EECS Seminar Spring 2021 - Dr. Ashwin Gopinath, MIT - The Bridge from Top to Bottom

So most of this work that I've been talking about is stuff that I've been doing for the last, I don't know, nearly 7, 8 years. So part of it was— most of it was done at Caltech. And towards the end, those are stuff done with my collaborators over at Arizona, as well as at MIT. I've started at MIT only a little over 2 years ago. And in case you guys didn't notice, last year has been— Interesting. So starting a new lab from scratch during— in the middle of COVID has been an interesting experience in and of itself. So anyway, so, so what I'm going to be talking about is DNA origami bridge from top to bottom. But before I get into the thick of things and like get started with all the results, I want to kind of contextualize what it is that I'm going to be talking about to kind of set the stage, so to speak. So the first thing that is, that one needs to kind of understand is what do I mean by top-down fabrication, and what do I mean by bottom-up fabrication? It's important to actually understand that, and to understand what I mean by the bridge from top to bottom. So top-down nanofabrication is the way in which everything around us is constructed. Basically anything that is not biological, or that you see in nature, is constructed by something top-down, some top-down nanofabrication. And this is just an example, a canonical example of what can be created with top-down nanofabrication. So this is the cross-section of an A2— A4 chip that was there in one of the earlier iPhones. And at the bottom, you have the silicon layer, and these little gray spaces that you have, gray features, that would be the gates of different transistors, and then a whole bunch of other layers on top of it. A lot can be said about the size of these different features. And, you know, that's what Moore's Law is, and, you know, how small you can create these structures. I'm not focusing on that. What I am more interested in is the fact that once you construct something on a surface and you know where it is, you can register with respect to that particular feature multiple other structures. So in this particular chip, there are about 2,000— between 200 and 400 different layers. 200 or 400 different processes were done, patterns were done with respect to features that were defined on the surface. So you have something on a surface and you're building a whole bunch of other stuff around it. So once you know where something is located with respect to that particular feature, you can do extraordinarily high-controlled fabrication. And everything in that particular picture was constructed by one of these 3 methods, by repeated application of this particular surface, these, one of these processes. You pattern it, and then you either deposit a material, you etch a material, or you dope it, you change the material characteristics. And by repeated application of it, you construct this, this, this set of this entire stack on silicon, and that is how you create all the microelectronics. There are details, there are different chemistries and different amount of tricks one needs to play to get very small features and different materials and things like that. We are not going to go into that. That encompasses what is top-down, generally encompasses what is top-down nanofabrication. And the same philosophy can be used to construct micro— microfluidic devices wherein you have— you control the flow of fluids. You can do optical devices where you control, so to speak, flow of electrons. You can also have microelectromechanical systems wherein you take mechanical structures, sort of like a gyroscope, and reimagine it on a nanoscale or microscale on silicon to get accelerometers and gyroscopes that you find all the time inside your iPhones and whatever phone that you have. So that's top-down nanofabrication. On the other end of the spectrum, you have bottom-up nanofabrication, which is basically everything that is not top-down, which is not artificially constructed, falls into this bottom-up fabrication, wherein you have, in an abstract setting, you have a bunch of atoms or a bunch of molecules coming together. Let's call it self-assembly process. This could be just heating and cooling, it could be electro— like chemical setup, or you're changing the electronic conditions, and it'll basically come together to form a structure. Like, for instance, everything that you see around you in biology is some kind of biomolecular structure, wherein you have proteins that come together to form larger protein structures. Or the other example would be quantum dots, wherein you have semiconductor molecules that come together to form a nanoparticle of a certain shape and certain crystallinity that gives it certain kind of optical properties. You have carbon nanotubes. The examples are endless in— within the bottom-up nanofabrication world. And within this context, I like to imagine that the question that my lab loves to actually ask, or part of the question that, like, my— that drives my lab is, what will integrated chips look like in 30 years? So this is a particular picture. This is a MEMS microphone that you have, or some version of it would be there in many of the, uh, in your— many of the cell phones. In the middle, you have a diaphragm and a whole bunch of electronics around. So diaphragm acts as like the mic, and it vibrates, and then you— it converts it into electrical signal and then does everything else. So what will integrated chips look like in like 20, 30 years? I think that it's gonna look very similar to this, except that you might end up having like nanowires positioned to create some kind of a logic that performs much faster, much quicker, things like that. You might have like sing— like biomolecules sitting on the diaphragm so that when you are speaking into it, it's also doing some kind of biosensing to tell what is the, what are the molecules in your breath. So, or you might end up having like single quantum dots or single atoms sitting inside these kind of exquisite structures to doing, to encode the information, to do quantum information processing and like encrypt it and send it off rather than the encryption that we are using right now. Now, this seems kind of futuristic, but the reality is parts of this, these constructs or these structure, these kind of devices are already being constructed. Specifically, you have nanowires being connected together to form electronic devices. You do have optical devices wherein you have single quantum dots or single emitters positioned inside resonators to give it like interesting quantum optical properties. You have biosensors wherein you have single molecules or single biomolecules interacting with single optical devices, and you're interrogating it one at a time. In fact, on the right, rightmost column, you have this biosensor, which is part of Pacific Biosciences sequencing platform, wherein you have a single biological protein, a polymerase sitting inside an optical device. And as it's processing the DNA, you are reading out what it is processing. So it is used for doing sequencing. So that's actually commercially available. Now, the problem with these hybrid nanodevices, when you have top-down devices and bottom-up components kind of connected to each other, the problem is how do you construct it? The problem is, you know, they are— and that's where interesting properties start to happen. So the electronic device over on the top left, this was constructed by a process that, let's call it hunt and pick. What you do is you take down— take the bottom-up components, throw it down on a surface, literally just incubate it. You just position them randomly, find out one of them, and create contacts and write a paper. Or the second one is scanning probe assembly, wherein you construct exquisite optical devices, take bottom-up components, take a small little tip like an AFM or a scanning probe, move the component into position, and then study it, uh, for a certain purpose. Or the last one is spray and pray, wherein you basically make an exquisite device and then you just basically throw down your molecules and hope that one of them goes and sits exactly where you want. If you get 2, that can't be used. If you have nothing, you can't use it. You want just a single molecule sitting exactly where you want within these kind of, these nanodevices so that it can be utilized. So as you can imagine, none of this is scalable. None of them are modular. What I mean by scalable is you can't construct billions of devices and make sure that their properties are all identical. None of them are modular. What by that, whatever you learn from one of these processes, you can't apply it to something else. So there's a fundamental problem there. And it is in this context that we ask this question. And I think that some version of this is going to become a reality because the opportunities are just too much here. Like this kind of integrated systems wherein you have bottom-up structures and top-down devices kind of merging. There are a lot of interesting questions to be done, not just physics, not just fundamental questions, but applications as well are there. So some version of this is going to be a reality. And this is the quick kind of— And so my lab likes to ask 2 questions. One is, how do we actually achieve this integration? Like, how do we actually integrate bottom-up atomically precise components, whether it is a single molecule, single quantum dot, single atom, single ion, integrate that with top-down devices that we know how to construct? How do you actually merge that? And the second question, which I won't talk about much in this particular talk, is, you know, what are the functional devices and what are the fundamental questions that you can start asking once you create this capability? So that's setting the, setting the stage of what I'm trying to work on. And my tool of choice, or my material of choice to achieve this, is DNA. But before I go into DNA, I want to kind of give a very brief introduction for those who don't know about DNA, like what is DNA. And, and everything that I'm talking about, uh, as far as, uh, DNA here, uh, uh, is views the DNA as a— we view DNA as a polymer. We don't use it, at least in the context of this presentation, we are not using it as a— for its biological properties. We are using it purely from it for its polymeric properties. So from that context, the structure of DNA, it looks sort of like this. It is—

it's—

it's— it— you can think of it as a cylinder, uh, with 2— that is 2 nanometers in diameter, and it has 2 parts to it. You have one strand, think of it as one thread, and another thread that is basically intertwined. And on each of the— the backbone of the thread is made out of phosphate, and, uh, there are— and so it has certain directionality. So you have a phosphate, you have a sugar, a phosphate, sugar, and wherever— and so there is a directionality to it. And on that phosphate backbone, you have one of 4 different decorations. You have adenine, thymine, guanine, and cytosine. So think of it as 2, 4 molecules, 4 types of molecules that you are kind of decorating on a single thread. And the distance between the 2, uh, any 2 adjacent, uh, molecules are going to be about 3.4 angstroms or 0.34 nanometers. And adenine likes to interact with thymine, guanine likes to interact with cytosine. So in this context, you can take one of these DNA strands. It's a single strand of DNA. You can represent it as a straight line. You give it an arrow there is this directionality. And then you can put it— you can enumerate it with ATGC. So you have— so that, that gives you the representation of a single strand of DNA. And once you have a single strand of DNA, you will always have another strand which is in the other direction, where the arrow is pointing in the other direction, which means that the phosphate and the sugar, the, the sequences are in the other direction. And wherever you have A in one, you have T in the other. The vice versa, and wherever you have C in one, you have G in the other. That is complementarity. Adenine likes to interact with thymine; guanine likes to interact with cytosine. And this is about all that you need to know to understand the rest of the talk. So this, like forget about all the biology and all that stuff. You just need to think of it as a polymer, which in which you have this kind of directionality, and this just and and you have ATGC. That's about all you need to know. So DNA nanotechnology, this idea started in the mid-'80s or like early '80s by Ned Seeman, who's at NYU, wherein he asked the question, if you take a single strand of DNA that has 2 parts to it, abstract 2 parts of it, the first half, the one of the halves of the first strand interacts with the second half of a second strand, the second half of the second strand interacts with the first half of a third strand, and the second half of the third strand interacts with the first half of the first strand. So essentially it forms this kind of abstract cross, and there's nothing like this, like from a just base pairing interaction, what, what I mean by ATGC, you can create these kind of molecules, these 4 molecules. And then he asked the question, if you take this abstract cross and you had short sections on each of the ends such that the top likes to interact with the bottom and the right half interacts with the left half, what happens? He proposed that if you make this, it'll fall— if you, if you were to create the structure, it'll tile and create this kind of extended crystalline lattice. At least that was a proposal. And after a few years, people started actually creating this. You can create these kind of mesh kind of a structure wherein you have single strands. A single strand of DNA is about 2 nanometers, and that's what they were trying to construct. And, and experimentally it was explored, and there were several papers that played on the themes and variations of this. You can follow extended structures, you can make, uh, other kind— this is a square lattice or a hexagonal lattice. You can even do computation here, but it's not important. I mean, but the point is to create extended lattices wherein you're treating DNA as a polymer that you can engineer and you can create materials out of it. And for nearly 2 decades, from the mid-'80s or early '80s to mid-2000s, this was the theme and variation of what made DNA nanotechnology. And in mid-2000s, somewhere around 2004, 2005, the concept of DNA origami came to the front. Now, the idea of DNA origami is where you take a long strand of genomic DNA, which is basically the single-stranded genome of a virus that infects E. coli. It's called M13MP18. Its single-stranded genome is about 7,300 bases, 7,250 bases long. And you know the exact sequence. You know exactly where ATGC, what the sequence is of this long thread. The idea was to— the idea that Paul proposed was to take a whole bunch of short DNA strands, and put it together such that it'll take different parts of this long strand, pull it together, and fold it into a particular shape. And that was the idea of DNA origami. And I will go into a little bit more detail, uh, a few slides down, but this is the idea of DNA origami. And if you were to take, um, uh, animation of this process— this is by courtesy Sean Douglas, who's at UCSF, but, uh, this is not a simulation, this is merely an animation. So you have a long strand of DNA, of scaffold. Here you go. This— these are the scaffolds, and these are all the staples. You will observe that the staples, because of its sequences, will go and find its own location on this long scaffold and bring different location, different parts of this long scaffold closer to each other. So if you continue with the animation, this— you see that as it's forming, as you mix all these structures together, the scaffold will fold into a particular shape because the staples will staple different locations to each other. It will, because of its sequences, it will find its own different locations and fold it into a particular shape of your choosing. In some vague way, you can think of it sort of like knitting. Just like how you're making these kind of cross patterns in a, in a long thread, you find these knots. You— the, uh, uh, here the knots are being constructed by short DNA strands, and the long DNA strand, a single long DNA strand, creates like the skeleton of the structure. It's not exactly like knitting, But it can help you kind of think about this. So it's— if you want to go into a little bit more detail, let's actually go into the exercise of designing a DNA origami. Let's say you want to make this kind of a triangular structure. What you would do is, depending upon the shape and the length and all of that, you will take this triangular structure and then you will raster through this entire shape. This is 2D. You could do this even in 3D, except that if you're doing it in 3D, you need to raster it in 3D. And I didn't represent it because it's kind of harder to observe rastering in 3 dimensions. So in 2 dimensions, you raster through this entire surface, entire shape, and then you will abstractly represent— you'll start enumerating this. This single raster represents that long scaffold, long genome of the virus, and you know the exact sequence of the— of that particular genome. So you will put it as ATGC something. Then when you know the sequence, you know exactly what it is. Once you have that, you can start asking which shorter DNA strands you should have to hold the structure, because the structure will not be held together because there is no reason for it to stay in this shape. However, if you have short staples, you can think of short DNA strands that are complementary, the second strand that is always part of a double-stranded DNA, and the sequence that finds different locations and bring them closer to each other. So that's the general idea of designing a DNA origami. And then, and then what you do is you'll enumerate all of that Each different dot in this particular map represents a particular molecule, and then you will, you know, you'll send it out to IDT or a particular DNA synthesis company, and they make it and send it back to you. But what's so special about DNA? You can make arbitrary shapes out of it. You can make 2-dimensional shapes, you can make 3-dimensional shapes, you can make curved shapes. All of these were made from the same scaffold. You're just changing the staple molecules, and they're extremely modular. You can make 2-dimensional and 3-dimensional shapes. They are all identical to each other. So you know exactly what is the size of these structures. The yield is close to 100%. You can make this by biology. So at approximately $300, you can make a gram of origami, and that's enough to cover multiple football fields. So, so, so, so you can make this at scale. You can crystallize these structures. You can design, design it just like how Ned talked about crystallizing just DNA structures. DNA origami can be crystallized as well to make like extend lattices in 2D or 3D. Excuse me, and you can make it modify it so that it's structurally stable up to 100°C boiling water in arbitrary buffer or ionic conditions. And the quality that I like about DNA origami is its quality as a molecular bread. So remember this map of the DNA origami. So what you have done is every single part, when you're designing an origami, after you've designed it, when you are buying these different components, you will send it off to a particular DNA vendor and they'll give you back a bunch of plates, like plates of DNA, but in which each location on that plate represents a physical location on your map. And, and when you mix it all together and heat it up and cool it down, you form the origami. And the next day you decide, I want to put just 3 molecules, 3 molecules in the inner vertex. Just a note, the edge of this particular origami is 127, the inner vertex is about 29 nanometers. So next day you decide, I want to put 3 molecules exactly on the inner vertex separated by 29 nanometers. You don't need to redesign the whole thing. All you need to do is find out what these locations are, what these particular molecules are in your map, and then change it such that it has a short section, like it has a tail that can capture whatever it is that you want. So it looks like this, and you have, uh, just like by just DNA, uh, hybridization, if you have a tail that is poly-T, so you have a section of TTT, you can actually have, uh, you always can make, you, you can be sure that another DNA molecule that is AAA, it will bind there. So you can take this molecule, this AAA molecule, and attach, attach, let's say, a protein or a quantum dot or an atom or a something onto it, and you mix it together, they go and bind there. So it becomes a very modular way to arrange other nanomaterials. So, and then you are left with another origami wherein you have just 3 molecules exactly sitting where you want. Now, what does this remind you of? This reminds you of an electric— and, and just a pointer, every single end is about 5 nanometers away from every other end. So this reminds you of these kind of Arduino, kind of these kind of breadboards in which you can put resistors and transistors and construct, abstract any electronic circuit that you want. But rather than actually doing it with electronic components, this origami allows you to do it with molecular components. You can attach arbitrary structures wherever you want on this kind of breadboard. And then what you do with it is up to your imagination. And this idea has been taken for the last 10 years since origami, or 15 years since origami was invented, and people have been doing this. You can put in carbon nanotubes over on the top left, people with quantum dots and, uh, nanoparticles, put conductive polymers, make quantum dot in a particular shape, or biological molecules. On the top right is personally one of the best examples of like precise, uh, biological device created by this technique. It's a little caliper wherein you have 2 nucleosomes. So these are molecules that are used to package DNA in your body. The, uh, this was from Hendrik Dietz's lab over at TU Munich. He made this little caliper and he can move these 2 molecules, these 2 particular protein molecules, at a half an angstrom step. And just for reference, half an angstrom is basically half the radius of a hydrogen atom. He can move it at that resolution and study the interaction between these 2 proteins. So, so it's just to kind of give you the the scope, how powerful the DNA origami technique is. And, you know, but there is a problem here. There's a— because everything that we are talking about is in a solution. So even though you create these kind of beautiful electronic devices with 2 carbon nanotubes forming a nice little transistor, and you write a paper in a really high-impact journal, there is an ugly part to this as well. Because when you zoom out out, this is what it looks like. You don't know where these molecules are going to go. You don't know where these precise devices are going to go. So you made a beautiful device that you can't see anywhere, that, that, that are not easy to create with any other technology, but you don't know where it's located. So even if you make a beautiful transistor, if you don't know where the transistor is located on a surface, you can't create a computer because you want to be able to remember when the first slide when I presented The real— one of the powers of top-down fabrication is that once you know where something is located, you can make interconnects, you can turn it on, you can create logic out of it by connecting them together. You can't do this with this technique. So, so how do you solve this? And this is the problem that I've been spending the last decade trying to solve. And the way in which I solve this problem is by using chemically patterned surfaces, wherein I take a surface that is neutrally charged, or it can be any other charge. You have 2 flavors of charges on a surface. You either have a neutral background, and you have either positive or negative charged shape location in the shape of the origami. And you create these kind of patterns by just using standard lithography processes. This is, this is a technique known as self-assembled monolayer, wherein if you just take a surface and you put a particular molecule, it will form just a single layer of particular group, a particular chemical group. And these are standard in the CMOS, like a nanofabrication, molecular fabrication world. So, so, so, and, and the size of each of these features is about 100 nanometers. And the reason why this is important is because 100 nanometers are patterns very easily created in almost any cleanroom. And we'll talk about towards the end how you can do this even without cleanroom. And these 100 nanometers are in the shape of an origami such that, and, and And the way in which we typically like to create it is by using e-beam lithography. Again, we use e-beam lithography simply because it's easily accessible. We can do this with standard tools that are available in foundries. So every part of this is scalable. So we take a surface, we make it, we put a self-assembled monolayer, put a polymer on top of it, use e-beam to create patches that are in the shape of origami, and we make it negatively charged using oxygen plasma. And now why is it, and then it's important to understand how origami interacts with the surface. So if you have an origami that is negatively charged, because remember I told you of phosphate groups on the backbone, phosphates are typically negatively charged. So if you have a negatively charged surface, there is repulsion. If it's positively charged surface, there is attraction. When neutral, then it has nonspecific van der Waals interactions. But in this repulsion setting, if you have a little bit of magnesium or like divalent cations, then there is like a weak interaction between them. And this is the regime that we like. We like 2 surfaces that are weakly, that are repulsive, but when you add a particular divalent cation or divalent cations, they start interacting with each other. And specifically what happens is the magnesium bridges the 2 negative surfaces. So you have methyl groups everywhere, that's neutral, where the origami can't interact, but you have anywhere that is negatively charged in the presence of magnesium, it will bridge it. Now the reason why this is very interesting is that any given magnesium bridge, it's not stable. But once it— once— but collectively, multiple magnesiums can bridge this large DNA origami onto the surface. So you can think of this, we can model it as like a self-assembly process in and of itself. So you have DNA origami in the shape of this blue triangle, and you have a surface patch which is red triangle. Initially, it will bind weakly with weak overlap, wherein you have— it forms some magnesium bridges. But remember I said that any given magnesium bridge not stable. So it thermally fluctuates, wherein some of these bonds break, some of these bonds are, are stable, and it basically fluctuates and basically gets knocked a little bit into the patch because it's slightly more stable. So it kind of acts as like an energy minimization process, and it slots itself into that binding site. This is not just a theory, we can see this in AFM as well. So this is like a high-speed AFM of an origami interacting with the binding site. And not only that, we can actually start modeling these systems as well. You can say what is the K-on, K-off of these systems. You can say how fast they align, and if you can have multiple binding events, things like that. Over here you can see initially it gets stable there, but initially it dances around the patch. A second origami will come in very soon. It'll try to bind to the surface and then it gets knocked off. There's a second— oh yeah, there you go. Second one came and then it knocked off and it reoriented. it itself. So very carefully, when you actually do everything right, you end up getting surfaces like this, wherein you have single origami sitting exactly where you want on silicon dioxide surfaces. These are surfaces and substrates that form the basis of your computer, computer chips anyway. And we get extremely high yield. We can do this on multiple substrates. We have very good alignment. We can also do chemistry onto the surface. You— so if you give us an XY coordinate and and tell me I want to have just a single chemical reaction at that location, we can do that with like precision of about 20 nanometers plus or minus, maybe even smaller errors. And that's a capability that currently doesn't exist anywhere. So, so after that, we decided, okay, how do we show— how do we demonstrate the scalability? So what we did is we basically took this general idea, and I'm not going to go into details of exactly what device we we created, or like a photonic crystal, but we created about 65,000 different photonic crystals. Each single pixel, we think of it as 65,000 different devices that emits a certain frequency and a certain intensity. And the reason why it emits a certain intensity is because of the number of molecules sitting inside that device. This was just by, you know, this was a picture that I really like, it's painting that I really enjoy. So I decided, okay, fine, this is what I chose. So each single pixel that you see here, each light source that you see here is a photonic crystal cavity inside which you have— you are putting just a single origami. And by changing the position of the origami and the number of origami inside this cavity, you can change the intensity. So if you want it to be really bright, then you want to be able to put up to 7 different origamis at specific location. But if you want it to be dark, there will be no origami. But if you wanted to have, let's say, a unit of 1/7 brightness, then you'll just put a single origami. So the reason why we did this is it's not an art piece. It is to demonstrate a capability. It is to demonstrate that once you create an optical device, you can give it a property based upon how many molecules you put there. It is to demonstrate that we have the power to put a molecule exactly where you want and do it using processes by which you create normal optical or normal microelectro devices. So this is the first instance of deterministic integration of molecules with devices. So once we got to this point, we were like, is it enough to control just the position? Like, it looks good, it looks like a good, good, good, good, uh, bunch of figures, and we have— we do this very reproducibly on mass, large scale, but it's not enough. It's not enough because if you look at it carefully, since the origami itself has— is a symmetric structure, it has up-down symmetry. Remember, it's a thin 2-dimensional structure. And it is a symmetric 3-fold symmetry, which means that what looks like a single binding looks like— actually, just a sec.

I'm good.

So what looks like a single binding is in reality one of 6 cases. So you can have 3-fold symmetry, you can have it in any of the 3 orientations, or it can be flipped and be in one of 3 orientations. What that means is that if you design a structure to look like this, if you want— if you were to have a symmetry-breaking feature on each of the origami, this is what it actually looks like, because you will not be able to actually figure out what is which direction any of the origami is going to go and bind on the surface. So it's important to break the symmetry. So we need to have— if you want to orient something, or if you want to position a molecule, you need to have absolute orientation. So you need to basically do this with asymmetric origami. In this particular case, it's like an asymmetric right triangle. So the big question is, how do you do this? There are 2 challenges. Remember, this is actually a 2-dimensional particle, which means that it has up— you need to break the up-down symmetry. and you need to break the in-plane symmetry. So the first challenge of up-down symmetry is if you take an origami, you have a right-handed face and a left-handed face. Think of it as an asymmetric right triangle. If you have a right-handed face and you have a left-handed face, you have 2 different faces to this. And so how do you actually control that? Since it has the same phosphate backbone, the probability of each of the face binding is identical. There is no reason why it should have an asymmetry. However, remember that we have arbitrary control over the position of the, of the, of the DNA molecules. We can make sure that there is polymers only on one face of the origami. So I can choose the face of the origami, whether the right-handed side or the left-handed face, and I can put a polymer on top of it. I can put polymers on top of it, which means that entropically it's more difficult to stabilize the polymers underneath than it is to have in a position— than if the polymers were pointing out upwards, which means that if the polymer— the side with the polymer will basically face upwards. And this is the theory, at least. And when we started doing the experiments, that's exactly what we saw. If we completely cover one of the face with a polymer, that face points upwards. And if we cover the other face with the same polymer, polymer, that face lines up. And then, no, that's as good as breaking up-down symmetry. We have— essentially, this is proving that we can break up-down symmetry. We did a whole bunch of analysis on figuring out how this happens and, you know, how— what are the rate at which it changes, all of that. And, uh, uh, uh, and, and, and, and that was all, you know, clearly understandable. Then the question becomes, how do you break the in-plane symmetry? What do I mean by this? So once we broke the up-down symmetry, we were like very happy. which means that this will work great. So we decided to basically make a surface wherein you had binding sites and you have the origami. So you had the right-handed origami, made sure that it will go— it's going to bind only right face upwards, made a binding site that matches, did the experiment, and this is what it looked like. Didn't work. So, so we were like, you know, we stared at this result for a while, we didn't understand it. But when we did some analysis, we realized that any single binding event always fell into one of 6 cases. So what you have seen here is this is like an average of all the bindings, single binding that falls in this particular phase or in this particular mode. So it always forms in one of these 6, 6 positions, 6 possibilities. And this is very interesting. So we decided to actually write a small little simulations to this to kind of explain this. So if you have a symmetric origami and a symmetric, symmetric binding site, that in blue and an origami that is red, you know, it binds partially with partial overlap, and then by thermal fluctuation it increases the area of overlap, which means that the energy minimization process. So as the area of overlap increases, the energy— it's like an inversion of the energy landscape. So it, it smoothly increases, it smoothly falls into this kind of locked position. So think of it as like an energy curve. It smoothly— the energy minimization is a smooth transition. However, if you look at the asymmetric structure, you start realizing that it starts somewhere. Let it repeat. Okay, it starts somewhere and it increases the area of overlap at constant, which means energy is reducing, free energy is reducing, reducing, reducing, and it gets into a plateau. Because it gets into a plateau wherein no matter what wiggle, how much it wiggles, it doesn't gain any energy. The free energy doesn't change, which means that it's a trap. It's an energetic trap So, so as it's going through this energy minimization process, it can get into traps. And then what we did is we, we did these simulations multiple times and figured out what are the probabilities of falling into a particular trap, and then we quantified it. And as it turns out, we can predict exactly what the traps are. We can predict what— if you give us 2 shapes, we can predict how those— how these molecules are going to interact with each other, how the binding site will interact with your Now, what does this tell us? It tells us that we understand how the process works, because if we didn't understand it, we couldn't predict it. So we know how this kind of origami interacts with the surface. You have a binding site of a particular shape and an origami of a particular shape, we know exactly how they are going to interact with each other. And using this, we ask the question, can we make a shape that is asymmetric and always have a smooth energetic minima into a single position, because that, that way it is asymmetric, which means that you have broken the symmetry and you're making sure that there are no kinetic traps. And as it turns out, these kind of asymmetric donuts meets the criteria perfectly. So the— and we did a whole bunch of simulations wherein it looks like, you know, what happens is it wiggles around initially, the centers align, and after the centers align bind, it twists and locks itself into place. So under all conditions, it'll work. And we can also theoretically explain why— prove that these shapes— there's an entire class of shapes that behave this way. Uh, and then we did the experiment. So this is a case wherein you have an origami and you have a binding site. The binding site doesn't have this symmetry-breaking feature, but the origami has the symmetry-breaking feature. And then we took all the single binding events and averaged it. And it looks like a donut. What does the donut say? The donut means that there is no preferential direction to the origami binding. However, if you do the experiment wherein the origami and the binding sites are— both have the symmetry-breaking feature, then you end up seeing all the origamis lining up and pointing in the direction that you want. And the reason why here— I mean, if you are a careful observer, you'll basically say, why is this a square and this is circle symmetry breaking, that is just a, you know, artifact in the sense that it's easier to make sharp features on origami, whereas making sharp features on a binding site is very difficult. You need to remember that these are the symmetry breaking features about 30 nanometers, so making 30 nanometer features with very sharp edges is very difficult. And the placement yield is greater than 99%, the alignment error is about 4 degrees, and this is an absolute alignment. What that means is if you get a particular molecule molecule there, it is absolutely— it's perfectly— if you, you can give me a direction, or like with an absolute coordinate, and give me an XY location, and I can put a molecule, a single molecule there with greater than 99% yield, plus or minus error of let's about 10— 5 to 10 nanometers, and orientational error of plus or minus 4 degrees. So you can control all the different free parameters of a particular molecule on a surface. And we can do this on, on a full wafer using processes that are used and scalable. And, uh, to prove this, we went, we went and we, we started having fun with this. So the one of the experiments that we had was basically making what we call as a DNA polymerimeter, wherein we have, we created this structure with 12, we, 12, um, arms, a wheel with 12 arms, wherein each arm has got origamis such that if you put a molecule, they are oriented in the direction of the arm. So every arm has got origamis carrying molecules pointing in a particular direction, and when you excite it with unpolarized light, all of them light up. However, when you excite them with a polarized light, only the molecules that align with the excitation polarization will light up. And when you rotate the polarization, only the molecules that that interact with those particular molecules that are aligned with it light up. And you can do this. I mean, and this was done under oil, so, and fluid. So that's why you see like junk kind of floating around. And then we were like, you know, can we actually do a little bit more fun with this? Okay, this was about 3,500 individual DNA molecules. And, you know, and, and they all emit— I mean, in which, you know, they, the, the the, the, um, you know, 3,500 different origamis were put, and each origami carries, uh, light-emitting molecules that are oriented in a particular way. And then we decided to have fun with it, uh, or— okay, I don't, I don't have those figures. I mean, I— we created a bunch of Star Wars pictures, but I'm gonna— I mean, just for the sake of time, I'll just ignore it. We can talk— I might show it later. So then what we did is we repeated the experiment with the same, the, the same device that we created, the starry night. But now, rather than changing the position of molecules, we changed the orientation. So we had a single origami sitting in the middle of the photonic crystal cavity, and we rotated its orientation. So we have a molecule, we have multiple copies of the same device. All of them have the same number of molecules sitting, sitting at the exact same location, and we are merely changing changing the orientation of the molecules. And the reason why this is interesting is because inside the photonic crystal cavity, you have a distribution of electric field. The electric field distribution looks sort of like this. There are 2 parts to it. You have an Ey component, electrical components, and you have Ex component. Obviously, if you have a molecule sitting in the middle, it's going to interact only with the Ey component, which means that is because you have the dipole moment and electric field, and it's a cross product— it's a dot product between them. Which means that it has a cosine theta interaction between them. So when they are oriented with each other, when they are aligned with each other, this value is maximized. This dot product is maximized, whereas when it is out of— when it is orthogonal, this particular product is going to be minimized, which means that it goes to 0. So we did a whole bunch of these cavities. In the middle, you have them orthogonal. Wherein it— this is how the AFM looks like. You have the cavity, you have the origami pointing in the, uh, x-direction, and on the edges you have the origamis such that it is pointing in the y-direction. So you have maximum interaction between the light molecules in the— on the origami and the cavity on the edges, while minimum in the, in the, in the middle. The reason why it is not zero is because the, the dipole orientation is not perfectly planar. It has out-of-plane components, which is why you have like some light still coming out of it. And when you compare it against the experiment and the simulation, it matches almost identically. So this is like the first instance wherein you are able to control independently the position and orientation of molecules, and at the same time integrate that with like optical devices, or potentially even electronic devices. The reason why we didn't do electronic devices was purely because, uh, we didn't have, uh, you know, I, I, my background is optical physics, so I just decided to do there. I mean, electronic devices, we will go into it soon, uh, uh, but, uh, you know, it's still work in progress. So once we reached this position, we were left with another question. Can we actually do this whole process without getting into the clean room? Because there are a lot of reasons why you would want to position individual molecules at specific locations, but you don't want to spend the money to go into the clean room and do all of that. So we wanted to democratize the whole process. And also, personally, I am interested in self-assembly. How far can self-assembly be pushed? So we decided to actually create these molecule arrays single molecule arrays for less than a dollar. And the idea was to take a whole bunch of microbeads. So you take a substrate, you put microbeads. So these are just polystyrene beads that you can get like essentially liters of these for like a few dollars. And if you put them together in the right condition, the right surfactant conditions, and you evaporate it, you end up getting a self-assembled monolayer of these polymers. So these beads, and from the side it looks sort of like this. And what is interesting over here is the, the region that is occluded by these particular beads depend upon the size of the bead. So depending— a larger bead and a smaller bead— a small bead will, will shadow only a very small region, whereas if you have a big bead, it will shadow a larger region. And the distance between center to center distance is determined by the size of the bead. So, and so what you can do is you can take a surface, put the beads onto the surface, and do a self-assembled monolayer. So you can actually— these, the beads will shadow, which the wherever the beads are touching the substrate, nothing happens, but everywhere else you can put a passivating layer. So you can make the surface, you can start with a surface that origami is sticky to, and then basically passivate all the regions other than where the beads were bound. So you end up patterning the without actually using any cleanroom technique. And then you flow in the origami and you're left with single origami sitting exactly where you want. And you can, I mean, and when you zoom into the surface, zoom into a, into one of these locations, you have just, you know, origami, just you can see single origami sitting. However, you don't have orientational control, but that's a different matter. And the errors here is depending upon, you know, it's somewhere between 7% 70 and 85%. Again, the reason why you have lower yields here is because the process itself is completely by self-assembly. There are errors. The chances of errors are a lot more here. But either way, the cost of this whole process is less than a dollar, and you can get surfaces like this, wherein each— this is an optical experiment where every single position has got a single molecule. This we have just, we have just, you know, aggregated. We have just— it's basically an average of the entire experiment. Wherein we have somewhere in the vicinity of 10 to the 4 molecules being simultaneously observed over a single field of view. This is fundamentally impossible using any other technique, like you— and we do it for less than a dollar. And the reason why we were doing that is to open up single molecule biophysics experiments, to, to, to make it easy for anyone to do single molecule biophysics experiments. So you don't need to go into the clean room, you don't need to spend a of money doing these patterning, and you can get a lot of data. And then what you do with the data is up to you, then depending upon how the experiments are set up. So, so in summary, you know, we— and I showed you data on how you can position molecules on arbitrary substrates using this origami process, how this is scalable, you can couple it to wafer-scale integration tools, and then I talked about how we can control orientation. So this is the Star Wars picture that I was talking about, wherein we made 2 different pictures where you have Darth Vader and Yoda. They are both basically in the same region, but if you put it in one polarization, you will see Darth Vader, and when you put it in the other orientation, you end up seeing Yoda. So, um, but, you know, that's a long story as to why this, this didn't show up in the paper. And, uh, you know, that's, that's a, that's a completely different story. And then we also are starting to do this whole process using completely without cleanroom-free, without ever getting into the cleanroom. So I'm going to pause there. And, you know, I'm going to pause there and basically take any questions that you might have. So, and the people who are all part of this, it's basically, I was in Paul's lab, you know, all these guys who are funding us. And Andre and Evan were involved in the Starry Night paper. David and Chris were very helpful in trying to get the proving that these kind of donuts, asymmetric donuts, help us orient the molecules. Harry and Anya basically were involved in the, in the orientation experiments. Rishabh and Rizal were helpful in the cleanroom-free experiments. And Jamie and Byungjin, they are at Caltech, They were working alongside, helping with some of these experiments over the time. And these are guys who, you know, funding agencies were generous with us over the last many years. And this is like MIT and Caltech have been, you know, been gracious with their facilities and, you know, giving us a chance. So yeah, I would love to take any questions.

Well, thank you very much. It's a very fascinating talk. And Now is your chance to ask questions. Okay, why are people thinking of questions? I mean, okay, I think Hang is— Okay, so yeah, this is very interesting, you know, and what's the near-term application for this?

Ah, so the near-term applications, let me see if I have the— I might have some. Oh, okay, the figure, let me, let me reshare it. So the near-term application for this is mostly in sensing of different kinds. Specifically, we are creating— there are different— so, so for instance, If you want to be creating, let's say, like we are. So we have a startup wherein we are trying to do protein measurements. We want to be able to take— make a single surface or a single array wherein we want to be able to measure single molecules, like count proteins and small molecules that are there in biological samples. So that's a near-term application. Another near-term application is basically trying to, from a, from a non-biological setting, it's going to be on doing things like quantum information processing, wherein you require single light emitters or a single quantum processing element, sort of like, let's say, a single point defect in crystal, in a silicon crystal. Yeah. You know, how do we actually manipulate single quanta of, you know, either a light emitter or an optical, like an electronic defect. All of those requires point modifications or like single unit modifications at specific locations. All of those areas, we can have immediate applications. More longer term, this has applications in like trying to make like patterning and all these other things. But for the time being, anywhere that you require single molecular interactions, the technique allows— technique is applicable.

Oh, okay.

All right.

Yeah, that's, that's good. Very interesting technique. Yeah. Okay, any other question? So I'm just— quick question about— you mentioned that, that the the origami structures can be stable at 100 degrees Celsius. So what trick do we use to make sure that they're not dissociated at high temperature?

Oh, so this is, this is the beautiful work by Hendrik and Hendrik Dietz's lab at TU Munich. So the way in which he does it, he puts poly-T, so he has this TT optical TT dimerization. So if you you have 2 thyamines that are adjacent to each other. If you hit it with like about 306 nanometers UV irradiation, the TT, both the thyamines will dimerize and form a particular— it basically covalently links. So what he has done is he has taken like small TT overhangs onto every single staple. This is a published paper. It's a, it's a Science paper from I think last year or the year before that. He took all of these, like, staples, and he has like T, like a 2T modifications at the edges, and he hits it with UV, 306 nanometers, and it just basically crystallized. It covalently links the entire structure, and that is not going to dry up. I think the first author is— first author of that paper is Thomas something. Okay. First author is Thomas. Last author, all authors Hendrik Dietz. It's a science or science advanced paper. Gotcha. It's pretty neat.

Like, so then for your photonic system, and let's say you create a wheel with different polarization orientation, does it mean that you have to create a lot of different origami?

No, you don't. It's it's the same origami. It's the same origami, and the reason why each of them light. up differently is because their orientation of the molecule is different. So a good way to think about this is sort of like, so you can think of a dye molecule, a good way to think about a dye molecule is, uh, you can think of it sort of as like a pendulum. So if you were to think of it as a pen— so, so a good way to think about it, let me actually just, uh, let me one second, uh, so So let me share it again. Okay, so this is what the origami looks like, and in there what we have done is we have taken an integrating dye. It's a TOTO dye. It basically integrates into the DNA strand, DNA strand, and what that means is every— since all the helices are pointing in the same direction and the integrating dye is gone in— is on the orthogonal plane, so if the, if the dye is going in one direction The, the dipoles are all in the plane orthogonal to it, which means that the net dipole of all the molecules in here is going to be in a plane orthogonal to that of the helices. What that means is, if you were to take a dye and excite it with an excite— with a particular beam and get an emission, and it has a certain dipole orientation, the electric field— when the electric field is aligned to it, it has maximum interaction, and when it is orthogonal, it is minimum interaction. This is very similar to pendulums. So if you have 2 pendulums and if you put force into it, it, when you put force, both of them will excite, get excited. But if you have one pendulum that is orthogonal to other, and when you put force, only the pendulum that is orthogonal, that aligned with the force will oscillate, the other one won't. What that means is that in this setting, when you excite it with a light, when you excite it with light on these arms, on the left arm, on these arms, you can see my cursor, right?

Yes.

So only the, only the, only the molecules in this particular we are— and this particular arm are aligned with the polarization, which means that maximum energy gets excited. It gets maximum energy in, into the dye and maximum emission. Whereas on these, these particular arms, since they are orthogonal, it doesn't excite it at all. However, when you rotate the polarization and it gets to the middle, now all of a sudden it can actually— those can interact with So as you rotate it, you're changing which molecule gets excited by the light. So you're keeping the same molecule, you're just rotating it. Ah, I see.

Gotcha. That's pretty neat. So mainly you have one structure, but it's because you control the orientation of the, uh, 360-degree of that, uh, donut origami. Then by rotating the, the polarization light, you could be able just to excite them with different intensities.

Exactly. Exactly, exactly.

Pretty neat. Okay, so now, um, you also mentioned about the, um, the case where you try to break the symmetry, right? And then you notice that that particular structure, you have 6 different positions.

Yeah, so that, that's, that is, that is, uh, so that is partly I think that what you're talking about is this one, this figure, I suppose.

Right.

Oh, is is not this the one that?

Not that the one with the the non-symmetry. There are.

I see. I see. I think it's this one. Yes. This one.

Okay. So so why six? Like how do you six?

It's it's just a it's just. it's just the shape. If you change the shape, you will have other kinetic traps. It's just so— how— it's just a, it's just a fluke that, uh, this particular shape had 6 kinetic traps. If you change the shape to some other shape, some other asymmetric shape, you will have other kinetic traps.

I see, because I'm, I'm, I'm in— I'm looking into a, a bonding between streptavidin and biotin. Okay, so try to— they have 4 binding sites.

Yep.

And if you attach a biotin with DNA label, then they also have the preferential binding. And they, and then they could form the obtuse angle, and they could form a linear angle. They can basically, you can control the angle, how the biotin— and I just want to say, if we could control that, like, sometimes I want to only have Just only 2 biotin, and I want to ignore the other 2 so that I can form a chain-like structure.

I, I, I, I, I, I don't know the exact interaction. I don't, I don't know off the top of my head how the, like, exact binding and, you know, what are the sizes involved, but I suppose there might be ways. I mean, we can talk offline about how that can be achieved.

Gotcha. And, and could you also comment about how the, the scanning probe tips affect the position of the— let's say, because you mentioned that they are not— the, the magnesium wouldn't bind to the—

Unless you hit it very, very strongly, it's not going to affect it much. We can— I mean, unless you— unless— it depends upon, it depends upon the exact, uh, uh, um, Scanning conditions, but you can scan it in a— you can be very soft about it. You can be in a condition wherein it doesn't affect it at all.

I see. So independent of the external force from the tip, the orientation naturally—

Yes, yes. Okay, great.

Oh, that's pretty awesome. Okay, any, any other question from audience? Okay, if not, let's thank our speaker again. And yeah, so happy that you could give us an awesome seminar.

Yeah, okay.

Yeah, I don't know, I mean, for those who are not— who shy that you want to ask ask questions after the talk. I'll hang around if you have any questions. Perhaps the speaker could hang around for another 5 minutes.

Sure.

If not, we could end the meeting at, let's say, 5:10.

Okay? Okay, that sounds, that sounds good.

Thank you. Thank you.