor. If you look at that compared to a conventional thermocouple, we have very, very smooth lines. We have much, much higher sampling and sort of much better data rate than a conventional thermocouple. We're combining this with some of the work we've done in thin film pressure censors. So in the last two years we've developed what we think is the world's softest, fully elastic material. It has a modulus of one kilopascal, which is about 10 times softer than most cortical tissue. We can spin this into a 20-nanometer thick layer.

We can build interdigitated electronics onto that material. So what we've built here is-- the whole stack is less than 10 microns thick, but it's a transparent thin film matrix address pressure sensor. So you can see when we're pushing sort of on the pixel, you can push and hold, and you can measure sort of the lightest finger taps all the way to very, very heavy forces. If we had a fly come and land on one of these materials, we could see how the weight was distributed among that fly's legs, which is really cool. You can see if you're pushing other pixels, there's almost no overlap in that signal than when we're over sort of the sensitive pixel.

And we can build these pixels with about 40-micron spacing. So we could have a screen. Imagine some sort of touch screen, that you could interact with that screen in three dimensions when you're outside of the screen, and then as soon as you touch it, you can sort of interact down into the screen. So we see a lot of neat applications, both for wearable and consumer electronics, but also for really interesting new ways to interface with biology in terms of pressure and temperature.

Another company that we spun out, Syzygy Memory Plastics, is doing some neat stuff in the audio world and in the oil and gas world. What you see on the right here is something called a frac ball. But these help us do hydraulic fracturing, which is pretty big in Texas. But we want to be able to maintain 5,000 PSI on top of this frac ball, and then have it completely degrade in a very fixed amount of time.

Most polymers have a linear degradation profile, so if the well temperature's a few degrees off, it will degrade too quickly or too slowly. We have this logistic degradation profile that maintains its mechanical properties for a couple days, and then all of a sudden the whole network falls apart. And so this gives us a lot of specificity to help keeping oil wells open. I gave a talk across campus this morning from our 3-D printing company, Adaptive 3-D Technologies. And the problem there we've tried to solve is that most industrially 3-D printed parts can't be used for direct applications. People print molds, they print jigs, and then they do manufacturing around 3-D printed parts, but not manufacturing with 3-D printed parts. If you look at the stock prices of some of the big polymer 3-D printing companies-- that's sort of shown-- there was all this hype leading up to 2014 that they could solve some of the grand challenges in additive manufacturing, i.e., print a part that goes into an automobile. Unfortunately, that hype didn't quite pan out, and you've seen a lot of these stock prices dropping in the last couple years.

But it's not a business model problem. It's a materials problem. You need to be able to print materials, this layer, then this layer, then this layer, then this one, that are as strong in this direction as they are in this direction. And with a lot of our study of interface physics, of polymer physics, we've come up with ways to delay that cross-linking during 3-D printing to build isotropically tough materials. In materials that are tremendously strong, this is a stress strain response that's very similar to nylon. We can stretch up to 400% with a stress of 50 MPA, for instance, and so we're really excited about this next generation of 3-D printable materials. Another company that we recently spun out is called Aeries Materials.

There's a problem in semiconductor processing today, and it's this thing called CTE mismatch, or the Coefficient of Thermal Expansion mismatch. It's sort of a rule in the industry that you can't design materials that are thermally mismatched, or when you cycle them, they will fall apart. So in this case, these thiol ene acrylate shape memory polymers that we've built, which have a CTE of almost 70 parts per million, should fail miserably when you put them in contact with thin film metals. But in fact, we can withstand multiple thermal cycles up to 270 degrees C, and we have misalignment of less than one micron per centimeter.

So if you compare that with materials up here, like polyamide, like biaxally oriented polyethylene naphthylate, like polycarbonate, all these materials, as you thermally cycle them, they'll shrink and warp, and you'll not be able to align transistors on a photo mask over a large area. The reason we don't have cellphones today that you can crumple up and stick in your pocket is it's prohibitively difficult to build electronics onto plastics over large areas and still align all these components with submicron precision. Well, people have thought that CTE mismatch is this problem that dominates that. But the reality is, CTE mismatch is a symptom of having a lot of cure stress built into your polymer. We've come up with materials that, as they polymerize, are completely stress-free. So if that material has the strain capacity to accommodate that mismatch, and we don't have these local stress concentrators, we can build plasma-enhanced chemical vapor deposited silicon nitride, which has a CTE of 2.3, onto a material with a CTE of 70, and thermally cycle this indefinitely to 270 degrees and not fail. And so we're doing some interesting work in building flexible backplanes in materials, again, based on solving this very small interface physics problem as related to how polymers and metals stick together. Lastly, something that we've spun out is something I'm very happy with.

Chris has actually helped us with this a little bit in his spare time. But we've written one of the world's most comprehensive mods for "Minecraft." "Minecraft" is ostensibly the world's most popular video game. It's been downloaded four billion times in the last four years. There are more YouTube videos about "Minecraft" than even cats, which is surprising. There are about 100 million daily players of "Minecraft." And so we've gone and written a layer of material science in and on top of "Minecraft," not to be educational-- it is very educational-- but to be fun.

Kids can build flame throwers and jet packs and scuba gear and pogo sticks and bouncy castles if they teach themselves the underlying materials processing. So we've added oil and petrochemical refining and tech trees and ways to make all these sophisticated materials. But they're teaching themselves what we want them to learn in college so that they can build overpowered items to either blow up their friends in "Minecraft" or get around the world a little bit more quickly, or show off a better base. And so we've found some really neat ways to tap into these new modalities of learning using programming and using video games. And so we just launched this a little while ago. UT Dallas is actually offering weekly $5,000 scholarships for the Polycrafter of the Week. So if any you guys have kids that are thinking about college and want to win some scholarships playing "Minecraft," this is their ticket.

So in review, I've shown you some interesting things on the implantable sensor side of things. We have thin film transparent sensors that are mainly photolithographically defined. We've done work with carbon nanotubes.

Here you see a high charge injection capacity titanium nitride interface with all these little columnar posts. This is a dipole fractal antenna. We've got a laser system, a neodymium-doped yttrium aluminum garnet laser that can cut out samples within just a few microns of electronics and not redeposit sediment on the electronics.

And so we're doing some neat things to build devices to help us understand what the body's doing. Some other neat capabilities we have back at the University, one of my colleagues-- he was the president of MRS, the Materials Research Society, back in 2013. And he works on a material called ultra nano crystal and diamond, which is one of the world's stiffest materials.

It has a modulus of 1,400 gigapascals, making it more than 10 times stiffer than steel. We can photo pattern UNCD and build exoskeletons in and around our devices to make them anisotropically stiff, but then very flexible when they bend and contort inside the brain. Another neat capability we have back in Dallas, this is Moon Kim, and he runs one of the 20 most powerful microscopes in the world. This is an aberration corrected transmission electron microscope that can resolve down to 78 picometers. So a single-carbon carbon bond is about 150 picometers. So we can see point defects in [INAUDIBLE] disulfide, in graphene, in carbon nanostructures, and really use this to study things like single electron transistors and high K dielectric materials, and a lot sort of in the nano and the quantum world. And so in the end, we've got great infrastructure back in Dallas that's really been the driver for being able to build a lot of these sophisticated devices on the plastics.

And in Texas, we like to think small. So with that, I'd love to answer any questions. Oh yeah, one more comment. Yesterday we found out we were picked as one of the panel winners at the big South by Southwest music festival in Austin, Texas in March. There's going to be a forum called Inner Space: Bioelectronics and Medicine's Future. And so I'll be there speaking with Kris Famm, who's the head there from Glaxo SmithKline of Bioelectronics, and the president of SetPoint Medical from New York, and then someone from the Metropolitan Museum of Art to talk about of the confluence of art and neuroscience. So I'd love to answer any questions. Thank you very much for your time.

[APPLAUSE] AUDIENCE: So after a few months, you gave the example with the rat with the implant. Does the neural signal degrade? You know, they get corroded, like tends to happen with some of these devices. WALTER VOIT: So for the one that we had in the rat for three months the signal was unperturbed for those three months. Unfortunately, the animal protocols we had, we were forced to sacrifice the animal at that time and do histology and immunohistochemistry. So that big Army grant that we got, we are looking at that over much longer time frames. But in that case, we had the 350-microvolt signals, which are the same as when the device started. Now the cortex is a little bit different environment than in the periphery, and so we're trying to study that phenomenon both centrally and peripherally to look at the effects of immunocompromised and non-immunocompromised areas.

AUDIENCE: Thanks for your talk. Really great work. You are using supercapacitors and some energy harvesting tricks for the power management of some of your systemms. So I wondered if you could talk about that. WALTER VOIT: Sure. Sure.

So there are a lot of competing approaches to getting energy inside the body. A lot of the devices that we've done preliminary research with are still wired devices, so we use omnetics connectors, and we have cables that are bonded out straight to our recording instruments. For the first generation of wireless devices, there are a host of competing technologies across the country.

Purdue is doing some really neat stuff in that area. There is some neat stuff that's being done with optogenetics and using power to power lasers, and then using lasers to communicate with light. In terms of the supercapacitors in the batteries, we've chosen these lithium polymer batteries, these 10-milliamp power lithium polymer batteries because they've got a small footprint, we can charge them with fairly simple circuitry from inductive coils that we can pattern straight on to our plastics.

Right now for the larger devices, we're still using coils of copper or other metals. But in the future, miniaturized devices, we've got some really neat technology to photo polymerize and photo pattern very small antenna traces that can behave like much larger antennas straight onto our plastics, and that would be enough to then power some of these implantable batteries. Great Batch is a big company. They just moved their headquarters to Dallas a couple years ago. But they've got some pretty sophisticated implantable battery technology that powers a lot of the biomedical implants that are on the market today. In terms of the super capacitors, we've done less work so far in integrating our devices with those, so I can't comment more on those right now, at least publicly. But for the devices we have, the battery powered ones are the ones that are working the best that are fully wireless.

AUDIENCE: Can you talk a little bit about moisture barrier and encapsulation? That's always one of the issues that everybody runs into with implanted electronics, and particularly flex electronics because you don't have it in a hermetic enclosure. WALTER VOIT: Absolutely. So as polymer chemists, we spend a lot of time thinking about how to protect and properly package these electronics. And we have a number of different approaches we're exploring.

So the most basic approach for a lot of these devices you've seen is a thin film parylene c coating, and that ranges anywhere from 800 nanometers to 10 microns in thickness of that parylene. The problem you get when you get parylene up to 10 microns in thickness, which is sort of enough to prevent pin holes and a lot of moisture, is that size begins to dominate the mechanical properties of your device. In a lot of cases, these electronics that we build are-- the whole gate stack is 10 to 25 microns thick. So if you put 10 microns of parylene on either side, that's dominated everything. So Rommel's actually been heading up a research into low stress nitrides.

We're doing work with silicon nitride. We've done some work with silicon carbide. We're doing work with composite stacks of materials to try to, in a very thin film way, prevent shunt impedance and prevent moisture from getting into our electronics. The parylene encapsulation has been good enough for our subchronic implants, so lasting three months or six months or so. As we move into the multi-year time frame, though, it will be probably more combinations of these composite approaches. We've also designed our substrates, though, to be pretty good moisture barriers.

We can tune the hydrofelicity, or the hydrophobicity of our polymer substrate. We can position these electronics into the neutral plane of our device. So we've got, let's say, 10 microns of our shape memory polymer on the bottom and on the top, which has been tuned to be a very good moisture barrier. Then we've got our gate stack that has other very thin film encapsulants. A problem with things like these low stress nitrides is they become brittle and they don't have a lot of strain capacity. So if you want a material that's flexible and bendable and stretchable, you're sort of at odds with using ceramic interfaces, basically. So we try to accommodate that geometrically by building highly serpentine traces where we can accommodate sort of longitudinally or torsionally a little bit of that strain so that these thin film ceramics don't crack. And that's, I think, one of the big challenges in implantable flexible electronics is to be able to move away from hermetically sealed tin cans to devices that are small enough and soft enough to interface with biology over long periods of time.

And I don't want to say for a second that we've solved that problem, but I think we're attacking that problem from a lot of different angles. And I think it's going to be a lot of combined solutions that get you a little bit of the way there that, in total, get devices that are going to be good enough for subchronic and potentially chronic implants. Thanks a lot for coming in. And we appreciate your time and attention. [applause].

Why Do I Need To Be Induced With Gestational Diabetes?

Hello, I am Ty Mason of, researcher, writer and I have type 2 diabetes. Today I want to talk about gestational diabetes and induction. After you watch the video…

Views: 1 219 By: Diabetes Healthy Living Channel
Occupational Therapist: Christine Khelfa, CHT, OTR/L

Hi, my name is Christine Khelfa and I'm an occupational therapist at Duke, and I work in the hand therapy clinic. I decided to go into this field because I like helping people return…

Views: 210 By: Duke Health
Phrenic Nerve Surgery

Hello and thank you for tuning in to the program today. I'm your host Neal Howard here on Health Professional Radio. Our guest today is Dr. Matthew Kaufman, he's the only known…

Views: 417 By: Health Professional Radio
Occupational Therapy FAQ

Hi I'm Ashleigh, an Occupational Therapist here at Therapy Playground. And I'm Kim, I'm also an Occupational Therapist at Therapy Playground. Today we're going to discuss…

Views: 97 100 By: TherapyPlayground
Pinched Nerve (Cervical Radiculopathy) Stretches & Exercises - Ask Doctor…

Hey everybody it's Doctor Jo, and today I'm going to show you some stretches and exercises for cervical radiculopathy. Let's get started. So if you have this, you know is…

Views: 127 877 By: AskDoctorJo
Audio Conference Shoulder Dystocia

Kate Reback: Thank you very much for attending HRSA’s third risk management audio conference of 2011 titled “Shoulder Dystocia: Managing Clinical and Legal Risks.” My name is Kate Reback…

Views: 2 730 By: HRSAtube