 Ready to go! So what better than an introduction by your students? So thanks Ahmed for setting the stage and I'm a professor here at Stanford, so I'd like to test people. So I want to start testing you whether you paid attention to the previous presentation. So I put here a collection of images and I want to ask you what do you think there is in common between all these different images. So we have a car, a soap bubble, a plastic bottle, pharmaceuticals, myself a few years ago, a cornfield and actually an enzyme. What do you think all these things have in common? Given the previous presentation. So we're gonna be talking about the importance of dot dot dot Exactly, so that's catalysis, okay? So catalysis, because catalysis is involved in all these processes. So we are making materials based on calories so 90% of any material around us, so 9 of 10 objects around us are made through at least one catalytic step. So we make detergents, we make plastics, we make pharmaceuticals, we make cars, we also make gasoline to power your car, but catalysis is also what nature relies on. So those cornfields can grow well because they rely on enzymes that do a lot of catalytic transformations and provide molecules and probably you also learn from the previous presentation that catalysis relies on surfaces. So we need to make these surfaces very active by making them very very small. So in our group what we try to do is to engineer materials at a very small scale of atoms and molecules and try to make nanostructure materials with high precision. And what I mean with that is exemplified in some of these microscopy images that show not atoms, so all of those dots, although those dark dots are not atoms, but those are particles containing thousands of atoms. But nowadays with nanotechnology, we can make these molecules with very high precision so we can bring together a very precise number of atoms in order to make these well-defined structures. And then we use these well-defined structures for catalytic processes that hopefully will run and will be very active, stable and selective. So the first question, some of the research that we do in our group is first of all, how do we make these well-defined materials because it's not that easy. So we rely on techniques and on collaborations at SLAC. So this is Liam, who is the postdoc working on this project and Krista Sone is our collaborator at SLAC to use SSRL, the synchrotron, in order to understand how these materials grow. And here there are some curves that are not as important to understand what they are. Those are essentially telling us how our materials develop in the solution. And what we can find, for example, is that if we want to make these well-defined materials, we have to mix appropriate components. And when we mix these components, we can also find surprising things. We can, for example, find that these materials form some small particles, some small nanocrystals, and then these nanocrystals self-assemble into crystals of nanocrystals. And this is something interesting that we found a few months ago and that would tell us how to make these well-better-defined materials that we can use for catalysis. Then once we know how to make these materials, we use them for catalytic applications, such as emissions control. And you heard that from Emmett's talk just a few minutes ago. For example, we're looking at how to limit emissions of methane in the atmosphere. And in order to improve the materials, you can do a couple of things. You can just screen random materials and you hope to be able to win the lottery and find the material that works. Or you can do what we call a systematic screening. So you change one parameter at the time and you understand how that parameter is affecting reactivity. So in this case, we prepared different nanocrystals based on palladium, because we know that palladium is very good for this reaction. And we change, we add one additional promoter, which are different metals here, manganese, vanadium, iron, and so on and so forth. And then we segregate these two phases in a controllable way. And then we can screen this material and see that, for example, in this turnover frequency graph, you see that some materials perform better than others. So we can understand how to improve these materials based on the systematic screening. But what I really want to talk about today is how do we get to the next generation of catalyst? Because if you look at calculating materials, yes, they rely on small particles and everything happens on the surface. But what enzymes do in nature to be so extremely selective and produce the molecules that we need are to have very small pockets. There are three-dimensional pockets where the magic happens. So we want to make the magic happen as well. And when you look at our materials, despite small, these are not three-dimensional materials. These are still surfaces. So if you look at conventional catalysts, for example, these are based on small particles that are deposited onto a support. So we have what we call a two-dimensional time material. But in order to engineer that pocket, what we are developing in our group is what we call embedded catalysts, in which you put a third dimension and you confine your reactivity in that small pocket. And for example, you try to affect the selectivity and how to build molecules that you engineer in length and you engineer in functionality. So this is something that we started a few, a couple years ago. Andrew Risco is a PhD student working on the project. And first of all, we started on polymer-based materials because polymers can be functionalized very easily and are analogs of enzymes. So this is what we call artificial enzymes. So for example, we deposit some of these nanoparticles that are our catalytic centers on this polymeric material. So these are the black dots are the nanoparticles supported on our polymer. Then we remove the ligands from the particles in order to activate the catalytic surfaces. And then we build this three-dimensional shell around the particles by putting more polymer around these materials. And here you can kind of see this overgrowth of the polymer around the particles. So here we think we are creating these artificial enzyme pockets where we are confining our surface reactivity to these pockets. Do they work well? First of all, they provide higher rates than conventional materials. So in green you have the dots for this particular reaction, which is the oxidation of CO2 to CO2, which is used to remove CO, which is a toxic compound from streams. And these green dots have higher rates internal frequency compared to conventional materials. So by confining the reactivity in those pockets, we can get higher rates. And we can also get control over the transport of species because you see those oscillation in reactivity that you see at increasing temperature means that we are changing the reactivity inside these pockets and that CO2 is confining the pockets and it's changing the reactivity of the system going from high activity state to a low reactivity state. And we can repeat these materials are very stable for many, many hours and they show this CO2-induced oscillation that we hope we can use also to induce selectivity. Then if we want to further increase the stability of this system, what we can do is to take the first three steps as we did before. So we get these polymeric materials that we call POS, around our particles. But then we use a nano-casting technique in which we used to infiltrate these polymeric materials with inorganic materials such as silica. In order to create materials that are way more stable even at higher temperatures and pressures, this is what Aizulou at Bekova is working on right now. And we hope to use these materials, for example, for transformations that are even more challenging such as the conversion of CO2 to alkenes or the conversion of methane to other compounds. So with that, I hope I just gave you a brief flavor of what we do. We use these nanocrystals in order to control, understand and improve materials for catalytic transformations. We work on the activation of what we call small molecules, but that produce big challenges such as the production of hydrogen, conversion of CO2 and methane and the activation of nitrogen. If you are interested, please go visit our website and I'll be happy to stick around a little bit longer after this session and talk to you. Thank you very much and thanks for the opportunity to talk today. Our next speaker is Lynette Segalski. She's an assistant professor of chemistry and she will be talking on discovery of a naturally produced, chemically modified cellulose and implications for energy research. Thank you. So in my group, we primarily work to measure and understand a structure and function in complex biological systems, in particular, wholesale systems. And I'd like to talk with you today about the discovery we've made that bacteria make a chemically modified form of cellulose that we envision we can use to impact energy research in particular to convert plant biomass, cellulosic biomass into ethanol. And one of the key bottlenecks in this step is the degradation of cellulose into glucose and then from there it's basically a solved problem to ferment glucose to form ethanol. And this is an emerging technology. There's a plant in Iowa by DuPont where they're working on scale to do this. And one of the premier challenges bottlenecks, rooms, opportunities for improvement is the need for intense cellulase enzymes in order to degrade the cellulose. And that's thought to accompany or account for up to one-third of the cash cost of this whole process. And so anything to ease this process would be an improvement. And our modified cellulose produced by bacteria is more digestible than crystalline and amorphous cellulose. We'd like to translate this into plants to help this process. And I'll allude to one other potential application at the end of the talk. Here's the inspiration for our work, the E. coli bacterial biofilm where you see this is a community of microbes with intact bacteria surrounded by a complex kind of slime-like extracellular matrix. And we're working to transform qualitative descriptors of these slime-like materials in general into quantitative parameters of chemical composition and architecture. And in this case we know two of the molecular determinants for E. coli biofilm formation. These are uropathogens found in the bladder and kidneys, are a really fascinating functional amyloid fiber termed curli and a cellulosic component. And I asterisked that. We now know this is a modified form of cellulose. And we have a remarkable set of biological assays one can employ, in particular when the bacteria are making both curli and cellulose polymers, they elaborate this hallmark wrinkled colony morphology. So this is about the size of a dime. You can see this on the auger plate. In the lab, whereas when they're only making one polymer or the other, they look rather smooth and mucoid. And I have a 10 second video, which is taking up part of my seven minutes, but it's worth it, of the development of these wrinkles. So this is what you see after about 24 hours of growth. And the next 12 hours in 10 seconds here is when they're producing the biofilm components. And you can see the orchestrated elaboration of these wrinkles. So it's really a wonder we can watch these movies all day in our lab. And when you zoom in on one of these by scanning electron microscopy, is when you can really appreciate this topography up off of the surface of these mountains and valleys. And then when you zoom in on one of these fissures, is when you see now really the intact cells in the extracellular matrix. And so in order to start studying the exact composition of these slime-like materials, we worked, developed protocols to isolate the matrix material away from the cells. We're stunned to find that we could isolate these intact, mechanically robust, super molecular baskets that surrounded the intact E. coli. And you see this is on the size of the cell, on the order of the size of one E. coli. And this is one of the E. coli cells in the process of having this type of sleeping bag material pulled away from it. And we can collect intact collections of these. This one kind of resembles a caterpillar, the flagellar, a hair-like appendages still on it. Where they've just been slightly perturbed to allow the E. coli to pop out. So really robust materials, building materials that E. coli uses. And so we have these complex baskets. We also have purified materials of each component. And we use solid-state NMR in order to study these systems because they're not soluble. One can't use solution-state NMR. And they're not crystalline, so crystallography is not applicable. And there's care and innovation in the way we do these measurements. Particularly using custom-built transmission line probes in order to look at, in this case, hydrogen, proton, carbon, and nitrogen nuclei. And here's one of our carbon spectra. The sum of all the carbons in this complex matrix, I compare that for you, to the carbon spectrum of the cellulosic material and the coli-protonaceous material. And we found that the spectral sum of those two samples was the perfect recapitulation for the extracellular matrix. And so this enabled the first determination ever of the composition of the extracellular matrix quantitatively. So it's 85% protein fibers, 15% of the cellulosic material. And then what drove us and fascinated us and continued our research for the better part of two years, myself and a very talented graduate student, was to identify this kind of mystery peak here at 42 ppm. And as you can see, that's a peak not present in commercially available and crystalline cellulos, whereas all the other six carbons are present. And what we've discovered using solid state NMR, measuring dipolar couplings, the distances between nuclei, that out of all the carbons in our material, there was one that was one bond away from a nitrogen. That's the modification C8 carbon. We also found there was phosphorus in the sample, and then through all these other experiments, couplings, etc., we determined that it's a phosphoethanolamine modified cellulose. And it's at the 50% level. So 50% of all of the glucose units have this modification. It makes it zwitterionic and it gives it, ascribes it a hydrogel like properties, which you can see here. So it's able to be semi soluble in solution whereas regular cellulose precipitates out of solution and it is more digestible by cellulases. And so cellulose is the most abundant biopolymer on Earth. And this is the first example ever in nature that there's a biosynthetically modified form of cellulose. And so despite decades of research, and I think it was because it took the solid state NMR to analyze this material. If you digest it to do mass spec or typical conventional assays, what happens is you get glucose, phosphoethanolamine, ethanolamine, glucose 6-phosphate, all things that would look like contaminants in a cell. And so then of course our pressing question was what gene or genes could install this modification that would be required to then enable the transformation of this kind of approach into plants. And we targeted some putative genes and ascribed function to a previously unknown enzyme of unknown function, this BCSG gene encodes an enzyme which is a phosphoethanolamine transferase. And so now we have the identification of the gene to potentially bring in to plant species. In the case of bacteria, it is responsible for function. If the bacteria are not making modified cellulose and they're only making regular cellulose, it doesn't interact with the other polymers, doesn't build this robust building material. And so our next target then is collaborating to bring the gene into plants. See if we can incorporate to some extent the modified cellulose not to interrupt structural integrity of the plant, but to ease digestibility of the cellulose. And then one last very recent result which inspires a new line of thinking is to look at these really beautiful fibers. So this is now lighting up with a special imaging, just the cellulosic fibrils of the modified cellulose. They're very elongated beautiful fibers, whereas with regular cellulose, which is still wonderful fibers, but they're very short and stubby comparatively. So we're thinking along the lines of if there's a way to transform these into carbon fibers and literally transform the equalized building material into our own. And of course the production of the carbon fibers is very dependent on the starting materials. And there have been some efforts in using cellulosic fibers. There are challenges, but it's very empirical as to which starting material precursor you begin with. So we'd like to look at that next. I'd like to end by thanking very much the Precourt Institute for Energy for the scientific inspiration for this work with their seed grant opportunity, also the funding and exceedingly talented graduate student Wery Aung San Buon, originally from Thailand, who really drove this entire cellulose modification project. And thanks very much for this opportunity. Thank you very much, Dene. So our next speaker is Arlen Baker. He's a consulting scientist in the robotics lab. This is the lab that's run by Osama Khatib. And he's in the Department of Computer Science at Stanford. And Arlen will be talking on human-robot collaboration with physical and cognitive interaction. Thank you very much. Thank you. Now for something completely different, but perhaps equally scary. Talk about robots. The goal of our research lab is to distance people physically from dangerous tasks, things they can't do, things they can't reach, things that would harm them. And the challenge there lies in connecting the human capabilities, skills, the intuition's experience with the muscle capability of the robot. And its ability to be someplace the person can't exercise his muscles at remote distances. Robots have been with us for a long time, 50 some odd years. They've been in industrial applications. And what distinguishes them there is that they're in very structured environments, very controlled. Everything is positioned exactly what has to be. There's no uncertainty in any way. When you begin to think about moving robots into the world that we occupy, it becomes important to think about other issues, like what happens in terms of uncertainty. Robots move, people move. You want to move in an environment with a couple's people and objects and other robots even. There's a lot of issues to deal with, including dealing with the uncertainty, trying to measure things, trying to make sensing that's reliable. And operating safely in a way that doesn't hurt it, doesn't hurt the people or the other things around it. So what becomes very important is understanding developing systems that are responsive to the uncertainties, responsive to the environments. And what distinguishes the work at this lab is the use of force, measuring it and inducing it to make changes in the environment. Here's a typical example of a robot doing something. It's grabbing that container and opening it up. Every little articulation, every movement has been pre-programmed. That's typical of how robots work. Clearly, people don't do that. Children can pick up bottles, open them up. All kinds of tasks, we learn how to do very early and we learn them sort of naturally. And we can't anticipate having to program all these things. An issue is how do you develop these skills? And we talk about that in the work we do. There are a bunch of reasons why this would happen now and not like earlier, although we've been dealing with this for 25, 30 years in the lab. There's a lot of physics and force measurement and understanding analysis, mechanics, and material sciences that go into building good robots. But what's happened most recently is the result of the smartphone wars, if you will, where we've gotten a whole lot of sensing and a whole lot of computing that's very tiny and very cheap. So computationally and with new sensors, and that's led to things like the proliferation of machine learning. So we're able to make big headway, take advantage of those recent advances. So this is sort of the domain we're looking at putting these robots into very complex, dangerous environments like Fukushima, Deepwater Horizon, disaster sites where things are broken, underwater, great depths where people can't go, perhaps in mines where there's great risk, and great expense in trying to get people down to those sites. And what's required to make this work really is seeing, touching, hearing, lifting, moving, all the things humans do so readily but that are so difficult to get into an automated mechanism. So what we've done is addressed a few of these in particular, the issues of vision and touch. As the beginning of an avatar, so the goal is not to replace humans totally but to augment them with this skill injection and the guidance of the human and use the automated processing capabilities of the devices we build. You see on the right hand side there are one of the devices that our lab has produced. It's a haptic interface. The user moves it and the robot at the other end moves precisely the way the human does. And the human can feel what the robot is grasping when it touches things. So there's a coupling between the device that is making the experience and the human who's inducing it and understanding what's happening by the feedback process. And on the left is what I'm going to show you now, which is our embodiment of some of these ideas in a robotic device that has done some pretty impressive things. It's an avatar. We sent it to the Mediterranean a year and a half ago to go to a shipwreck that sank off the coast of France about 350 years ago. And it was discovered about 20 years ago. They'd lost sight of it. It was Louis XIV's flagship called La Lune. So there's sort of a sketch of the robot and for some reason a sketch of Louis XIV. But the video here shows you this has audio. This has audio. This is the video made by the French research team that helped us in this as we deployed Ocean One into the Mediterranean. What was hilarious to me was this was developed in the lab tested in the pool at Stanford at a distance of one meter and then it was deployed at 100 meters in the Mediterranean. Talk about bravery. I was quite impressed by the courage involved here. So here he is going in the water. We began with some experiments with a human diver at 15 meters just to make sure things worked and seals worked and that sort of thing. So there was some initial experiments like this just to see how it worked. Now the human is in the boat communicating over a cable. We have optic connections as well but this is with the cable you see. The robot itself has like eight thrusters on it, stereo vision, something like 16 force sensors in the various parts of the body. The arms in the head, actually the arms are the real business end. Everything else is controlled and it's positioning and it's automated. It's reacting to all the forces, turbulence in the water, it picks up something and knows how to accommodate forces across multiple arms. So a usual robot here would have ripped that plastic cage in half because it doesn't understand the forces, the reflective forces. So here's the diver saying, I can't go any deeper than 15 meters because it'll kill me. Would you please go and see what you can do down there? So the next day we took ocean one down to 100 meters. They had sent human divers down there in these very complex suits, big heavy things that protect them against the pressure. With those suits they're unable to do anything other than look. So the advantage here is that we can feel, we can touch, we can interact, not just look. Here are two of the cannons sticking out of the bed of the sea from Le Lune. And unfortunately one of the archaeologists wanted a particular vase that was situated between these two cannons. So boldly going, we put the robot down in between them and it got stuck because you're not quite sure if you're girth and it got hung up. The thrusters weren't sufficient to get it loose, began to get buried because the captain was saying, it's midnight, storm is coming in, we have to abandon it. We've got to leave. So Osama, who you see here with the white hair, took controls and just pushed the way a diver would have in the water. He would do the same thing. And lo and behold there's the ocean one free of his entanglement. Of course it might have just as well pulled some important cables loose, but it got up, it got free. We went back the next day to finish the job, to go down to the depth and to pick up a different vase this time. So this shows the grippers picking one up, that he is working some of the controls. So the gripper is quite nice design. Again, everything is Stanford based. The job is pick it up and put it in a canister that then gets lifted up through, a little button is pushed and it elevates to the surface. So here they are in the hull of the ship working on that, watching the proceedings and guiding it. So the hands are doing what the person wants. So he's really controlling the grasp, the risk where he is and movement in fact. And everything else is automated, all the balance and control and monitoring of forces, reflecting of forces. So he can feel with these devices in his hand here what the robot is feeling. So you can see this generalizing to other parts of the body for humanoid robots. So there it is. It's done his job, he can celebrate the moment. And there's the French lead, that was how he got excited little bees. So here it is, if someone made a point of being the first to touch this vase in 350 years, when it came up it turned out to be kind of an important artifact. It was special in some regard that perhaps only archaeologists can talk about. But there it is, it's got four handles. And the usual celebration with the ocean one is actually a sea vessel. So they celebrate with wine, or champagne rather. So this is the goal of this research, to couple man and machine, to use the intelligence and skills of the man to guide the machine. We're moving towards the point of automation for all of these capabilities, but that's a ways off. And we think there's an awful lot to be done just now, including things like pipeline repair and mining and putting robots where people can't be or people shouldn't be. And that's it. OK, thank you. Thanks, Alain. I think we can better appetite to think what are the energy applications for that. So next to speak is Professor Aaron Lindemurk. He's an associate professor of material science and engineering and of photon science. And he'll be talking about what happens after photon absorption in hybrid dwarf skies. All right, it's a really great pleasure to be here today. So I'm going to try and give you a brief snapshot of some recent work where we're trying to really understand what happens. What are the primary atomic scale and real-time events that happen following absorption of a photon in these very exciting materials for next-generation photovoltaic devices and so on? And as I'll try to show you, these kind of measurements which allow us to visualize what's actually happening can actually have a direct impact on our understanding of the functionality of these materials, the unique optoelectronic properties and efficiencies and degradation mechanisms, and so on. So if you ask someone what is the structure of one of these materials look like, they'll maybe show you a picture like the top left there. You have these kind of corner-touching octahedra. This is the prototypical structural methalymonium lead iodide. You have these corner-touching octahedra. Sitting at the interstitial, you have these organic cations. So that's the sort of crystal structure, as you might define it. But now the middle plot there is an actual MD simulation. And you can see that the structure is just as much a liquid as it is a solid in some sense. And so there's been a lot of discussion in the literature about what are the role of these kind of dynamical fluctuations? How does this impact the optoelectronic properties of these materials? What I'm going to show you is actually an additional surprise. And that is that when a photon is absorbed, when you shine light on one of these materials, you trigger a very interesting and dynamic large amplitude structural reorganization, a kind of rotational disordering that we call essentially a large amplitude distortion of these inorganic octahedra. And I'll try to show how these measurements can, in part, help us to start to understand the unique optoelectronic properties, the efficiencies of these materials. So how do we do these experiments? Well, this is an experiment done just up the road at the Slack National Accelerator Laboratory. We use a femtosecond optical pulse. This is kind of a prototypical pump probe experiment. A femtosecond optical pulse excites a thin film of these materials. And then we probe the structure with a femtosecond bunch of electrons. We're doing electron diffraction. And of course, from the measurement of the bragg peaks and so on, we can work backwards in principle and actually reconstruct where the atoms are and what they're doing in real time. So here's the kind of rough measurement. So there's a lot of data here, and I only have a short amount of time to describe it. But the black curve up there is the, I have a laser pointer here, I guess. The black curve is the static structure of these materials. The red curve, maybe zoom in on that. This is the differential change measured about 70 picoseconds after photo excitation. You can see there's a complex kind of range. Some peaks are going down. Other regions of the scattering pattern are going up. From this measurement, we can essentially, by doing an inverse Fourier transform, we can extract what's called the pair correlation function. This encodes the short range atomic scale correlations and how these are changing as a function of time. So to understand what's happening, I'm first going to show you what the static pair distribution function looks like. That's over here. Essentially, you have a peak, for example, right around three angstroms. This corresponds to the distance from the central lead atom to one of the neighboring iodine atoms within this inorganic octahedra. You have another peak at around 4.5 angstroms. This corresponds to the body diagonal, one of the diagonals, the distance separating one iodine atom to the next. This is the static structure. And of course, you have higher order correlations as well. So what's happening as a function of time? Well, here's the kind of main result of this measurement. And what you see is actually a really dramatic structural change. Quite surprising. We weren't expecting this at all. There's a very large amplitude dip right at the iodine-iodine bottom length, 4.5 angstroms. So this says that we're dramatically washing out those correlations following this optical excitation. In contrast, there's very little change at the lead-iodine distance. So what this says in a very coarse way is that absorption of a photon drives very large amplitude rotational disorder, a distortion in which the iodine atoms move on the surface of the sphere. This effect completely dominates over any kind of thermal effect in the material. And so this kind of light-induced, volume-preserving distortions actually really dominate the kind of photon induced structural response in the material. So let me spend a couple of minutes and just talk about the summary of what we've learned from these measurements. So the first thing is that from these measurements we're essentially directly watching the transfer of energy from electronic excitations into the lattice. And so from these measurements, we can directly extract things like hot carrier lattice coupling. This is important with respect to understanding hot carrier solar cells, the processes which determine the Shockley-Quizer limit, these kind of entropic thermalization processes that in the end determine these kinds of efficiencies. These time scales occur on roughly 10 to 20 picoseconds. And then kind of most surprising is this kind of observation of this dynamic rotational disordering. The kind of key takeaway here is that it's not enough just to consider these kind of intrinsic dynamical fluctuations that occur in these materials. And that actually absorption of a photon drives really large amplitude structurally organizations. These occur on picosecond time scales. We think these processes are happening really every time a photon is actually absorbed in a photovoltaic device. There are a number of ongoing experiments. We're probing a number of other related materials. There are actually a lot of interesting complementary techniques, all from the perspective of all optical probing of these processes and materials, for example, are interesting terahertz emitter. So every time the photon is absorbed, they emit with quite high efficiency terahertz photons. And these photons, in turn, actually also encode these kind of structural distortions. This is one of the ongoing works that we're working on right now. And then in terms of the future, it hasn't escaped our attention that, of course, the iodine atoms and these proticle probe skites are oftentimes the key mobile species that actually, in the end, determine the eventual degradation mechanisms and the lifetimes of these materials. So we think that in some sense we're kind of capturing the first steps in these ionic hopping events that are occurring on kind of picosecond time scales. And then really, from a very general perspective, we think these measurements can provide a new kind of feedback between experimental, between synthesis and the design and engineering of new materials and the kind of real-time feedback of the kind of fundamental processes that in the end determine the ultimate efficiencies of devices based on these materials. So thank you very much and I'll hang around afterwards, happy to take any questions. Thank you, Arun. Thank you. So next up is Arun Majumdar. We all know him as the director of the Precourt Institute for Energy. He's also the J. Precourt Professor of Mechanical Engineering and of Photon Science by Kutze of Material Science and Engineering. And so Arun will be talking to us today about two-step, semi-chemical redox reactions. While this is being set up, let me just take this opportunity to congratulate Lin or Chris Edwards, I don't think they are, and Sally, Richard, and all for putting this amazing program for the last 15 years that have really transformed, as we talked about yesterday, what's going on at Stanford and beyond and all the industrial partners, and that has truly been transformative beyond Stanford, not just Stanford. So what I'm going to talk about is what we have started doing in a group which is dramatically different from what we used to do when I was, for the last 25 years, especially most of it at Berkeley. And this is a collaboration with Will Chu, who you heard from in the morning, as well as Mike Tony, and these are the students, Hyungyu was a postdoc in our group, Shang Zai, Jimmy Rojas, Nadi Alborg, and then Kipil Lim. And this is to look at hydrogen production. So before I move forward, let me just give you quickly how hydrogen is produced today. Most of the hydrogen is produced by steam-ethane reforming. I won't go into the details of that, but this is a hundred billion dollar industry, mostly used for ammonia, as well as refinery, and some for methanol, and some cryogenics. And so there are, of course, carbon emissions from this process. But the cost of hydrogen production today is on the order of about $1.50 or $2.00 a kilogram. So this is an existing industry, and the question is, can we decarbonize? But hydrogen has other implications as well. If you could somehow produce carbon-free hydrogen, it has implications of what you do with CO2. So it turns out to be extremely important commodity. But the key is to produce it at, you know, hopefully less than $2.00 a kilogram. And if most of the cost is really in the energy, and if you do the math, it turns out that the cost of energy has to be less than $30 a megawatt or $30.00, where today, as you heard from before, wind is sort of getting there, it's already there. In some places, solar is getting there, et cetera. And if you look at the pathways to get there, you have, you know, electrochemical pathways, photochemical, biochemical, as you heard of earlier all, and thermochemical. A lot of work is going on in electrochemical because you get such cheap electricity today. What I'm going to talk about is a different approach. And in terms of research, so we are trying to create options out here. And one of the options is thermochemical. And that's because today's chemical industry is almost exclusively thermochemical, and there's a tremendous amount of know-how that is there in trying to take something and scale it, to large scale. And it is volumetric in nature, so it should scale in terms of the economy of scale should be there. And with the know-how there, the question is, do we have the right pathway? And if you have cheap electricity, if you can do electrochemical, you should be able to do, you know, inductive heating, and the heating part should be relatively easy. So how do you do water-splitting thermochemically? The most common approach is what is called a two-step reaction. It's essentially a redox heat engine. You take an oxide, you heat it, the oxygen comes out. Now you've got vacancies in there, expose it to water, and you reduce water to produce hydrogen. And then you, you know, turn this around, and, you know, you got heat addition at high temperature, heat rejection at low temperature. This is a heat engine. The work is a redox work. The challenge is the following, that the materials that are used today, and will shoot at his thesis on this at Caltech, the materials are cerium oxide and ferrites. And the temperature, the high temperature part of it is 1500 degrees Celsius. And that, in terms of the industrial setting, it can be done, but it gets to be expensive. So the question is, can we reduce the temperature so that the material degradation actually goes down? And this has been a challenge. So how do you get to 1100 preferably 1000 degrees, which is sort of the holy grail, and can we go even lower than 1000 degrees? Now this may seem easy, but it's not, because it is a, it's a thermodynamic problem. If the thermodynamic doesn't work, the kinetics don't matter. And if you look at these, the two reactions and the two step, the delta G for both of them has to be less than zero. If you are in the red line, which is the high temperature one, anything below that line is thermodynamically feasible. And for the low temperature reaction, anything above the blue line is thermodynamically possible. Now, if you are operating at 1500 degrees Celsius, the high temperature, the region between the 1500, the dash line, and the blue line is relatively large, you try to get to, in 1100 degrees Celsius, and that window shrinks. And so you've got a thermodynamic sweet spot, which is really small. And the question is, can we get materials in there to be able for that to be able to do both the reactions simultaneously? And as you can see, sodium oxide kind of starts in there and immediately gets out of the region. But you're at 1500, it's fine. Ferrite's do the same. So the question is, how do you find a material that has the right enthalpy, but frankly has large entropy change, which is, you know, difficult to get. And we were looking for materials that would undergo phase transition because in a phase transition, you can get large entropy change. We ran into this material that was basically discovered two years ago called entropy-stabilized oxides. And what you do is you take a bunch of oxides, in the magnesium oxide, cobalt, copper, et cetera, and you mix them together and heat it. And it is known that when you heat these oxides, they, you know, you form essentially a stable structure. And in this case, you know, it's stabilized by the entropy of mixing of the cations in a matrix of oxygen in a matrix. And so this goes to a phase transition. And when you go from a spinel to a rock salt structure, the oxidation level changes of the cations, of the transition metal cations, and oxygen is kicked out. And so there you have now an interesting approach. And they have shown that this is indeed a phase transition because if you change the number of cations, you can change the control of phase transition temperature. Now, they discovered this material, but they didn't think about the water splitting part, and said, we asked the question, can you use this phase transition for a water splitting reaction to produce hydrogen? And indeed, we find a whole class of materials, not just one, whole class of materials that can split water by a phase transition from a spinel rock salt structure to a mostly rock salt structure, a little bit of spinel. And so when you go from a spinel to a rock salt, again, the oxidation level of the various transition metal changes. The question is, how good is it compared to sodium oxide in ferrites that are the high temperature materials? So here's some data in the interest of time. I'll keep it short. This is at 1,100 degrees Celsius. And you find that the gray bar out there for the sodium oxide is the thermodynamic limit for sodium oxide. And you find also cobalt ferrites in two different ways of synthesizing this material. One is a sol-gel technique, which seems to be working better. The other is a solid-state sintering method. And you have nickel ferrite and cobalt ferrite. That's the performance of the existing materials. And the material that we have found, this iron, magnesium, cobalt, nickel oxide, is much better. This is 1,100. And we said, can we try 1,000 degrees where the other materials don't even work at all? And we find that indeed, you can get this material to work to produce hydrogen at 1,000 degrees Celsius. And we're trying to go even lower. The question is, of those cations that are there, iron, cobalt, and nickel, which one is actually redox active? Is one of them redox active or all three of them redox active? The way to find out is to look at extra absorption near its structure. So you hear this is where the oxidation level changes, shifts the core absorption of the electrons. And what we find, the bottom line is that we find iron to be the redox active species. Cobalt and nickel are redox inactive. What we're trying to find out is what is cobalt and nickel doing in this lattice? And that we still are trying to figure this out. Since iron is the redox active, the question is, is there a optimal concentration of iron? And indeed we find that 30 to 35% of iron in this material seems to be working really well and thus producing these amazing results. And cobalt really doesn't change much because it's not redox active. So this work has been funded by the DOE, Office of Fuel Cell Technology, a little bit of money from there, a little bit of money from Slack, and a little bit from Suncat. So what are we trying to do next? We're trying to find the limits of kinetics and the thermodynamic capacity and how can we approach these limits? We have done some measurement of kinetics. They actually look pretty good, but we don't know what the rate limiting steps are. And depending on the rate limiting steps, then you can tweak the kinetics and the kinetics are then now very important in terms of the cost estimation, et cetera. We're trying to figure out the role of the polycatein oxides. We think it is the entropy of mixing that reduces the phase transition temperature, but we're trying to find out more about this. And the question is, can we go down below 1,000 degrees Celsius? Because that really reduces the material degradation and opens up all kinds of opportunities. The question is, can we reduce CO2 to CO? And if we can do that, then you potentially can sin gas and other things. These are all reduction reactions. We're also exploring, which is a very difficult reaction, can you do partial oxidation with this material? Because now you can mix and match various cations and you can statistically, you can think of copper and copper atoms around the oxygen and maybe a zinc atom elsewhere. And then you can do the oxidation from the material itself to see if you can partially oxidize methane to form methanol. This is, of course, a very, very difficult experiment. And because of this, we have sort of, because of some of the results, we've got some funding now from the Office of Naval Research and Tomcat Center and another office from ERE through Idaho National Lab to try out all this variety of other things going on. Thank you very much. Thanks, Yvonne. Okay, next up is Balaji Prabhakar. Balaji is a professor of electrical engineering and of computer science and by courtesy of Management, Science and Engineering and of operations, information and technology at the GSB at the Stanford Graduate School of Business. Balaji will be talking to us today on the future of transportation, zero commuting. I'll open it. Hi, good afternoon everybody. I was not expecting to be teaching this quarter when I accepted to speak here and I'm teaching a course and it's a core course, so I didn't have time to prepare for the future of transportation, zero commuting. So I'm gonna speak on another topic in energy. But with some background, for several years, since 2008, my students and I and collaborators have worked on incentivizing and reducing road congestion through getting people to commute off peak and using monitoring incentives and other things. This has led us to also doing things like wellness and energy reduction programs and I'm gonna talk about energy reduction program. But the idea of zero commuting is this notion that what happened to congestion anyway? I just feel that the only way to really bring congestion down is to get people to give up driving and just work. I don't want to use the word telecommuting because it has a sort of limited sense in which people understand it. The more exciting thing would be if you imagine how future workplace is really some thing in your house or you don't have to go to work every day, but you have full VR type interaction with your colleagues, I think that'll do the work and many companies will be incented to adopt that because it solves all sorts of problems including just real estate and parking issues and various things. So that's for a future time. Right now the future really is about building this sort of highly interesting commuting sort of communication mechanisms and putting it in the cloud. But let me tell you a little bit about an energy project which I'm in the process of reviving. This is started with energy usage reduction in Japan on the heels of the Fukushima disaster. But thinking of reviving it now in the context of high schools, local high schools like Pali and Gunn especially. So as you all know, there was this disaster and Japan has two types of frequency, 50 and 60 Hertz and they couldn't borrow and the population had to reduce the energy by 25% quickly, okay? And the ministers of I think trade and industries came out with half-sleeved shirts and no tie and all that and said we can all dress down so we don't need air conditioning and we will have brownouts and we will forego many things. They helped but what NEC who was sponsoring some research we were doing was interested in is can we actually go to every house and have them reduce energy really fast? In a time scale that would make sense but the problem is that there just wasn't enough energy meters deployed. But imagine if NEC's project could be successful. So we came up with a scheme, we could have the roughly 100,000 employees of NEC use this, reduce energy and then take it to all the other companies like Toyota, Honda, they're all very big employers and they all wanna do something for Japan. That was the whole goal. What I wanna share with you is the proposal because they're very what do you do when you don't have smart meters? Okay, how do you incentivize people? How do you measure? So our proposal is just these five pictures. Using smartphone as a sensor, take energy meter readings at least once, maybe two times a day, okay? Just take pictures. And so Damon Rishik who's seen this picture is the person who came up with some of the app designs, I'll tell you. And the app will automatically upload the meter reading and you earn points for low energy use and these points are redeemed by playing a game of chance. When we pitched this to NEC and I was visiting Japan and we wanted to work on this, some of the reward money was to come from, potentially from at that time, someone like 7-Eleven, for example, because they were also interested in helping energy consumption go down in Japan, okay? So, and then the idea would be that you save money and Japan saves energy. But let's look at the technical components of this proposal for a second, because I claim that much of this is now quite easy to do. This was done back in 2011, 2012. So think of it in the modern sort of, in the post machine learning sort of success story concept in our world. When we get someone to take a picture, we had them also enter the meter reading because we found that first we thought Optical Character Recognition could do this job, but these meters are sort of, have cracks. The pain has cracks and so it's not very easy to read your meter, okay? And so we had people enter it and there was an augmented reality widget that gave the meter a personality so the meter could speak to you. That was proposed here, but actually done. Now, looking forward, what does that mean when you have all these meter readings? Typically this graph is looked at shifted 90 degrees. So time is increasing downwards vertically and meter readings are on the, in this case, the x-axis. So your goal as a user is to stay within the green region in the energy consumption sense on a weekly basis. You get points for staying there and if you leave and go into the yellow zone, you get fewer points and you go into the black zone, you don't. But the way to visually convey this, we had a caterpillar, okay? So depending on energy readings, you can see these small white dots inside this picture. That's when you took the readings. Your caterpillar will follow that. So as long as the dots remain inside the green area, your caterpillar's gonna be green because all caterpillars wanna be green, okay? And otherwise, it'll turn yellow or black and which is when it gets bad. So it's just a visual cue. We had this as to incentivize you to take energy readings frequently. We had these success, two successor readings. The interpolation isn't sort of just the usual. Shortest line, the shortest line, you would always go horizontally and then come down. So if you took one reading here and one here, your caterpillar would turn black, okay? So we had it, you would take it frequently. This is all very simple and easy to understand. We also had you build cohorts for you based on your neighbors, so something that has to be your location and some to do with your friends at work, for example, NSE colleagues. And so all of this was implemented in subsequent projects in various places, including the Singapore Public Transit System, the BART, commuting incentives program and in various of the places. But this was proposed in the context of the Japan Energy thing and I wanna take this soon to schools over here because this looks like the sort of thing that every student has a smartphone and we could run this. And they also like to compete with other schools. And so this is basically how, we have to think of how to incentivize here but other than that, the components are all there. So hopefully by the time you come back next time, you'll hear more about this. You'll certainly hear more about the transportation, how to prevent people from commuting or how to reduce commuting as much as possible without losing the benefit of physical interaction, okay? So let me stop with that, thank you. Thanks, Gladgy. Good to hear about your work, both in transportation and in other energy applications. So next up is Professor Juan Vibas. He's an assistant professor of electrical engineering and he will be talking today on very high frequency power electronics, research directions and new applications. Welcome, Juan. Control energy. Hi, everyone. I don't think it's showing the right screen. So hi, everyone. So I wanna tell you a little bit about the work that is going in my group. And essentially we work with power electronics. Essentially power electronics is pretty much what it's responsible for making your laptop adapter or your cell phone charger too big. So particularly the part that I'm interested on is in improving the power density of power converters. We want to make power converters smaller, cheaper, more efficient. And one of the few handles that we have as engineers in this area is switching frequency. As you can see in the power supply shown here, usually the size and volume and weight is heavily dominated by passive components. And this means inductors and capacitors. And by having switches operating on and off as fast as you can, you can actually reduce the size of the power converters. But these techniques have their limits. Particularly in the industry in the past few decades, engineers have been increasing the switching frequency. The power density of the power converters starts increasing up to a point. And then we reach a point of diminishing returns in which further increases in switching frequency are not translating anymore into smaller converters. You can see for example in the converters in the bottom of the figure that as we increase the frequency from 250 kilohertz to 500 kilohertz, we can increase the power density from 10 to 13 kilowatts per liter. But if you double the switching frequency to one megahertz, the power density only increases to 14 kilowatts per liter. So it's kind of questionable whether another further increase in switching frequency would translating something smaller. So like what my group is trying to do which is a little bit counterintuitive is essentially just in keeping increasing the frequency but a heck of a lot more. And like the reason is as it's fundamental that you increase the switching frequency with the size of the inductors particularly and transformers become smaller. And it's more to the point that you can actually eliminate the magnetic core that is needed to make these inductors. So essentially you completely eliminate a source of loss which is core losses in a transformer. And this essentially allows you to give you the opportunity to fabricate inductors by simply printing traces on a board or using just a little piece of coil around a plastic core or just essentially print them on a printed circuit board and avoid any fabrication techniques. And we take advantage of the geometry so as to improve upon reducing the electromagnetic field that are accidentally emitting when we design these power supplies. And essentially our roadmap is to make converters also cheaper to fabricate. For example like we can go from the power converters that I built when I was in grad school to converters that essentially are just like printed with very inexpensive batch fabrication process like the one that is shown on the right. And essentially we want to build essentially building blocks that can allow you to have very compact, very power dense power electronic circuits. But we want to take it even further. We want to take advantage of new fabrication techniques like 3D printing in which we can actually start fabricating components simply in the computer. And then analyze them in a FEM program and have predictions about the inductances and performance parameters that we may be interested on. And then we can just simply 3D print it in a low cost 3D printer to fabricate a scaffold that are plating it can render a component that is electrically effective and very, very, very lightweight. And essentially later on, we want to essentially 3D print converters. So like instead of like making complicated process you can just get your very low cost 3D printer from the Kickstarter project that you supported and just start making efficient power converters that are actually very lightweight. So for example, here's an example of some converters that we built that shows a 50 watt converter that uses this 3D printing fabrication techniques and only weights five grams. And we're using this to essentially ignite plasmas for very interesting applications. So this is a very lightweight converter that is capable of striding a very powerful plasma. Where do we want to take this? To infinity and beyond. Essentially, we want to use this to make thrusters for satellites. If you imagine like there's now low cost satellites called CubeSats that for a few tens of thousand dollars you can actually start sending to space hitchhiking a ride on a rocket or they can be released from the space station. But unfortunately, they tend to be placed in an orbit that very rapidly decay limiting the lifetime to between six months to a year. But it'll be interesting if someone can provide a solution for propulsion that can actually keep your satellite flying much longer. Essentially, we want to go from systems that has ion thrusters like the one shows in here to something that is much, much smaller. So we're collaborating with the University of Australia National University to make CubeSats satellite propulsion thruster that we can fit in a CubeSat. And essentially we've been in the past few weeks, few months fabricating this type of structures in a way that actually can have multiple functionalities. So for example, here in the middle you can see a CubeSat and it consists of several panels that forms a structural frame that we can replace with one of these planar converters that we're fabricating in my group. And actually, just a couple of weeks ago, we presented in the International Astronautical Congress in Australia a full implemented CubeSat 3D printer, CubeSat thruster that actually has quite a lot of room to spare. So now we can try to collaborate and find opportunities to see if we can fly one of these things. We can use them then to monitor natural resources, disaster areas, et cetera at a much lower cost. And we are also trying to use this plasma generating techniques in other, for example, to implement plasma medicine applications. We want to be able to use plasma to essentially treat kill bacteria and biofilm. And also like recently we've been working with some high voltage power converter technologies that allow us to generate very high voltage at a very, very small volume. The reason we want to do this is we want to apply this in order to purify water, for example. It turns out that if you have a couple of electrodes and you apply a very high voltage field between 20 and 50 kV per centimeter across liquid, if you happen to have bacteria within that liquid, the bacteria kind of explode. You essentially electroporate the membranes of the bacteria. And usually this is done in a very, very large volume systems. Thanks to the technology that we're developing, we're actually able to reduce the size of this system to something that is kind of a kingdom the size of a brita filter. And we've been able to test this with liquids in which we're able to demonstrate that with like a very short pulse and little energy, we can actually eliminate bacteria that is existing in the liquid. But unfortunately, water is not a very, it's not a good commercial idea. So instead, we're trying to see if we can take this to a market that is probably a little more interesting, which is milk. It turns out that about 40% of milk that is produced in rural places is actually gets spoiled before it's reached a pasteurizing distribution channel. So one of the things that we want to build is a system, a pasteurizing system that is very, very small that sits essentially next to the cow that is maybe powered by solar power that essentially that a farmer can use to treat raw milk and pasteurize it to levels that like can give them more time to for Smith to redistribute your channels. And this is just a picture of one of the conversions that we made. And with that, I'd like to thank the Precord Institute at the Top Can Center for supporting us in part of this work and NSF and all my students. Thank you. Thank you, Rana. We're now at our final speaker in this session. And she is Professor Xialin Zhang. She's an associate professor of mechanical engineering and she'll be talking to us about light-driven fuel cell with simultaneous production of hydrogen peroxide. Xialin. So what an honor to be the last speaker. My parents has the wisdom to name me. There is the combination of X and Z in my first and last name so that I'm always listed the last. So I'm still the last one listed in the mechanical engineering faculty directory here. So that's a very easy way to find me. And it's very hard to pronounce as well, my name. So I want to talk about today is a little bit about my group specialty on flame synthesis of metal oxide. And how that leads to you to look at today's topic is basically we're using water, oxygen from air and when we shine light and we're generating hydrogen peroxide and this is spontaneous, it produces a little bit of electricity. So when we look at renewable energy, we want to think about how to convert renewable energy either to fuel or electricity. As an engineer, we think about devices. There are a number of devices to do that like a fuel cell or solar cells. And if we look into all those devices, they have many things in common. One common thing is they all contain metal oxides. So when we think about energy problem in scale, that means we have to produce metal oxide in a scalable fashion at low cost. And many of those applications require executive design of the nanostructures of metal oxide. So the challenge is how to produce those metal oxide and nanostructure at a scalable fashion. And my tool, it goes to my rule down combustion is to use flame as a reactor to synthesize them. Flame is very cheap and scalable, has high temperature and it operated ambient pressure. You may now know actually all of you own products from flame synthesis, like the tire carbon black is made by flame synthesis. So anything like a rubberage and that contains flame made material and actually flame made material annual production is almost $20 billion. So that's a huge market that shows the scalability. And over the past decade, my group has trying to synthesize different nanostructures of metal oxide. We started making a metal oxide nanowires. And recently we were able to control it down to a mono layer sub-nanometer metal oxide material. And we also have techniques to dope those metal oxide to change their electrical and the chemical properties. So how those material perform in applications. One of the application as many of you have looked at is our solar water splitting. I like it because it has the photo light absorption property, transport property and also surface transfer property. And many of our flame processed material shows better performance than other techniques. We also use flame try to replace the traditional furnace. In this case, we look at a dye sensitized solar cell or perovskite solar cells. They all evolve a process to use furnace to bake the TL2 paste. And normally it's about 30 minutes and we replace it by high temperature flame. Now we can done it in less than one minute. And our efficiency actually is a little better. Our flame made material because it's high temperature is high quality. They can also serve as a very efficient whole doping material for many of the semiconductor devices. So that led us to look at a lot of solar water splitting using our flame made oxide or doped material. We even develop flexible electrode for this case. We also look at the other side, the hydrogen side. We start to look at the molytide sulfide. We activated the basal plane for hydrogen evolution reaction by generating sulfur vacancy. So this is exciting. But when we look at the water splitting, as many of you have said, this is really challenging. The efficiency is still not far from satisfactory. If we look at a little bit, there are side reactions going on here. If we look at the oxygen side, the water actually can be oxidized to hydrogen peroxide. And they look at the hydrogen side, if there's oxygen there, we can reduce oxygen to hydrogen peroxide. And hydrogen peroxide is a very important chemical. It's a strong oxidizer. If both side can generate hydrogen peroxide, it's a liquid phase that may solve some of the problem with water splitting. And if we do this, do the electrochemistry is not a spontaneous reaction. It means we require some kind of bias. So I mentioned this is a side reaction. And we have to compete with the regular water splitting. So we work with a Young-Snosskopf group, and they did a lot of calculation. We did experiments. So the bottom line is we do find conditions and the material that prefer to form hydrogen peroxide. And now we can use that knowledge to build a reactor. And it looks like a fuel cell runs in a bucket of water. And we sign light. As I mentioned, this is not spontaneous. But if we choose the right material, we can use photo carriers to make both reaction hypen spontaneously. In this case, we're using a BIVO form. And now the whole thing becomes spontaneous. To show you this is really happening. So I have some plot. The first one shows a chopping curve to produce hydrogen peroxide without any bias. It shows it produced hydrogen peroxide. On the other side is typical fuel cell measurement. It means while we are producing hydrogen peroxide, we produce a little bit electrically, not a lot. The bottom shows we have the fuel cell in the beaker and we have our ED light. When we don't run the light on the fuel cell, it's dimmer. But when we turn the light on, our hydrogen peroxide fuel cell is generating additional electricity. It made the light brighter. If the goal is to produce hydrogen peroxide, we can also apply a bias. So if we apply 0.5 bias, we can actually accumulate hydrogen peroxide to thousands of PPM level. And the really exciting thing is this also works on tap water. So if you just put tap water, put 1.5 volt, and it can also generate hundreds of PPM level. For water disinfection, we only need tens of PPM level. So it's not a very high requirement. So what's the future? And this is one of the vision we have with this spontaneous hydrogen peroxide generation system. We call it a solar-powered water disinfection. So we can use dirty water with bacteria heavy metal, run through this fuel cell, like driven fuel cell, and we can kill the bacteria, and only needs light. Another possibility is hydrogen peroxide is a strong oxidizer. It's on the, you probably have the 3% at home for disinfection. And as a clean oxidizer, it can be used for many additional chemical production. So we can envision here, using hydrogen peroxide as intermediate for other oxidation reaction, and use solar to power those reactions. So that's really exciting for the possibility. So with that, I want to conclude with light and water, a little bit air, hopefully we can produce hydrogen peroxide. And thank you all for your attention.