>> Desney Tan: John Rogers is here from UIUC. He does nano and molecular scale fabrication materials, patterning and builds some really crazy electronics as you will see today. He has got a chemistry/physics background and in fact has a physical chemistry PhD out of MIT. He spent time at Harvard Bell Laboratories where he was in the Condensed Matter Physics Department and in fact ran the department for a number of years. He is now the Swanlund Chair in the Materials Science and Engineering Department at UIUC with joined appointments across just about every department you can imagine. I am not going to list them. His publication patent press awards list is massive and again you can look at his bio for a small snippet of that. He has also had multiple startups starting with his PhD work and recently with a company called MC10, which I don’t know if you are going to talk about. >> John Rogers: A little bit yeah. >> Desney Tan: But if you look at this project list as well he has done everything from stretchable lithium ion batteries and silicon integrated circuits to flexible silicon [indiscernible], electronics on balloons for surgical catheters, cellular-scale optoelectronics, negative index metamaterials, and the list goes on and on. The thing he is going to talk to us about today is in the bio-integrated electronics space, specifically epidermal electronics. So rather than get in the way of a great talk I am going to hand it over to John. >> John Rogers: Okay, thanks a lot Desney for that introduction and for the invitation to come here and spend some time at the labs. It is bringing a lot of good memories back from my time at Bell Labs, sort of a research lab in a company that makes products. So I think it’s a great environment to do innovation. I will tell you about some of the things we have been doing at Illinois recently that may be of interest to hardware interfaces to the body of relevance to future computing, health, wellness, sports and other things. And it’s referred to, this is a term we made up, sort of epidermal in the sense that the electronics themselves have physical properties pretty well matched to the epidermis. So they go on in a way that’s more or less invisible to the person wearing the device, which we think is important for an intimate high functionality interface to the body to have that kind of wearability. So I will start by giving you motivation and what we are trying to do with perspective on how this kind of electronics kind of relate to more conventional stuff. I will give you some ideas around requirements in materials, mechanics and manufacturing in order to do this kind of skin integration. And then the last half, maybe the last third or so is just step you through a bunch of examples of things we have built with the idea that this is still in the early days. And I think there are a lot of opportunities for new ideas and higher levels of functionality, but just give you a snapshot of where we are right now in the development of this kind of technology. So let me just start by telling you something that’s obvious, you already know it, but it’s a useful reference point which is the future of electronics where it’s been in the past, where it is now and where it’s going into the future. And it’s really all about making things smaller, faster and cheaper. So in the old days you had giant boxes that filled desktops or filled rooms and were used primarily for specialized industrial academic computation, but Moore's Law miniaturization in the transistor sizes, reduction in their cost and increasing levels of functionality. Really transformed the hardware into completely new device formats and that’s an engineering accomplishment. Maybe the more significant one is that it introduced a qualitative shift in the way that computing devices are used in our every day lives. It’s much more ubiquitous, much more personal as communication tools, productivity enhancement, systems and entertainment devices. And if this is where we were, this is where we are and were are we going in the future? It’s a pretty easy answer: You just go to the semiconductor map and it tells you exactly where this integrated circuit technology is going to be 5, 10, and 15 years into the future. And if you think about folks who are in electronic materials, electrical engineering and thinking about integrated circuits, doing research on them, the vast majority are working on scientific problems around that scaling. And that’s a very valid set of problems to be thinking about because this is a trillion dollar global industry and there are a lot of interesting fundamental science problems in the service of that enormous market and enormous impact. It’s not the future though that we are thinking about these days. We are trying to do something a little bit different and complementary, not competitive with that future, but something that might open up a range of opportunities that are currently unavailable because one thing that has not changed over the years is the platform for all of this electronics. It’s always been silicon wafer, which is a wonderful class of material in a wafer form. It’s a wonderful type of substrate, but it has certain geometrical constraints that make it tough to say, “Take your iPhone and wrap it onto your brain, melt it into your skin, or envelop your heart with it, because the wafer is rigid, brittle and planer. That’s fin for lots of applications, but there are some that require intimate integration with biology that cannot be addressed with commercial integrated circuits that exist today. And the idea is that if you could do some of this stuff and bring all of this power to bear on problems of human health or brain/machine interfaces that could lead to another qualitative shift in the way that the technology is used and expand out some opportunities there. So that’s a 30,000 ft view of what we are trying to do. Let me pound that message home a little bit further in the context of the brain. And we have done a lot on the brain, the heart, and I will focus this talk in the skin, but the brain is a good illustration to further help me make the point that I am trying to make. So if you think about the brain it is a very sophisticated electrical system. And if you want to do therapy on the brain, you want to diagnose abnormality in the brain, or you want to establish a brain/machine interface the best way to do that might be to take the most sophisticated man made electronics system, which is Silicon [indiscernible] Electronics and interface it with the brain. But that’s not easy to do for the reasons I have just told you. The wafer itself it planer, it’s a huge geometry mismatch. What is not quite as obvious is that there is also an enormous mechanics mismatch because silicon is a very stiff hard material. It has modules of about 150GPa. The brain is a soft squishy piece of Jell-O. Its modulus is about 5kPa. So you have geometry and mechanics problems there. So what do people do today? This is the work horse for the last 20 years. It’s really a pretty nice piece of technology and it’s been very useful for a variety of neuroscience studies. And it accommodates that geometry mismatch through a micro-machined array of doped silicon pins that penetrate down into the tissue on one side and on the other side terminate on a flat platform where you can mount a chip that you would dice out of the way and establish the electrical interconnect that way. And you take that integrated thing and with an air hammer you mount it onto the surface of the brain. These different pins can penetrate to different depths in the brain, thereby solving the geometry mismatch problem, but you damage the brain upon insertion. Then what’s worse is that over time the interfaces between these hard shafts of silicon in this soft Jell-O degrade, because the brain is not static in a mechanical sense. It’s swelling, de-swelling, it’s rattling around in the skull slightly and now you have a pin in a moving bowl of Jell-O and that’s a bad mechanics interface to think about. So this is great and it will likely remain in wide spread use. But what we are thinking about is what could you do to enable a future in which the electronics are configured to match biology? So soft in a mechanics sense, shape, conformal and then made out of bio-compatible materials. So that’s a pretty clear example. The heart as well where you can kind of appreciate the motivation. But the same constraints and considerations apply to skin as well, because the skin; although not as soft as the brain and not as reticulated in geometry as the brain is rough, stretch and flexible. So what’s done today usually involves point contact with conductive coupling gels, tapes, individual wires that run to a separate box of electronics. So electronics really not integrated with the skin. It’s electrodes and this is great for a hospital or research environment, but nobody is going to wear that long-term because you have got wires running all over the place. And then also this tape interface to the skin is irritating over time for reasons I will get into in a few moments. So, it’s not great for long-term wearability and also not great for improving the sophistication of the interface. Because as you add electrodes you add wires and pretty soon you are swamped with wires. And it becomes a difficult scaling, which is a shame because the electronics themselves are racing ahead according to Moore’s law, but then your interface to the body is stuck in this kind of old approach. You can get better and there are commercial devices and are more sophisticated then this obviously. And this is kind of one example where you have bare dye or individual chips mounted onto flex PCB. So a tape of [indiscernible], you have adhesive on that and you can put it on the forehead in this example, but still it’s not great for the same reasons I just mentioned. You eliminate some of the wires, but you don’t eliminate the mechanics problems at the interface which are driven by the fact that the skin is soft and stretch and even a tape is only flexible, it does not stretch. So it’s constraining the natural motions of the skin and that’s kind of a problem. So if you think about that set of circumstances you might conclude, “I have got to use something other than silicon if I want to do better”. And we have been working in this space almost 20 years by now and we spent a lot of time looking at alternatives. And the semi-conductor is probably the most challenging materials component of an integrated circuit. So you can think about organics like polymers, small molecules, single crystals, maybe carbon nanotubes. I don’t think graphene is interesting tubes. Let’s say these classes of materials and you might say, “Okay, polymer might be a good choice for a semiconductor because it’s already kind of soft, bendable, whatnot”. We have spent a lot of time on this. We have built flexible displays like this one in the early days. The problems with polymers are that their electrical properties are lousy. So you can do a slow nonemissive display like this. If you want to do a liquid crystal display that’s a lot tougher and if you want to make a high-functional interface to the brain that’s out of the question. So the topic might be then, “I need a synthesized new polymer so I can move this metric for transistor performance in terms of the mobility and centimeter square per volt second from this regime up to something more comparable to state of the art in organic electronics”. But that’s a large number of, you know, that’s orders of magnitude is the point. So you have got a huge gap and how you are going to span that gap becomes a topic of research. A lot of folks are still working on that. We sort of moved away because that just seemed a little bit too daunting. And instead we come back and ask the question: Is silicon really ill-suited for this kind of thing? Certainly in the wafer format it is and certainly the intrinsic material properties are not well matched, but could you do some simple things to effectively bypass those apparent limitations? And that’s where we focused and that’s what I will talk to you about today, but not at a super deep level. I will just give you a sense of it so I can skip over to what kinds of devices are enabled. So if you think about the wafer as I already mentioned and as you know you can’t bend it. And if you try to bend it, it will crack very quickly. Why is that? That’s because partly the material property. So you think about the bending stiffness and the degree of bendability as defined by the peak strains associated with bending given radius of curvature. It depends on the intrinsic elastic properties of the material. That’s the Young’s module and you are stuck with that if you want to go with silicon, but it scales with thickness in a pretty straightforward, easy to understand way. It is a really strong scaling. So the bending rigidity or the flex rigidity bending stiffness is the product of these two things. So what that means is that if you think about plotting the bending stiffness as a function of thickness in a [indiscernible] plot you have a power law behavior. So for a wafer that might be a millimeter thick the bending stiffness is maybe 10 Newton-meters, but if you take that wafer, you shave it down to something that’s sort of nanoscale in thickness, so this is 20 nanometers in thickness, you have reduced the bending stiffness by up to 15 orders of magnitude. So now it’s 10 [indiscernible] Newton-meters. And although that’s just linear elasticity Newtonian mechanics, there is nothing quantum about that, whenever you change a material property by 15 orders of magnitude that changes qualitatively the way you sort of think about it. And in fact silicon with that thickness is floppy, very, very flexible as illustrated with the SEN [indiscernible] image, in the sense of a low bending stiffness and also in sense of degree of bendability. So the peak strains associated with bending are going to define how far you can bend it before you crack it, because silicon can only withstand about a one percent tensile strain. And what you see is for a given bend radius R that bending strain goes down linearly with thickness. So not quite as strong as scaling as the bending stiffness, but nevertheless thinner is better, so very flexible, very floppy. Now, if you think about silicon sheet with that type of thickness it’s great, it’s bendable, but it’s incredible fragile. If you touch it, it will shatter. It’s just not a sensible platform for an integrated circuit in the same way that a silicon wafer is. In fact that’s why people use wafers that are this thick. It gives it some robustness so robots can move it around and you can put it on a PCD for example. So the next question is that maybe you think about this as a building block material for an integrated circuit that you would build on a sheet of plastic or a piece of rubber. So that brings up the next challenge then which is: How do you go about gluing a piece of silicon to a piece of silicone rubber? And that’s a hard problem because the material properties are so different. The thermal expansion coefficients in particular are very different, a hundred or thousand times different. As a result it’s hard to come up with an adhesive that would keep a silicon chip bonded to a piece of rubber in kind of a realistic way. But there again thickness comes to your rescue, because if you think about what drives interface cracks it’s this parameter, which is the energy release rate. And you can see how it depends on the differences in thermal expansion coefficients of the two materials. Also the change in temperature is driving interface stresses, but here again it’s scaling in a nice way with thickness. Which means that as I decrease the thickness of my silicon the energy release rate goes down the propensity for a crack to open up is diminished. So it relaxes a lot of constraints on what kind of adhesive you need to think about in order to keep these dissimilar materials bonded together. Here’s an example: So this is a very thin plate of silicon. We have printed it down onto a micro-machined ridge on a plastic sheet. There is no adhesive at all right here. This is just van der Waals forces keeping it in this cantilevered state partly because of that scaling. So think silicon is what we like, its floppy, it’s very flexible and it’s sticky in some sense. So you can do this kind of heterogeneous integration. So those are nice features. Thin silicon looks pretty good, but not good enough because this kind of characteristic is not what you want or you need if you want to get on the human body. So this is bending, what you really need is something that both bend and also stretch, because if you only have bendability you can only wrap cylinders and cones. You can’t even wrap a sphere, much less a brain or heart and you can’t accommodate the kind of complex motions that organs are undergoing all the time either. So you need stretching and there is no thickness to which you can make the silicon that will impart stretchability. So that may seem like a daunting problem, but even that one is really easy to overcome it turns out. So if you take this thin silicon and you bond it to your elastomer substrate, you know ultimately you want something that stretches. You probably need to use a stretchy substrate. So you stretch out your silicon rubber and then you use surface chemistry such that upon contact this silicon will bond very well at the interface with the rubber. Such that when you relax the pre-strain it induces compressive forces on the silicon. And what happens is it will cause the silicon to spontaneously buckle into that type of shape. And that’s pretty nice because it’s an accordion bellows made out of silicon on silicone. And in particular if I stretch it out now I am not stretching the silicon itself, but I get effective end to end stretchability because I am trading out of plane displacement for in plane deformation. So the amplitude goes down; wave length goes up if is stretch it out. And then the reverse happens when I compress it back down. So that’s simple and it works provided you understand what’s going on with the mechanics because this kind of thing is a hard soft composite. You have to know where the stresses and strains are if you want to do real engineering with these kinds of ideas. And we spent a lot of time on the mechanics, wrote entire papers just on that type of structures so we understand what’s going on. So that’s the idea, thin silicon, buckling physics, off you go. You can make integrated circuits. This is not an integrated circuit, obviously. It’s just a test structure and it’s not even a good starting point for an integrated circuit because you are never going to do device processing on rubber; it’s just not going to happen. But, you can make ultra thin circuits fully formed, flexible and then as a back end packaging step you bond them to this pre-strained rubber. You create all this buckle wavy shape and now you have a composite that has the soft mechanics you want. And this is an example of one of these stretchy integrated surfaces based on silicon. This is a picture taken at our Beckman Institute. We have an imaging technologies group and so they made a nice image here. So this is a glass pipette, they are just pushing down on the center of our circuit and you can see how it is able to form in a continuous curve linear way. You can see the different wrinkles and the interconnect traces. And here you can see that particular trace is stretched out by this local force. So all the wavy structure that was originally present has been pulled out. You have pulled your accordion out all the way out to its full extension. If you kept pressing you would probably crack that line. So the key is being able to understand what the range of defamation that your devices can accommodate and do the design up front so that your application requirements can be satisfied. So, circuit designed combined with mechanics designed to allow you to do what you want to do. Now the key thing is that you haven’t sacrificed any of the electronic properties by doing all this stuff. So its thin silicon, its wavy circuits, but the electronic performance is very much more comparable to what you see in a wafer based system. So, factor of 1,000 better than what you can do with the organics, very nice switching characteristics in both types of transistors and this is just a simple ring oscillator. So you take in this stuff, you crumpled it up and yet you have done it in a systematic way with understanding of the mechanics so you haven’t disrupted the electronic function. And that’s the key part of this obviously. >>: I have a question. >> John Rogers: Yeah? >>: Do those electrical properties say constant when the material us under mechanical mode? >> John Rogers: Yeah, so that’s a good question, in this geometry, no. So there are actually finite strains at the device positions. So this is a p-channel/end-channel MOSS fed and it’s a ring oscillator. Mobility or the switching characteristics of the transistors change slightly with strain. That has been exploited by the Semiconductor industry to increase the speed of integrated circuits. At a modest level I would say that kind of a 20 percent type shift in the output currents. So in some cases you can live with that. In some cases you cannot and in cases that you cannot you actually take this circuit and you can do things to it to strain isolate the active devices. To eliminate that dependence, but it’s coupled in this very simple implementation. But, I won’t step you through. The basic idea is thin materials, hybrid hard/soft composites and buckling and from there all kind of elaborations and different things that you can do to accomplish strain isolations. It’s a very good question, but you know you can side step, it’s an engineering problem. So if you can make soft circuits that way then what do you want do with it? And I outlined at the beginning what we have at mind and that’s putting things on the body. And it could be inside the body; it could be on the body. We got interested in skin mounting maybe four years ago I guess. And we started to think about what’s out there today that goes on the skin unlike these tape on electrode wires that is comfortable and works. And we thought about kids temporary tattoos as maybe a model to think about how to do circuits on the skin because they look great, you put them on your skin, you move your skin around, it can kind of wrinkle, stretch, deform with the skin and you don’t even know it’s there. So that’s what we set out to do. To think about whether we can make an integrated circuit and sensor suite that has that kind of mechanics or something as close as possible to that. And what does that mean? That means as thin as you can go. We had a target of 5 microns. We got as close to that regime, as light as possible in an areal mass loading sense, maybe a milligram per centimeter squared and as low in modulus you can go. We had targeted something kind of brain modulus level 5kPa, but the skin is actually much stiffer. It’s about 100kPa, so 100 would be good. We went a little aggressive on that, but 5kPa. It needed to be at least as stretchable as the skin itself. So skin will stretch to about 30 percent before it starts to tear. So we wanted something that had that kind of elastic level stretchability. And then some additional requirements: You need your sensors and your electronics electrically isolated from the skin. So it needs to be waterproofed, but at the same time you don’t want to disrupt natural processes of transepidermal water loss, so you need some level of permeability as well. So that’s we had hoped to do and there are a lot of challenges in getting anything on the skin and they relate to issues in long-term viability in the interface and you have got to have access. Adhesion is always a challenge, you know, you want something that’s soft and conformable, like a tattoo. And then cost, power, weight and size are also relevant considerations. This one is a big one and has been over the years, you know, how do you do robust bonding or binding without irritation? And it turns out that problem is directly related to the mechanics goals as well. So I am not going to go through all the details. Let me just highlight the key points. So if you think about modeling the mechanics of a piece of electronics, just a generic material bonded to the skin you can run different scenarios in terms of the modulus of the electronics and its thickness. And you can calculate interface stresses in particular that tend to drive delamination. This can be sheer stresses or peeling stresses and you can do it for different scenarios. A silicon chip glued to the skin, flexible electronics or plastic sheet on the skin, thinner, lower in modulus or all the way to this kind of tattoo like construct that I put forward in the previous slide. And what you see in both cases is peeling and sheer is that the stresses peak up at the edges. So if I stretch on the skin delamination will start at the edge, which is sort of intuitive, but then look at the magnitude. So the magnitude for otherwise similar level of deformation for the silicon chip is large, smaller for plastic sheet and smaller still for this skin like and same thing over here. But the actual magnitudes are what notable here. So if you think about the ratios of the stress, silicon chip to plastic sheet for skin tension or under a bending deformation going from silicon chip to plastic chip helped because it’s reducing interface stresses, but not by a huge amount, by about a factor of 4 to 5. Now if you think about the ratios though for silicon chip to this electronic skin, this skin-like scenario now you are talking about 5 orders of magnitude. And that really changes the way that you think about adhesives to the skin to the point that you don’t even need adhesives at all. These things just cling to the skin because they are so soft and deformable. And so you can see that it’s easy to do that experiment just to see the consequence of that mechanics. So this is the same chemistry in an elastomer, a silicon based material. All we have done is we have changed the density of cross links between polymers in this material to change its modules, the same thickness, same interface chemistry, just different modulus. And you can laminate it onto the skin; it’s sort of near the wrist. And then you can flex your wrist and you can see what happens. No adhesive, this is just a tackiness of the material itself keeping it bonded. So here’s a silicon material with a modulus of about 1MPa. So you flex your wrist and it delaminates. You stretch the skin and again it drives the delamination front that begins at the edge. Now if you do the same exact experiment with --. Let me see, can I get rid of this? Now with lower modulus here’s what happens. So same thing, same region of the skin, same interface, chemistry, same adhesive strength, it doesn’t delaminate. And that’s just because you have minimized the interface stresses. And it sticks now, just clings and you can move your wrist around and it stays put. So that’s the basic --. >>: Sorry, what was the change between the first and the second? >> John Rogers: Modulus stiffness, just softer. >>: So what about isolated clusters of microneedle and nanoneedles to give you a --? >> John Rogers: You could do that, yeah penetrating pins, those are options. We were trying to get the thing stuck on the skin without any straps, pins or tapes. But you could add that to improve the robustness of the adhesion. You could definitely do that. So how do you get something that has very low modulus? So I told you some tricks to get stretching, but now you need stretching and also a very low effective stiffness and it’s a tricky problem. You want a device that has a modulus of let’s say 100kPA, but you want silicon in there and that modulus is 150GPa. So, how do you do this kind of composite engineering? And this is one way to do it: So illuminated by full 3D finite element modeling. So you take a circuit, an ultra thin one like I mentioned before with silicon nanomembranes, but don’t use it in the form of a continuous sheet. Cut it into an open spider web mesh like this in terms of these intersecting filamentary, serpentine wires and devices built into that. And then bond it to a low modulus silicon substrate. Think, you know, 30 microns in thickness and 50kPa. If you do that fully optimized with modeling guidance and then you stretch it along this direction or along this direction you can measure stress/strain behaviors. And that’s telling you how stiff it is. So the first thing you see is that the measurements along X and Y are quantitatively captured by the modeling. The other thing you see is that the modulus is very much comparable to the epidermis itself, which has a modulus that’s slope is about 160kPa. And you can see it starts to go non-linear around 15 to 20 percent strained. So this is actually a better material in some ways. You can stretch it further, but in this small strain regime it’s very well aligned with skin itself. So that’s a framework into which you can drop transistors, integrated circuit sensors and so on. So this example of a field effect transistor there is a resistive trace of very thin silicon cut into that geometry. So here’s what it looks like. The transistor performs well even though it’s in this funky configuration. You can do sensors, these are strain gages, and you can do [indiscernible] detectors, other kinds of things. We have done many, many classes of devices in this context. So here is what it looks like when it’s mounted up onto the skin. It’s very soft, no penetrating pins, no adhesives, you just laminate it on and it’s good to go. It’s the same kind of experiment. This is laminated on the skin of the post doc who did the work and then we poked him with a glass rod. So you can kind of see how that moves around. If you grab it at an edge you can peel it off. This is what it looks like. It is so flexible, floppy and soft that it doesn’t even support its own weight, it just kind of collapses down onto itself to give you a sense of the mechanics. The practical --. Yeah? >>: So a question: are spin coding a assume silicon over [inaudible], is that the substrate? >> John Rogers: Yeah, so these filaments are [indiscernible] and then in certain regions there are silicon, metals and dielectrics to make the devices. >>: And then you are taking that substrate and probably doing it on a wafer, pulling it off and then --. >> John Rogers: substrate. And of processing. much. It can’t question. Exactly, yeah, and then transferring it to the soft the key there is to keep the elastomer out of any kind It’s not going to happen because it moves around too get registration from layer to layer. Yeah, good So in terms of this mechanics the practical question is: How do you hold this stuff and manipulate it around? And there we just use water soluble tape. So PVA, put the circuit on there, flip it over, put it on the skin, wash away the tape and that constitutes the mounting procedures. I have a movie to show that. So this is just a demonstration platform that has a number of different types of active devices; transistors, diodes, inductors, capacitors, LC oscillator, temperature gages, strain gages, there is an inductive coil and then an antenna that runs around the outside. So what you are seeing is this transparent tape backing allowing the student to place this circuit directly into contact with the skin. So its skin circuit, silicon rubber backing and then that PVA tape. And he is washing away the tape by applying a little bit of water. And eventually that will disappear and just leave the circuit right here behind and the circuit substrate which extends a bit further out around the periphery here and you can kind of see that. So once it’s gone then it’s just this epidermal electronics on the skin. And you can see then by pinching the skin you can get a sense of what the mechanics is all about. So here is just the silicon rubber backing. Here is the rubber backing plus the circuit demonstrator. And as you pinch this thing what you see is that the skin wrinkles and folds into forms in the region where the circuit is located almost exactly the same as the way it deforms in regions where the circuit is not located. So that’s sort of validating the design approach. So the circuit is not constraining the skin or preventing it from behaving in a natural way. And as a result you put these things on and you don’t even know they are there and that’s with no adhesive. So you can squeeze it, move it around and the interface is strong enough to stay bonded, but not in an absolute sense, it’s not strong bonding. So you can just grab it at an edge, you can peel it off and that’s what you are seeing here. >>: Can you stick it afterwards? >> John Rogers: Yeah, so there are a couple of ways to do that. You can come in with one of these water soluble tapes, stick it there and then peel it off. It will come with the tape and then you can do the mounting again. But in this case you could not reuse that. [laughter] >> John Rogers: So the absolute strength of the bonding is not great, but if you want to do better one way is to just use existing temporary transfer tattoos as substrate. You already have FDA approved glue mounted on the back and then go through the regular thing. So this is an [indiscernible] pirate, just somehow pirates in tattoos, but there is no ocean near Illinois, so this doesn’t make a lot of sense, but in the context of tattoos maybe. So you flip it off. And now the additional benefit of this is you can conceal your electronics. So if you are into covert things, we work with CIA and so on that turns out to be a relevant capability. So the circuit is kind of right there in the facial region of the pirate. So I used to travel around with these circuits mounted on my skin just so people could see it, but it requires students to make lots of circuits. So I don’t do that anymore; I have given them a break. >>: [inaudible]. >> John Rogers: Yeah, so you can. I will come back to that in a second. There is a finite time over which you can stay over the skin and remain viable in terms of function. I will describe that in a second, but instead of wearing them I just carry some representative examples around. So this one is one of these kinds of circuits, but embedded in about a millimeter thick sheet of rubber. So this is not epidermal, but you can handle it, kind of twist it around and get a sense of what’s going on. The other one is one of these epidermal constructs on PVA so the mechanics here are dominated by the tape. And just to give you a sense of what it looks like I will pass these around. So to your point: How do you get these things on and how long do they stay on? So a tattoo is not bad, just van der Waals force is okay for an office environment, but if you go swimming it’s coming off. So how do you get it to say on better is something we started to think about? And what we found is a very simple way to get very robust integration with the skin that does not compromise this epidermal construction is to just use spray on bandage. So I don’t know if you guys have ever used it. You go to Walgreens, you get a little spray can of this stuff. It’s a copolymer of an acrylate and a silicone. You can spray it down as an adhesive and then you can spray it down on top as an encapsulate. In that case you can print just the circuit, so no silicon backing at all. It’s just the circuit printed off of a silicon stamp and that’s what’s shown in this video. It’s a little more advanced way. So there’s the spray on bandage, so a few squirts. That’s only about a micron thick sheet of that spray on bandage material, but it’s a nice adhesive. You print your circuit right on to your skin like that and then you sprits it up with some more sprays on the bandage on top. You let that dry and now it’s much thinner than it used to be and it’s also much more robust from a mechanical standpoint. So here is a video we have. So now you can go swimming, you can take showers, you can exercise, you can do essentially whatever you want and it won’t come off. Typically after a few days in the morning you add on a little spray on bandage to maintain the integrity of the encapsulating layer. And just to give you a sense of a little bit bigger piece of electronics designed to do something a little bit different then the one I showed you before. This is actually a functional think and you can wash it and do your thing. So time frame: These devices will say on about two weeks. That’s about it and it fails not because of a deficiency in the adhesive or the encapsulating layer; it fails because the skin is exfoliating cells from the stratum corneum all the time. And the timescale for that exfoliation is about two weeks. The entire surface of your skin is gone in that time scale. So you are not sitting on a stable platform. It’s shedding those cells and your circuit with it. So that’s a fundamental biological process. I don’t see a great way around that. You can imagine taking the device off maybe for a day and then you could reapply it or something like that, but that’s just the practical reality of where these things are. So this is what it looks like on the skin. This is spider web stuff and the key thing is that it’s conformal to all of the very challenging topography of the skin except the most daunting crevices. And that’s good from the standpoint of adhesion, but it’s also good in terms of the interface to the skin. So the skin is a window into health, wellness and physiology of a person wearing the device. So you can imagine put a cell phone in this format. I think it’s more compelling to think about applications in wellness, sports and healthcare. So that’s where our emphasis has been and this kind of contact insures you can make good measurements. And you can make those measurements on any place on the skin, like places where you would not be able in a practical sense to have a wire and a piece of tape. So like a finger, back of the neck, forearm, and jaw. So if you come into the group you have to be willing to do this kind of thing. And then you can make all kinds of measurements. I will come back to the details of what these measurements are, but there are a number of different measurement options we have shown. And one of them involves just measuring electrical activity associated with muscle contractions, so either the heart, skeletal muscles or muscles that are controlling your eyes or throat. I will show you speech and I will show you examples of all of those. This is the heart, so you can measure EKG with high fidelity and the detailed waveform there is saying something about health status if you put it on the chest. If you put it on the leg you can measure EMG, that’s electrical activity associated with contraction of the skeletal muscles. So you can measure things related to gait for example. So the subject is walking and then standing. This is that epidermal electronics here. Here is the conventional Pace non-electrode with conductive gel just to do a comparison. So we are not claiming that the fidelity of the measurement itself is higher or comparable, but it’s different because you don’t have a piece of tape. You have something matched to the skin and you don’t have just an electrode you have full electronics. So it’s really a lot different, but the fidelity is about the same. >>: Was there circuitry on the device that was actually filtering [inaudible]? >> John Rogers: Not in this particular case. You can do that and we have done that, but not in this particular example. >> How did you get the wire off of that non-electrode? >> John Rogers: I will come back to that in a second; very good question. Then if you put them on the forehead you can measure EEG and you can see all the standard things. You can do Stroop tests and you can look at alpha rhythms. And so if you put it on the head you might think about a brain machine interface, but these devices because they can go on unusual parts of the body you don’t have to think just about brain machine, but just more generically human machine. And we found that if you put these devices on the throat area you can pick up complex patterns in EMG associated with muscle contractions involved in speech. So if you look at the spectrogram of data that you collect when the subject is saying different words, either vocally or sub-vocally, you can see different patterns. And we work with folks in electrical engineering to do pattern recognition so you can train the system on a finite vocabulary of words and then use that as a brain machine interface, maybe for someone who suffers a disease of the trachea or as a prosthetic control mechanism. My students wanted to develop like a game interface so this was a very simple example. This was a couple of years back, but you could just move a cursor up, down, left or right. I have never played this; it’s some kind of strategy game, but just with your neck. So if you want to play a video game with your neck we have the technology for that. So it may be interesting, but maybe not that compelling, but anyway it’s a demo. We got a little bit better over the last year and a half or so, so that things can happen faster in terms of the interface so it doesn’t have to do be just a strategy game. It can be more of an action game. And there are companies out there now trying to do things like this. The hardware is totally different, but let me just show you that you can do this by EMG control. So this is a helicopter drone. Actually if you just out the lights momentarily then this movie shows up a little better and then we can go back to the lights on. So this is a post doc over here so it’s by manual control. He has these epidermal electronic patches on both forearms and then there is a computer system that takes that data and then translates it into a set of commands for the helicopter. And you can see that go here. So the helicopter, he will rotate his fists and the drone takes off. And then he is tilting his fist this way and it rotates in that direction. He will stop that, it will cease rotating. Then he clinches his fists and it will fly off in that direction. Rotate the other direction, it will land and then he will rotate again. It will take off and he will tilt back the other direction and it will rotate counter clockwise in this case, stop and then he will clinch his fists. And he is trying to fly the helicopter back to try and get it to land in that box. And he will cause it to land and you can see he missed. So he needs a little bit more practice, but you get the idea. It’s just a demo, academic level demo. But we think that these devices could be useful in a more commercial setting and for that reason two or three years ago we put in place a startup company in the Cambridge area. And they are trying to push these kinds of ideas out more into the real world. And they are interested as one application area of sports, wellness monitoring and as an example of a sporting event to the extent that you think about NASCAR as a sporting event. We put devices on this guy. He is a really interesting guy, an undergrad at Duke. So he is a really young driver on the circuit, but he is interested in advanced technology. So we put these things on and he races. He has done it two or three times now. And it’s a great proving ground for the technology because it’s incredibly hot in these cars. There is a lot of sweating and it’s an endurance type of thing. So that’s been a great interaction. We also have as head of the sports segment at MC10 former NFL guy Isaiah Kacyvenski. He is the highest draft pick ever out of Harvard. He played for Seattle Seahawks actually for eight years. He played in the Super Bowl in 2006 and he is a symbol to sports advisory board. So, usually in a small company you have a technical advisory board, we have that and then we have this other cool thing; sports advisory board. And you have got all these famous guys who are part of that, providing input on what kind of function you would want. It’s not just lending their name, but actually really intellectual input to what MC10 is trying to do, which is pretty cool. So MC10 makes these kinds of patches now, here and here. And they launched their first product in July; July 3 was the launch date. It’s a joint offering with Reebok and it’s a piece of flexible electronics that goes in sort of the brim region of a soft good, a skull cap, that measures in a very precise way three axises accelerometry and rotations. So it can quantify the detailed physics associated with an impact to the head and then it uses advanced algorithms to take that data and then bend it into three categories, moderate hit, severe hit and then a hit that could represent a concussive impact to the head. And the idea is to provide additional information to folks on the field to determine a path of action after a hit to the head. And Matt Hasselbeck who used to be a quarterback, starting quarterback for the Seahawks, then the Titans and now he is a back up for the Indianapolis Colts has been a real evangelist. So this was kind of publically announced at CS in January. That’s Matt with the device and then there is our device right there. So that was pretty cool. We got that on a throwing arm of a starting NFL quarterback. milestone. It was kind of a nice But this is a CHECKLIGHT, you can go and order them up if you want. You can go to Reebok website and you can buy these, so order early, order often, it’s a great piece of technology. This is what it looks like. It’s about 130 bucks and it keeps a running tab on the total number of hits in these three categories to the head during the course of the season. So it does overall dosimetry I guess, so to speak in terms of impacts. And this is the indicator system in the back piece of this. So that’s one example. We have a lot of other things going on at MC10. Our focus of my own university group is not, I mean we are interested in sports, but back to sort of health care. And you know the wires, straps; tapes and so on are problems for adults. It’s a real issue for premature babies that require a lot of monitoring. So this is just Photoshop, we have not put a device on a baby yet. But this is our vision of the way that this stuff could be useful. And to do this you need wireless, you need a radio, and you need onboard power or a strategy for that. And you need sophisticated sensors as well. Let me talk about sensors a little bit and then I will talk about interface and radios, wireless approaches. So we were doing this work and we were approached by an expert in skin surface temperature measurement at NIH. And when were approached I was thinking at the time that temperature is not that interesting, but it turns out it is. If you have ability to spatial map temperature and measure it at millikelvin precision. So instead of saying your temperature is 98.6 you can evaluate temperature to 98.653. And what’s happening in that last digit tells you a lot about health, wellness and what’s going on with the body. But you have to be able to do it with that kind of precision and also have some spatial mapping capability because the surface temperature at that level on the skin is extremely heterogeneous. Here is an example of how they do it now. So these are quarter of a million dollar very sensitive IR cameras that are imaging temperature profiles in space and time. And this is a close up image of a thumb. And what you see are these little dots, so each one of these dots is a black region. That’s a little bit cooler than the rest of the thumb and if you play this, this is in real time, and you can see that pattern of dots is constantly changing. The pattern and timescale, it’s almost periodic type of thing. Each one of those black dots corresponds to an individual sweat gland. It fires, a little bit of seat comes out, it evaporates and it cools locally. So that seemed pretty cool to me. Like if you could do that without the expense of hardware of the camera, without all the motion artifacts, so getting the thumb to stay still enough to do this is a challenge then that might be interesting. The thumb, you know, that’s kind of curiosity, but if you look at full body temperature maps during exercise you see, at least for me, things that are really remarkable. So this is an IR image of an individual on a treadmill and then the images are manipulated to keep these four points fixed. This is time so this is average body temperature varying as this person is jogging, kind of running and then resting. So, if you look at this variation and say, “Okay, that’s kind of interesting”, it’s not jumping out at you as something that’s high in information content. But if you look at the spatial patterns and again with high precision and spatial resolutions then you see some really funny things. So he is walking, he will start to jog very slowly. At this point he begins to sweat and you will see the overall surface temperature of the skin decrease because of that sweating. So you will see the body starts to go dark and then as it get’s darker and darker you begin to see an amazing level of texture that begins to build up on the body. And these lightning bolts there, that happens to be associated with vasculature, blood vessels that are transporting blood from the deep muscle regions up to the skin. And there is a thermal transport associated with that. So those are hot regions, it’s not sort of dripping sweat. That’s really telling you about blood flow and in turn cardiovascular stress and health associated with exercise. So that seemed interesting. So can you do that kind of thing with this epidermal patch technology? It turns out that you can. Let me give you a couple of examples. Here is one of these devices. Now in this case not only mass loading, but thermal loading becomes a very important consideration. And that was not on our list of metrics when we started this project. But if you want to put a sensor on the skin it can’t load the natural processes of water loss, sweat and it can’t heat as a result. So that becomes an important aspect and we spent a lot of time on that. But you can use the high end cameras at NIH. You can measure this way and you can measure that way and then you can look at the precision and the accuracy in this kind of sensor. And it turns out that without a huge amount of difficulty you can do this. So here is temperature as a function of time, measured with the IR camera and one of these epidermal sensors, just offset for clarity. So what you see is that when the patient is just at rest there are some periodic variations. This is associated with natural vasoconstriction and vasodilatation of blood vessels in the near surface region of the skin. There is characteristic amplitude and frequency associated with that which itself is interesting, but the purpose of this graph is to show you what happens and the kind of subtle things you can pick up. So here at this time the patient was told to do a series of operations in mental math. So division and subtraction and what you can see is you can pick that up, the effects of that because now your skin is sweating a little bit. And so you can see that cooling. So you can say something about cognitive state. The other thing you can do is you can really pick up subtle effects of physical activity. So here we are again, sensor IR camera, rest, now physical stimulus and not rest. So in principal you see the same kind of trend I showed you with the treadmill. Temperature goes down then it stabilizes, but this physical stimulus is very subtle. So you have one device measuring on the left palm and the physical stimulus corresponds to rubbing two fingers together on the right hand. So just to give you a sense of the precision and the kind of coupled response of the body are sort of interesting to me. So I am running out of time here so let me just skip through and let me just get to the last few slides. So there was a question about interface IO. So what we have done is things in two parallel tracks. One is let’s use a soft cable like this hard wire connection just so we can understand how the sensors are behaving. We can do thermal mapping, we can do EEG, we can do other things and we can also look at device lifetime. So here’s EMG, we are measuring on the forearm right after the device is mounted and then a week later where the subject is doing showers and all the normal stuff. You come back and it’s pretty good, maybe a little bit of increase in noise, but not too bad. So here this is a releasable cable, we laminate it on, make the measurement, peel it back and then off they go. That is not the vision though of how this technology could best be used. You would like to do wireless interfaces. One way to do that is a simple passive wireless coupling, a mutual inductive coupling between an external coil and a stretchy epidermal coil that coupled electrically to the skin. This is one way to do it. You can measure thermal conductivity of the skin. You can measure hydration state, impedance of the skin, you can do different things in that way and that’s fine for certain applications. It’s not completely generic however and so I will show you at this point some things that are still un-published, but it’s the result of integration of a full active radio on the patch. So I won’t get into the details, but this is full wireless radio transmission in EEG. The green traces are traces measured with the epidermal patch. The blue with a standard wired based system and you can see good correlation there. There is ECG in a sitting state and again very nice correlation between epidermal and standard wire approaches. This is EMG, same thing, wave forms look a little bit different here, but again this is ongoing work so we are trying to tease out the details here. This is measuring EOG, so you can put these things near your eye and you can track eye motions because there is muscle activity associated with that. So this is moving the eyes up and down. You can see conventional EEG and then this is horizontal back and forth motions of the eye. And we have shown, this is not great I am going to input a movie in here, but you can do game control with that kind of thing; eye motion, track using an electrical measurement and a wireless link. And we did PAC-MAN, but I just have a still image here. So that’s kind of all I wanted to say. I would say what we are trying to do is still early days. There are still lots of opportunities, as I was saying before, for new ideas or advanced technology development. But I feel like we have some ideas in place and some functionality that’s headed in the right direction I would say. It involves a lot of collaborators, professor Y. Huang does all the mechanics, professor P. Ferreira does all the manufacturing systems, T. Coleman we have worked a lot on EEG, these two guys on brain machine interfaces. But ultimately the students do all the work. They come up with a lot of the best ideas and so it’s best to kind of end the talk by acknowledging them and thanking you for your attention. I will be happy to answer questions if you have any. [clapping] >>: So maybe the power is also inducted in the sense how does it get powered? So not only for taking the signal out and do you have enough storage built into it? >> John Rogers: So those are great questions. So the answer is yes to everything you said. So you can do inductive, you can do Far Field RF power transfer. We have done both of those. This is an example of inductive powering of LED’s. These are for implantable devices, you can get them subdermal and you can light them up in that way. I don’t have our Far Field RF transmission data, but you can do that as well. A fair amount of RF comes out of your phone and so you can harvest even a small fraction of that. You can begin to do those things, depending on the power requirements. >>: What kind of power requirements do you have? >> John Rogers: Yeah, so for R radios, so we have a multi-functional system that does EMG, temperature and accelerometry with a voltage controlled oscillator for the radio frequency communication, about 2mW. So it’s low duty cycle, but that’s not very sophisticated. I mean think low power Bluetooth, sub-threshold logic, there are a lot of opportunities for driving down the power requirements. But if it’s a couple of milliwatt, you have got a watt coming out of your phone you are pretty close to something that’s reasonable. I think in the case of a confined environment like a hospital or a research lab you can always put an RF source. A patch antenna in one corner of the room and do it that way. And in fact we haven’t done that in a continuous mode on humans, but we have done it for LED’s injected into animals for optogenetics studies at Washington University. So that’s full wireless power and controlling when the LED is coming on and off. So, inductive Far Field RF is two options. And in some of these passive systems you don’t need power directly on the device. The other would be storage, right, battery. And I think that batteries are possible, but with batteries I would just be cautious. I think batteries are challenging. So we are working on it. I think you can do it. There are some practical aspects around how do you achieve robust encapsulation when you are in a soft rubber situation? And I would say that’s a challenge maybe for the community. It’s not one that we have solved. We are still thinking about it, but we can do batteries that are soft and stretchy. And this is an example of one. So this is a rechargeable Lithium-ion battery. Not a huge capacity, but you can run an LED. And the way to make it stretchable is basically to just take conventional Lithium-ion anode and cathode materials and create very small cells and then connect them with the deformable wires. And then package a soft elastomer top and bottom and an electrolyte gel in between and so you can do that. So this offers extreme wide range of stretchability, maybe up to 400 percent biaxial as you will see here. And the LED stays on during this defamation. So not out of the question. I personally prefer a Far Field RF because then you just need an antenna; it’s very simple. It’s constraining a little bit, but again I keep thinking about the smart phone and that you might be able to use its battery essentially, right. And ultimately the phone is going to be part of any kind of monitoring system anyway. So yeah, power is a good question. We also do mechanical energy harvesting. So you can do piezoelectric materials. We have done a lot of things on the heart, the heart is moving. You can generate enough to power a pacemaker, but for things that are on the skin you have to think about where you are going to put it to get enough continuous power to make sense. And it’s not clear to me how that would work, but we are still tinkering around with that. Yeah? >>: All your examples and your answer to that question assume that everything on that device meets all of your flexibility constraint and all of your air still get’s to the skin okay constraints. If you are willing to sacrifice those constraints for a few millimeters here or there can it [inaudible]? >> John Rogers: Absolutely, you can think about it. It really depends on the application requirement. So what I didn’t show you is we have ways to take unpackaged bare dye and get it into a format, it’s not full epidermal, like the full academic thing that we kind of strive for, but in a lot of cases it’s good enough. You talk to our sports advisory board folks and we hand them the thing that’s made with the bare dye type approaches. They look at it and think fantastic. It’s like it doesn’t need to be that simple. >>: Don’t you think you could put something a little bit rigid there, because you want that part to not bend. Can you kind of put a little piece of plastic under that? >> John Rogers: Yeah, it’s all about mechanical engineering doing things like that. >>: Are you putting wire bonded attaching in those cases or how are you attaching to the bare dye? >> John Rogers: Yeah, there are a few different ways to do that. Maybe offline, this is being recorded. We haven’t published that yet. There are two or three different options. You can imagine what they would be. >>: So the temperature work is really interesting. sort of subsurface blood flow and [inaudible]? Have you considered >> John Rogers: Yeah, absolutely. I kind of ran out of time there, but we are doing a lot around core body. I know you guys are too. You can do blood flow and that’s interesting because the precision you can pick up temperature changes. You can see effects of flow on thermal diffusion. So if you put one of these devices on the arm you constrict the biceps to prevent flow through and ulnar artery down the arm and then you release that. You can see a thermal effect, essentially a pulse type response associated with flow. And again you have to be able to do mapping, because you don’t know. It’s hard to locate that artery, it’s deep in the tissue. But here is one of these devices. This is while the artery is constricted and then you release it. So, that’s the artery right there and you can pick it up and some of these channels have very little response. They are not overlapped and spatially [indiscernible] with the artery. But some have a very strong response like that. And this is the IR camera so again an almost perfect match with that quarter million dollar piece of equipment and the details of how far it overshoots here, what the time scale of this is can be related to the flow properties. And that is an important indicator of cardiovascular health. So you can do this. What we are finding though is you can also do something that doesn’t even require the constriction. Like if you put the patches on the arm the thing that you can do with these patches that you cannot do with an IR camera is you can operate the temperature sensors as heaters. So you can light them up. Here we have heated up these four in an array of 16 and blood flow will cause anisotropy in the resulting thermal distribution. So if there is an artery running down in this direction then heat will tend to spread more this way then that way or that way. And you can map blood flow in that manner. So this slide doesn’t show that, but this slide is showing that you can determine thermal conductivity of the skin. So you can pulse the heat here and depending on how efficient a heat sink you are sitting on that determines the ultimate temperature of those heaters. And with a simple model you can back out thermal conductivity. Now why is thermal conductivity interesting? And it’s because it quantitatively correlates to hydration of the skin. And so we use a conventional moisture meter. It’s a handheld thing that makes the measurement electrically, but we see a very good correlation. So you can do temperature, hydration and blood flow. >>: But you can imagine that there [inaudible] that’s being laid over this [inaudible]. >> John Rogers: Right, so we can do [inaudible] measurements of the skin and look at Galvanic response, things like that. >>: [inaudible], can it stick to other things, internal organs? >> John Rogers: Yeah, so I didn’t put slides, but we have done a lot of work on the heart and the brain. And I didn’t put it in because it seemed like it was kind of outside a Microsoft realm of interest. It’s sort of hardcore clinical medicine is the context there. Yeah? >>: Have you guys thought about using these devices for presentation of information as opposed to gathering information. So gave the little LED example, but. >> John Rogers: Yeah, LED’s are not great because they are power hogs and you are trying to minimize power. So we have thought about electrophoretic display elements because they are bistable. You have to think about the mechanics there. We have done some work on thermochromic materials. I think Coors beer has a thermochromic thing. So that kind of material and it turns out you can embed those dyes into rubber. So you can make a fully stretchable little simple indicator. And it’s being switched by heat and so if you run the numbers it’s more power efficient than an LED. It’s not bistable like electrophoretic ink. So we made a little stretchable indicator using leuco dyes. These heat sensitive dyes. But there may be some other concepts and that is kind of what we are thinking about. The Reebok device is just LED’s but I think there is often times value in having the indicator right on the patch so I think you are asking a great question and we have thought about it. >>: Can you play the IR radio again please. >> John Rogers: Which one, the phone or this one? >>: And I have a question related to it next. [laughter] >> John Rogers: I can send you a copy of the movie. I should really highlight that this is Alex Gorbatchev and he is our collaborator at NIH and he is a world’s authority on skin temperature and he is a great collaborator. And this is his movie. It is not something that we collected. >>: In terms of cost I assume that this is low cost, is that true? there anything that makes it expensive? Is >> John Rogers: No, I mean it depends on the devices. Like those passive devices I have one of this little hydration monitors here. This is one that uses inductive coupling for the readout. I mean this is just metal traces, this could be pennies. Some of the things that involve radios obviously would be more expensive, but I think back to our initial motivation. We are trying to do things that are aligned with conventional electronics and then just do a few things on the back end in terms of rubbers and stuff and mechanics. So our thought is that it might be incrementally more expensive, but it’s not going to be qualitatively different in cost structure, but qualitatively different in mechanics and user interface. >> Desney Tan: Thank you a lot. >> John Rogers: Okay, thanks. [clapping]