DNA Origami: The Bridge from Top-to-Bottom, Professor Ashwin Gopinath, Samsung Forum

So I'm gonna be talking about DNA origami, and I know that seems like a little— it's a little strange title, and it's about the bridge from top to bottom. And most of this work was done at Caltech, where I used to be before, and only recently did I move to MIT. So before I go into my work, I just want to take a few moments to kind of contextualize what it is that I'm trying to talk about. What do I mean by top? What do I mean by bottom? So it's pretty— I don't think anybody will argue with me if I said that this phone, this is Samsung, is a good representation on where we are as far as society is concerned, you know, in our ability to create complex devices. And everything, if I were to actually take one of these electronic devices and if I were to, you know, section it, this is an SEM of, I believe it's one of the processors, not the most recent one. And on the bottom you have the silicon, you have the transistors, and on top of it there are about 200 different process levels. These are the interconnects. And everything that you see here was formed by repeated iteration of one of these 3 processes, which is basically you take a surface, you pattern You either deposit something, etch it, or you dope it. And you repeatedly do it. Each step you build from a surface. And a lot can be told about the size of the structures. But what I want you to appreciate is not the size, not nearly the size. It's appreciation of the registration capability. If you know something is there on a surface, if you know exactly where it is located, you can build an entire world around it. You can build it, you can connect it to the external world, you can connect it among itself to do all the computation. computations that we see. And the same technology has been used for doing microfluidics. This is a chip in which you have a whole bunch of these little tubes in which liquids go around and mix together. You can build optical interconnects. These are exactly like microfluidics. You have a whole bunch of channels for light that you traffic around, you move it around, and you do computations or you do communication. You also build mechanical structures. This is an accelerometer which replaces this particular kind of a gyroscope. And you see this in all the electronic devices nowadays. Now, this is all top-down fabrication. You take a structure, you slowly build around it, you etch it away, you deposit something, and you build around that. And there's an entire different philosophy of fabrication, which is bottom-up fabrication. And everything other than top-down is built this way, from galaxies down to molecules to everything. Everything that you see around you was built bottom-up. This is just 3 examples. One, all the biomolecules, these are proteins or cells, DNA, RNA, everything. You have quantum dots, these are little spheres of semiconductors. Depending upon the size, you have a certain light that comes out of it. You have carbon nanotubes, these are little tubes made out of carbon that can be used for semiconducting or for— as a wire. And every— in an abstract sense, self-assembly is just taking 2 components, 2 or 3 components, either by attraction or repulsion, they come together to form a particular structure. And depending upon the condition of assembly, you have a certain size, certain shape, certain composition, and a combination of that gives it certain properties. What I want you to appreciate is this mode of fabrication can do complexity and can create properties that cannot be achieved top-down. There is no top-down mechanism, top-down fabrication tool that can be used to create even a single simplest biomolecule or to make very nice carbon nanotubes. There's no top-down method that you can do it. So in this context, what I like to think about and what my lab likes to think about is what will integrated systems or integrated chips look like 15 or 20 years from now. And this is just a MEMS microphone that you find in most of your phones and computers, and I think it's gonna look very similar to it, except you might have like a little nanowire logic somewhere over here. You might have a whole bunch of little protein molecules sitting on the diaphragm of your microphone so that when you speak into it, it's also doing some kind of diagnostics. You might have a little quantum dot sitting over here or an atom sitting over in an optical resonator that is encrypting all your information and sending it out. So I think that this is what it'll look like. And within this context, it seems like, you know, it seems kind of futuristic. But the reality is parts of this have already been done. We already have nanowire-based logic. We already have single quantum dots sitting inside resonators. We already have devices. This is a commercial tool used for sequencing. This is a Pacific Biosciences-based sequencer. in which you have a single protein sitting inside an optical resonator, and as it is functioning, it's reading out the DNA signal. So this— these devices exist. The problem is how you make them. For instance, this particular structure, this carbon nanotube-based transistor made by IBM, was done by what I call as hunt and pick, which is basically take a surface, throw down a whole bunch of carbon nanotubes, go find one, make connects, write a paper. Or in an optical device, this one involved making the resonator, using a scanning a small little needle, move your quantum dot into this— into the position wherein you want to study it, and then write a paper. Or in this particular case, in this particular kind of biosensor, you make the structure, throw down the carbon— I mean proteins, hope that some of them will go and sit in the right position, and you do your analysis. So this is just spray and pray. So the problem is None of this is scalable, none of this is modular. So what I mean by that is what you learn from making a carbon nanotube-based structure cannot be applied to making a protein-based structure. And what I mean by scalability, you can't make billions of these. You can't make billions of these with a guarantee that all of them are going to perform the same way because they're not going to have the same number, they're not going to have the same orientation of the molecules, they might— they will not have the same interface with the device. So how do we make these structures? This is the question that we think about. So half of my talk, or like majority of my talk, is gonna be talking about imagining, introducing this new mode of fabrication that gets us towards interfacing bottom-up structures with top-down devices. And the second half, I'll just touch upon functional devices and new kind of devices and new kind of architectures that can come out of this. And everything that I do, I do it with DNA and DNA origami. So just a refresher on what DNA is. DNA is just this polymer. You can think of it as a polymer. It looks like a cylinder. It is 2 nanometers in diameter. There are 2 parts to it, one strand, another strand that comes together. One of these strands have a phosphate backbone. Think of it as a thread. It's phosphate, and it has a directionality. By directionality, I mean you have a phosphate group and you have a sugar group, and they form a chain. There are 4 kinds of molecules that are decorated on this phosphate chain: adenine, thymine, guanine, and cytosine. They are separated from each other by about 3.5 angstroms. Adenine likes to interact with thymine. Guanine likes to interact with Cytosine. I told a whole bunch of things. You can summarize that. You can take a single-stranded DNA, think of it as a straight line. It has an arrow associated to it because it is asymmetric. You can now label it with A, T, G, C. That represents adenine, thymine, guanine, and cytosine. And once you have one of these strands, you always will have another strand which is going in the other direction. Wherever you have A, you have T in the other one. Wherever you have G in one, you have C in the other one. So that's about all you need to know for the rest of my talk as far as DNA is concerned. So the field of DNA nanotechnology started back in the '80s by Ned Seeman, who's over at New York University. And he asked this question, let's take one strand of DNA and abstractly imagine it connecting to a second strand of DNA. Half of it is connected to a second strand of DNA. The half of the second strand binds to a third strand, the half of the third strand binds to a fourth strand that loops back. So it forms this abstract cross. And he asked the question, what happens if you have tails on the edges of the structure such that the left side likes to interact with the right, the top likes to interact with the bottom? And he proposed that such a structure, if you make it, it will form crystals. And for starting mid-'80s till mid-2000s, even now you find papers, people have explored this idea of fabric. Fabrication. By making these molecules, putting them in solution, they self-assemble into lattices of hexagonal crosses, square lattices. You can even do computation on this. In this particular case over on the bottom right, it was doing Sierpinski triangle growth. It's a fractal growth from a particular seed, and all of this happens because we can program the DNA to have certain interactions, and you put them in a solution and they form a structure. These are all repeatable structures. In mid-2000s, Paul Rodman over in Caltech proposed taking a long single strand of DNA from a virus, it's a genome of a virus, and designing short strands that basically staples it together to form a particular shape. So you can think of it as like crochet, wherein instead of forming knots, you're using short DNA strands to form connections. And, you know, since then, people have made 2-dimensional shapes. Each of these structures were made from a single strand of viral genome folded by helper strands. And the length scale is gonna be about 100 nanometers. The edge of each of these structures is about 100 nanometers. Can form things in 3D as well. This was work done by William Shi and Hendrik Dietz. Hendrik Dietz is a professor over at TU Munich. Sean Douglas is over at UCSF. And they form these kind of really complicated structures. And in all of these cases, every single point within this structure is known. We know every single— where every single atom with inside this particular particle, we know it and we can modify it. So let's actually go through an exercise of designing an origami just to illustrate how easy it is. So for instance, if you want to make this triangle, what you do is you make— you abstractly represent a triangle and you take a pen and raster through the entire surface. Now the same idea can apply to volumes as well, wherein instead of rastering over a surface, you would raster over a volume. Now this represents the single-stranded genome and it has a certain sequence. And since we know the sequence, we can have complementary strands, short strands that bring together different sections of this long single strand. And what you do is you do it in a computer and you have about 200 molecules that form this shape. You put it all together, heat it up, cool it down, you end up getting a structure that you intended to make. Now the scale bar here is 30 nanometers. It's— and don't take my word for the fact that the yeast yield is about 99%. If you zoom out, this is what it looks like. Every structure, we know the yield is close to 99.6 or 99.7. Statistically, we are missing probably half a staple in one of these structures. This is all AFMs. All of them are very homogeneous. They're— if you're doing it at scale, the cost of synthesis is about $200 for 1 gram. So just to put that in perspective, 1 gram of one of these structures can cover about 1,000 square meters. And every point within this can be uniquely labeled. By that I mean, the next day I decide I want to put 3 molecules on the inner vertex. What I do is I go in there, I look at my map. The map is basically each strand, each color is a unique molecule, and every molecule is separated from every other molecule by about 5 nanometers. We identify 3 locations, swap it out such that it can now accommodate a new molecule, reform the structure, you end up getting a structure with just 3 molecules in the inner vertex. It's as simple as that. And every material that can be attached to DNA can be organized this way. So, and over the last 60 years, people have been attaching DNA to every nanomaterials. So as soon as somebody invents a mechanism to make a nanoparticle or a molecule, you stick it to DNA. So for the last 60 years, people have been doing this, so there is a lot of chemistry that tells us how to attach molecules and nanoparticles to DNA. So all those materials can now be assembled on this origami structure. So you can start thinking about this origami as a molecular breadboard. Similar to how you have breadboards on which you can assemble transistors and capacitors to form a particular circuit, you can start imagining the origami as a molecular version of that. So you can think of putting proteins, carbon nanotubes, quantum dots, whatever you can think of, attaching it at specific locations, creating connections within the origami. People have been doing that for the last 10 years. So over here there is 2 FETs forming an isolated junction to form 2 carbon nanotubes to form an FET. This is a quantum dot surrounded by gold nanoparticles. This is a conducting wire routed on top of the origami. These are quantum dots, these are biomolecules that are organized in certain fashion. This is not the entire landscape. People have been doing this for the last 10 years. If you just search DNA origami on Google Scholar, You can literally find hundreds of examples wherein people have assembled proteins, carbon nanotubes, quantum dots, polymers on these structures. Now the issue here, this looks great and from a publication perspective this is really nice. You form a nice little transistor and you look at its characteristic, write a paper. The problem comes when you zoom a little bit out. When you zoom out, this is what that surface looks like. You don't know where the— where that beautiful transistor was located, which means that you make the transistor, it's in solution, you put it down on a surface, spend a whole lot of time finding a good transistor in the right place, making interconnects onto it. So you can't scale this. Again, you are back in the same world. These techniques are not scalable. So what you learn from carbon nanotubes, what you learn from a protein assembly or an ion assembly, you can't reuse it and you can't scale that up. But once you stick it with DNA, all of a sudden assembly of the structure on the origami becomes scalable. Because instead of assembling the carbon nanotube on the origami, you just need to care about assembling this small little DNA on the origami. Similar to that, instead of worrying about assembling the quantum dot, you need to worry about assembling the DNA on the quantum dot. So it makes it much more modular. Assembling the structure on the origami is very modular. Now the question is, how do you assemble origami itself? And my work thinks about this problem. If you can assemble the origami in a particular grid, automatically you can assemble the nanoparticle, you can assemble the nanowires, you can assemble the proteins, you can assemble the molecules, any molecule that can be tagged with DNA automatically becomes accessible. You can organize them on a surface. And the way in which I do that is by doing chemical patterning of a surface. So I take a surface, make it neutral using HMDS, it has methyl groups on it, then I do photolithography or e-beam lithography or nanoimprinting to create shapes that correspond to the origami shape on the surface, and I use the— I use standard chemistry for this, standard silane chemistry which is deployed in most of the foundries. And the size of the features are about 100 nanometers. And once you assemble the origami on top of it, even though the structure is— even though the size of the feature is 100 nanometers, once you organize the origami inside it, you have access to 5 nanometers. Now you can assemble structures at 5 nanometer separation even though you use a 100-nanometer feature, because the origami allows you to actually assemble structures at that length scale. So this is the way in which I do my fabrication, I just do standard lithography and do a short oxygen plasma to activate the surface. So essentially what has happened is the origami itself is negatively charged, you have 3 options, either negatively charged surface, neutral surface, or positively charged surface. If it is negatively charged, there is some repulsion, if it is positively charged, there is attraction, If it is neutral, there is nonspecific interaction, but when you actually add a little bit of magnesium ions into it, there is a weak, weak attractive force because the magnesium forms a bridge between them, and it is this interaction that we use. Specifically, you will end up making surfaces like this. You have methyl groups, hydroxyl groups, and negatively charged phosphate groups, and the magnesium bridges them. Now, the point that I want you to appreciate is every single magnesium in this case is not a glue. It's not strongly binding it. It's there or not there, which means that every single point here is flexible, which means, you know, when I put it all together, when an origami is sitting on a binding site here, red is a binding site and the blue is the origami, it starts, even if it starts with partial binding, it wiggles around because of thermal fluctuation and locks into the patch. And as it turns out, the experiments, this is exactly what happens. This is This is a liquid AFM movie, every frame is, I think, 100 milliseconds, and you see this origami dancing around this patch, eventually settling in there. And as it turns out, we can actually abstract away the features, abstract away the conditions, and figure out knobs that basically can increase the k-on rate, k-off rate, the readjustment rate, and all that. And you end up surfaces like this. Now this is a silicon dioxide surface which was patterned by e-beam. We do e-beam simply because it's easier for prototyping. And every single spot here is a single origami. And just a reminder, on the origami we have 5 nanometer accessibility. We can organize structures at 5 nanometers. And we can do this with greater than 98% efficiency. What I mean— yield. What I mean by that is there is greater than 98% probability that one site will have just a single origami. And we can do this on silicon, silicon dioxide, gallium phosphide, gold, graphene, and silicon nitride. And not only can we position structures, we can also do chemistry onto the surface. So the origami can carry reactive molecules. Once it assembles itself, you can trigger a reaction and lock it into the surface or do chemistry onto the surface. So you can think of this as a path to doing molecular liftoff. rather than actually trying to put a single molecule at a particular point, you can use the origami sort of like a resist. It carries your molecules, goes to a surface, leaves it, and then you remove the origami. So, a practical question would be, or a practical thing to think about would be, you know, how do you actually control dopant at a discrete number level? Can you just put 4 phosphorus atoms at a particular site, or just a single phosphorus atom at a— in the middle of the gate of a transistor. How would you go about doing that? A method like this can be modified for that particular case. So, after we got here, the question was, what do we do? Do we build some nanoelectronic structure? Do we do photonic stuff? Or do we do biosensing? And simply due to my background, we decided to go down the nanophotonic pathway, and the specific problem that we wanted to solve was putting discrete emitters inside resonators. By that I mean, can you put just a single molecule inside one resonator and make sure that you can do it repeatedly. And the application of this technology would be for sensing, for quantum computing, for making ultrafast modulators, for doing non-classical light generation, for doing metrology, as well as for quantum computing. I'm not going to talk about details of the experiment, but the punchline is this is the kind of surfaces that we end up getting. So this is 65,000 different cavities, optical resonators made on a single chip. The intensity of each of the pixel is determined by the position and the number of molecules inside it. So this is not an art piece. This is to basically demonstrate the scalability of our process. And rather than doing it electronically, we did We chose to do it optically. So, and this is just representing Van Gogh's picture. So if you take a subset, each pixel is an optical resonator that looks like this. This is a photonic crystal cavity and inside the photonic crystal cavity I have just a single origami and depending upon where it's located and the number of origamis that I put inside it, it'll have a different intensity. So we can control the number of molecules from 0, 1, 2, 3, 4, 5, 6, 7 and its position. And that gives me 8 possible levels of light that comes out of it. And from that, I construct this particular chip. Now, again, I'd like to emphasize this is not an art piece. This is to basically demonstrate the capability. And this is, as far as we know, the first instance of discrete molecules being interfaced with nanofabricated device. And so once we got comfortable with this process, once we kind of confirmed to ourselves that we can position molecules, we asked ourselves, can we do more? Specifically, is it enough to just position molecules? And for the astute observer, there is a problem here because of the symmetry of the structure. You— even if you decide to position a molecule there, a triangle there, there are 6 possible orientations to it. Because it's a flat structure, so it can go in 3 in-plane orientation, flip over, and have 3 out-of-plane orientation. So there are 6 degenerate states. In a practical setting, what that means, if you design to form such an array wherein you have the symmetry-breaking nanowire on each of these structure, the reality is it's going to form something like this. So only 33% of it is going to be good. So we didn't want to do that, we wanted to actually make it better, and as it turns out, a shape like this works. So by playing around with the shape of the particle, the shape of the origami, we can make sure that if you have origami in the shape of an asymmetric doughnut and a patch that looks like an asymmetric doughnut, it'll always orient itself. And we can theoretically prove that first it'll align themselves center to center, and then it will rotate and lock itself in place. And we, I mean, we can theoretically prove this. So purely from computational geometry, we can prove that this is how the assembly will occur. We did a whole bunch of simulations, made sure that it works all the time. And, you know, it increases the center, it does center to center alignment, and then it locks itself in place. And on over the right, it's just basically energy. I mean, the free energy change. And we also like the shape because it looks like this. But as it turns out, we couldn't make it, you know, that was no moon, so this is no Death Star. We made it, we had to make it look like this, you know, look like an asymmetric coin. And we end up getting surfaces like this. So this is a shape, the binding site doesn't have symmetry-breaking feature. The origami is the asymmetric donut. And if you— if I do a class averaging of every single binding event, it looks like a nice little doughnut. That means there is no preferential orientation. However, if I have the binding site matching the origami, and the reason why the origami— the binding site is circular is merely getting sharp features by lithography is very hard, so we just went with a circle. And all the structures are oriented. All the molecules, all the origamis are all pointing in one direction. And then we asked, and also as it turns out, the placement efficiency goes up. We can't find errors in like 100 micron regions. We can't find errors of having multiple bindings. And the alignment error is about 3.5, 3.8 degrees. And now the alignment error here is, this is with an absolute frame of reference. By that I mean, if I design every molecule to point due north, every one of them will be pointing due north plus or minus 3.5, 3.8 degrees. And then we wanted to actually explore this from an optical perspective. We took an optical molecule, this is a dye molecule, and if you think about, and when you create such a molecule, it will have a preferential dipole moment. In this case, it's horizontal, and the reason why the dipole moment is important is because when you have an electric field excitation, when the polarization matches, you have maximum light going into the structure and thus it becomes brightest. And when it is orthogonal, not that much excitation happens. Now, an analogy of this would be if you take 2 pendulums and apply a force, when both of them are aligned, both of them will have the approximately same energy goes in. When they are orthogonal, the energy orthogonal to the side of the direction of force will not swing. So essentially it's very similar, it's analogous. So if you have a mechanical oscillator, when you— depending upon how you excited, you have a certain amount of energy going in. Similar phenomena occurs even with optical dipoles. So we made a structure like this over here. It's this 12-pronged wheel where each wheel had origamis oriented such that the molecules will be oriented along the axes of the wheel and we— and when you do an unpolarized excitation, this is what it looks like. And when you do an excited polarization, this is what it looks like. And as you rotate the polarization, you have only a certain number of molecules turning on or off. Again, this is not an art piece, this is to basically demonstrate that we can actually position molecules, we can control the orientation of molecules as well. So, a practical use of some technology like this would be if I want to tell what is the excitation, what is the polarization of photons going down a waveguide, I just need to position 2 molecules orthogonal to each other and I can figure out what are the polarization. Depending upon the intensity ratio, I can immediately figure out what would be the polarization of the light that is going through. And just going with— and there are about 3,500 single molecules here. Just to have fun, we decided to do this. So, this is 100,000 molecules forming a shape of a character and the excitation was in this direction and when you go to orthogonal excitation another picture shows up. Right. So, this is basically just to say that we have— and in this particular picture there are about 100,000 molecules and 100,000 molecules we determine where it's located We determine its orientation and we do this repeatedly. And every part of this uses standard cleanroom fab. So there is nothing specific, there is nothing non-scalable in this process. And we can also do this on optical devices. So in this particular case, rather than positioning the molecules at different locations inside of photonic crystal, we merely changed the orientation of the molecule. So, in the middle, the dipole orientation of the molecule and the dipole orientation of the cavity were orthogonal, which means that it is— there isn't that much interaction between them. On— over on the edges, they are aligned, so you have maximum interaction between the cavity and the molecules on the origami, and it matches the simulations as well. There is a little bit variation in the middle, and that is because of dirt, and it's on the surface of this resonator and not on the inside, not completely buried in the device. This is, to my knowledge, nobody has been able to control molecules at this precision. It's not possible with any other technology to position molecules It is no other way to orient the molecules within an absolute frame of reference. This does not depend upon the crystal lattice, this does not depend upon external fields, this is modulated purely by the shape of the origami and shape of the patch. And any asymmetric molecule that can be attached to DNA can be controlled this way. So for instance, if you have a ferrocene or like if let's say you have a graphene nanoribbon nucleation site. You can use techniques like this to get it to point in a particular direction and have the growth occur only in that direction. I'm not saying I have figured out how to do the nucleation site, but technologies like this can be used for applications like that. Now, the state of the technology is going to be we are able to position the molecules. On multiple substrates, we can do it at wafer scaling, and we can do it, it's generalizable to control the orientation of the molecules as well. So now let's actually think about some of these devices that can come about from this technology. So one of the first devices that we are building is going after protein quantification for personalized healthcare. And the reason why we are looking at this problem is because currently people think of personalized healthcare in terms of genomics. So you look at your genome and from that you can tell whether you are going to get cancer or whether you're going to get Alzheimer's or something like that, higher probability. But it doesn't tell you whether you have— you're going to get heart attack tomorrow. It doesn't tell you whether you're going to get flu or whether you have some other ailment at this moment. So for instance, there are 5 instances of this guy. His genome is exactly the same in all of these cases. So the genome is exactly the same in all of these settings, but the proteome is completely different. So that is to say, the proteins in your body is going to represent what is happening at that moment. And just to remind you, genomics is currently about $20 billion market, and that is to say you are looking at one individual's genome once. Proteomics is every time you go into a doctor or every time you want to know what is the state of your body, you want to be able to measure it. So it's a much, much bigger problem and it's a much more potentially lucrative market. And so what we want is to be able to quantify all the proteins in your body or at least a subset of it in real time at single molecule resolution. We can go into the details of this offhand, but this will not work if it is not at single molecule resolution. So at a single molecule resolution, how do you actually quantify all your proteins? So we are working on this kind of a contraption which is basically 2 molecules that interacts with one protein and it's all assembled on a surface such that— so think of it as a Venus flytrap that closes only upon capturing one molecule. And in such a setting, what happens is whether you're looking at a TIRF imaging or like an electrical detection, in this case it will be like something like ion-sensitive FET or cyclic voltammetry wherein like the top lid carries some signaling molecule and when it comes close to the surface, it gives out a fluorescent signal or it gives an electronic signal. So, you end up getting time traces like this. So, it's basically wiggling around on the surface, and as soon as it captures a molecule, it stabilizes. So, and we are starting to construct these structures. Over here, we have a negative control. This is just a cartoon. As you increase the analyte in solution, you have more and more of these structures turning on, and we can get down to about 10 picomolar right now, and this was work done with Rishabh. and Rizal Hariadi from Arizona State. So, we can get down to about 10 picomolar on optical detection. And as we move forward, what we are trying to do is envisioning chips in which we have about, you know, large arrays of single molecule— the single molecule measuring tools. So, Each of these 5 by 5 grids that you see looks for the— looks for one particular protein marker, and upon capturing a molecule, it'll give you a signal. So, now you can think of this— so, depending upon the signal that you get from this, so you're looking at the concentration of not one molecule, but you're looking at concentration of N different molecules in real time at single molecule resolution. As it's changing, this value also changes. So we can actually start envisioning methods for monitoring somebody's, you know, health at a proteome level as they progress. So we can start imagining having, you know, longitudinal proteome-wide association studies. So that is to say, you know, look at a population of people, look at how their proteome is changing from every hour to hour, and ask the question, okay, this guy, or like this subset of people, went on to get a heart attack in the next 12 hours, which means that if the proteome looks at a certain pathway, pattern, good chance that you're going to have an issue like this. So we want to reimagine healthcare in this fashion. So thinking about what kind of tools would you have 15 or 20 years from now if you go into a hospital, you know, instead of putting an EKG, you would have a little catheter that you stick in your blood and that will basically give you a time trace of how the proteome or other small molecules are changing. Another area that we are exploring is large area patterning. Remember, I told you 1 gram of origami costs you $200 and it's enough to cover about 1,000 square meters. So the question becomes, can we actually use this as a mechanism for patterning extremely large areas? Imagine, you know, if you want to pattern entire facades of a building to make it look like a butterfly wing, for instance. So like butterfly wings have structural color and it's not just color, you can give it optical properties, you can make sure that it reflects all the IR so the internal heating requirements goes, or cooling requirements goes down. So all of these things are from patterning, patterning optical features on the surface. How do you actually do this at scale? So biology does it. If you do a back-of-the-envelope calculation, the surface area of the remaining butterflies is enough to cover about half Earth. Half of Earth's surface area can be covered with just the butterfly wings, just the butterfly wings. We are not talking about any other insects and structure or any other organism that has structural color. So how do you actually do patterning at that scale? Because you can't do that, you can't use photolithography or optical or even beam lithography for, you know, such a massive area. So origami allows you to have that flexibility because I talked to you about 2 ways, path 1 and path 3, to organize origami on a surface. There is a middle path wherein the origami can be assembled to form lattices and then the origami itself can be converted into an optical device or Or it can be converted into graphene, it can be converted, it can be used as a liftoff layer to make metallic devices, it can be used as an etch mask to create dielectric devices. So there is a lot of flexibility, and when you think about the design space available, since the origami itself can be very easily programmed at very high yield, this opens up a whole range of devices and a whole range of structures that can be created. So this is work that I did with a bunch of collaborators from Finland, wherein we took a surface and merely by putting— merely by designing the shape of the origami can we give it optical properties. So on the left, we had little crosses that we put it onto a surface and converted it from being DNA into gold. And over on the right, it's a chiral structure, it has a right-handed symmetry. If you put it all onto a surface and convert it into gold, over on the right it gives you a clear chiral signal, and on the left it doesn't have a chiral signal. And this never went into a cleanroom. Every part of this was done purely by designing the origami shape and using, you know, chemistry to create these structures. Another area we're going after is discrete like next-generation light sources. On the left and the right, these are 2 grids. On the left, we have every single grid point has got a single europium ion. So the reason why we had europium is because europium is in the same family as lanthanides. It's a lanthanide family which includes neodymium and erbium, both of which are used for telecom wavelength light emission. But rather than doing IR experiments, we wanted to do visible experiments, so we went with europium. Each of those spots has a single europium ion. europium bound to it. And on the right, we have single quantum dots bound to it. We have not started very detailed optical experiments on this yet. We have not done single molecule, the single photon, you know, anti-correlation and autocorrelation experiments. But we're getting there. And this is what we imagine our capability to enable, to be able to actually position individual structures and then building structures like this around So this is an experiment that we are doing with Harry Atwater and over at Caltech. Anya Mitskovich is his grad student who's working on this. We're making these kind of resonator structures and trying to position single light emitters at specific locations so that now we can get extremely pure light out of it. We don't need to rely on defect centers that have— that are Poisson limited or some kind of quantum dots that are just thrown down and looking for a particular structure that works all the time and things like that. So with that, I go back to this particular picture. What I imagine my— what I imagine structures, devices to look like 15 or 20 years from now. I think in some version or the other, this is going to become real. People are going to put quantum dots in specific locations to make quantum computing possible or quantum information transfer possible. We're gonna have like nanowire-based logic, we're gonna have these biosensors integrated into devices. I like to imagine that the way in which we are gonna get there is having a little bit of DNA underneath us. So I leave you at that, and these are the folks that I've worked with. Paul was the lab that I was working with, Paul Rotemond lab at Caltech. These are all the other folks that have helped in small and very big ways. And with that, I'll take any questions. Thank you.