SMolSTM: an open hardware scanning tunnelling microscope for creating single-molecule circuits - EMF

[applause] Hi everyone. Uh yeah, I'm Sam. Um I'm a researcher from the physics department at Lancaster University. Um and yeah, today I'm going to talk to you about um how I use homemade equipment to create circuits with just uh single molecules. So just to give a kind of a bit of a background on on the kind of scale we're talking about here, this is uh just a cartoon image of like a grain of sand you might find at the beach. Uh and it's about uh on the order of 500 microns across, so about half a millimeter. And this is about the smallest thing you you'll in your day-to-day life be able to see with your eyes. Uh but there's of course a lot going on below this length scale that's really important and affects our day-to-day lives. For example, this is a red blood cell. So these are responsible for carrying the uh the sort of oxygen around our bodies. Uh and they're too small to see with your eyes, but you could see these with a decent optical microscope. But on this particular image, you might be able to see just a single white pixel in the middle of that blood cell. And this represents the scale of the smallest things that we as humans routinely mass-produce, which is a transistor of the type that you might find in like a computer processor. So there's right now as we speak a few billion of these transistors in this laptop generating this image for you. So they're kind of making an image of themselves right now. Um, but we make these kind of the same way we humans have always made everything, which is we take a block of material and we take stuff away to leave behind the shape that we want. But nature has a different way of making these things. So, as we reach kind of the length scales where we're being limited by these processes, maybe we can look for a different approach. And you might be able to see even a smaller dot now on the side of that transistor. And this is a carbon 60 bucky ball. So this is a stable structure that exists in nature and it's kind of a football shape of 60 carbon atoms. So it's a very complex shape. Um but it uh it kind of self assembles from the bottom up and that's that's how nature makes things. They form a low energy configuration that's stable. Um and this is just 0.6 nanometers across. So 0.6 of a billionth of a meter. Um, and what's interesting to me about these kinds of of small organic molecules is that we're not just limited to stuff we can find in nature. In fact, if you work with chemists, they've developed over the years an incredible ability to make almost any kind of molecule you can imagine. Uh, the degree of customization is actually incredible. Um, there's been estimates that there's on the order of 10 to the 60 small organic molecules that could feasibly be made. So that's so many that even if we used up all the atoms in the solar system, we couldn't make one of all of them. So we're never going to run out of possibilities for molecules we can make. But I was talking about transistors earlier. So why are these interesting in the context of electronic devices? Um well the reason is not just because they're small and we can pack more of them on a chip, but because they actually give us access to new physics. So the molecules here are so small that while in a normal circuit we might think of electrons as as charged particles flowing around. Uh we're at a length scale now where we can no longer ignore quantum physics and we have to start thinking of these electrons as an actual matter wave. Uh and then we can start thinking of our molecules not as like wires or switches but as waveguides and that gives us some really powerful tools to control the flow of charge in a circuit. So if we take this example here, this is a benzene ring. Uh and there's two paths that our electron wave could take through the ring to get to the output. And the wave actually takes both at once. So in this case where the two paths are the same, they're the same length. The paths arrive in phase and we get this large sine wave where the two waves have added together. But if we just move this situation by one atom and instead inject the electron up here, now the top path is short relative to the bottom one, the two waves no longer add up and they actually cancel out. So now we have like a single molecule switch that we can actually turn on and off. So we're already seeing kind of the the groundwork of a technology on the scale of a single atom uh and reaching kind of an idea of of a of a molecule that could actually compute. So another cool thing about these molecules is you don't just have to work with one. The same way nature will create a stable configuration of atoms to make a molecule. They'll also form stable configurations of the molecules themselves to create large flat structures. So you can self assemble from solution large planer films just one molecule thick. Uh and the density that you can achieve is incredible. We can have a trillion carbon 60 molecules on the surface area of the head of a pin. uh but before we can kind of understand these large thin films, we do have to understand exactly why the customizations that we make to the single molecule produce certain effects. So to do that, we need a way of actually putting some probes on our molecules and testing them. So we want to be able to create single molecule circuits. And clearly this kind of cartoon that I've I've drawn here is isn't really practical because we we saw earlier that where you position those electrodes on an atomic scale has a huge impact on your measurement. So we need to be able to precision uh position our electrodes with subnanometer precision. This is where the scanning tunneling microscope comes in. So this was invented in the 80s and it won a Nobel prize because it provided us for the first time with a way of seeing the atomic scale in real space. Um, and I won't go into too much detail about how this works, but it's sufficient to understand that it uses a very sharp metallic needle that it scans across the surface and it measures small electrical interactions at the quantum scale between the tip and the surface. And this is produced some really fantastic, beautiful images over the years. But what I think is actually really exciting about it is not the imaging, but the fact that you have this local atomically sharp probe that you can actually interact on the nanocale with your samples with. And this brings us to the break junction experiment. So this is a way we can reliably form a single molecule junction. And the way it works is we take a gold surface and a gold STM tip and we put some molecules down on our surface and we then deliberately crash our gold tip into the surface to form a sort of a nanocale metallic junction. And we can measure the electrical conductance of that junction as we pull back and stretch it. And at the point where that junction breaks, you would expect to see the conductance of the junction just disappear. But if a molecule happens to bind between the two electrodes, then we've formed a single molecule circuit and we see a plateau corresponding to the electrical conductance of that single molecule as it gets stretched. So this is a really powerful technique. The problem is that you need an STM for it. And an STM is typically on the order of half a million to5 million pounds. And even if you have one, they're not really designed for these kinds of niche measurements. They're designed for high resolution imaging. So then you have to take your super expensive nice piece of equipment and start modifying it. So I think that one of the nicest things in experimental physics is we get to make our own stuff. So I thought wouldn't it be cool if we could just make a sort of a homemade STM that is designed just for these types of measurements and is optimized for these techniques. And in principle an STM is pretty simple in terms of hardware. You just need some long range positioning to get the tip in the right place. some short range atomic resolution positioning, a sharp tip, and some electronics. Uh, but obviously when you're working at the atomic scale, even simple things become very complicated because any source of noise or interference is going to be a huge problem. And this has kept me entertained for quite a while. This is uh I've been playing around with this both as as my research and a hobby uh for for quite a few years now. You can see some early kind of crazy looking things I made in my shed during the the COVID pandemic. Um, and then when I started to move them into the labs in Lancaster University, they got a bit less mad scientist looking and a bit more professional. Um, and the one that I'm going to talk about today is this one here. And this is the instrument that I've chosen to open source because I think it strikes a good balance between being easy to make as an individual while still actually being able to perform real sort of research grade measurements. We still use this today in our lab to do research. So here's kind of a view of what this STM looks like. You can see a cutaway render of what's going on inside there. And I'm just going to go through some of the main pieces uh and show you some of the sort of design decisions here. So, the simplest part is the frame. You need something stable to attach all of your components to. Um, but when you're working at atomic precision, you need any kind of drift and vibration to be as small as possible or it's going to be bigger than the actual thing you're trying to measure. So, the materials that are good for this, like uh sapphire or granite, aren't really very accessible for a home builder. So, I opted for a kind of a a composite approach or like a two-stage uh setup. uh where we have a steel inner frame and this is just made of some laser cut top and bottom plates. So you can just send off these from the through the post uh for not too much money and then the rods separating them are just hand cut bar stock. Um and then the issue with just leaving it here is that if you ever take a steel structure and and hit it with like a hammer, it rings like a bell. So it's got a lot of resonances which are actually amplifying vibrations at certain frequencies. So to kind of deaden those resonances out, we made a 3D printed mold and we poured this cast epoxy granite composite, which is just epoxy compound with some granite aggregate. So essentially really coarse sand. And then after curing it and post-processing, we get our STM frame that looks like this. And in the center of the frame here, you can see the actual uh heart of the STM, which is the nano positioner. So I mentioned near the beginning that we we need some long range and fine range positioning in our STM. Uh and this scan unit is based on an open- source project by another group uh which combines both of these things into one unit and it does it in a really clever way. Um so the short range positioning is done with a poelectric crystal. So this is a crystal that changes shape uh when you apply a magnetic field to it. The atoms get further apart and closer together. So you can produce really really small movement with this and it's coupled to this linear slide with a magnet. So at slow speeds that coupling is rigid and the linear slide reproduces the atomic scale movement of the crystal. But if you move quickly in one direction and then slowly in the other. When you're moving quickly the magnet will slip a little bit. So you actually get a net displacement and you can move the stage in these kinds of steps to get longer range motion. So this is how that looks in real life. And we just assembled a threeaxis stage. So we have this motion in three dimensions. It costs about £150 to build one of these stages, which if you compare it to the kind of thing you could buy from a research supplier is is about two orders magnitude cheaper. Um once you have that kind of of resolution and capability and positioning, it's no good if you can't control external vibrations. um the vibrations in a typical room will be orders of magnitude bigger than the things we're actually trying to measure. So I don't know if you can make out the yellow trace on this oscilloscope. This is the signal from an earlier iteration of this STM and this is me just dropping a small bolt uh from a few meters away onto the floor. It's a very zoomed in video. Um, but you can see when the bolt hits the floor a huge spike in the signal and that is the junction actually rupturing and the the experiment breaking down. So if you don't have some kind of vibration isolation in a normal environment, you have no chance of doing one of these measurements. So what I built into this instrument was a stack of alternating steel discs and soft bon O-rings. So at each stage, this damps some of the vibrations as they move up from the base towards the more sensitive part of the instrument. And this has been really effective in helping to keep those sort of vibrations away. So once this is all built, we need a way of actually testing that we're we're measuring uh what we think we are. And I think really the coolest benchmark is to measure the electrical conductance of a gold atom. So this is like taking your multimeter, setting it to the resistance mode and putting the probes across just a single atom. Uh it's a really cool experiment. Uh partly because you can actually with some some fairly simple maths that fits on the back of an envelope uh predict what the electrical conductance of a gold atom should be. And it turns out to be this value G, which is equal to 2 E^2 over H. So that's the charge of an electron and plank's constant. So these are two fundamental properties that should be the same everywhere in the universe that we're probing with our STM here. And if we do our break junction measurement where we're crashing the probe and pulling away and stretching metallic junctions without any molecules there, then at the moment that the metallic junctions about to break, it should start to thin down to a countable number of of atoms. So we should start to see integer steps of this 2e ^2 over h as we go from 3 2 and finally one atoms in our junction when we do this for real that's exactly what we see. So this is a real measurement from from this STM on some gold atoms and you can see the three two and finally one atom steps. So this last plateau is a single gold atom being measured in our junction. And if you do this thousands of times and plot like a histogram, like a probability map, we can see peaks at these integer multiples which are telling us that we are reliably forming uh at atomic scale junctions. So you'd think that if you can form an atomic junction, a molecular junction should be easy, right? Because this octane diile molecule that we've chosen to study is eight carbon atoms long with a sulfur atom at each end. So it's 10 times larger than the the gold atom. Uh but there's actually quite a lot of complexity in the binding between the gold and the molecule. So you get measurements that look like this where we're seeing molecular binding events and and single molecule circuits, but every circuit is a little bit different. So how do we handle this kind of variation and say something actually useful about our molecule? The typical way to handle this is to take 10,000 or so of these measurements and collapse them onto a 2D heat map here. So brighter points correspond to where there is more of this data in the data set. And we can read off the shoulder here and we can get the average electrical conductance and breaking length of our single molecule. And we can see that we are consistently forming single molecule junctions. But we then went a little bit further and we decided to apply some machine learning based clustering to this data set. So it's splitting the data up by the similarity of of the traces to each other and we get this. So that actually that big messy heat map actually it turns out is made up of a few distinct event types and these actually correspond to the different physical types of binding that the molecule can have with the electrodes. And this is really powerful now because this gives us not only um access to the average properties of the molecule, but we can also look at the different ways that the molecule can bind and see the relative likelihood of those binding uh uh modes. And that means that if for example for a future device we would want just this type of binding we can now go back and modify our deposition methods or a molecular design and we can do this experiment again and see if the proportion of this group has increased in our data set. So we now have a pathway towards iterating towards practical consistent devices. But there's also interesting physics here because uh scientifically we understand what is going on in in the electrodes really well and separately we understand the molecule very well but this kind of coastline between the molecule and the electrode is extremely complicated and still not entirely well understood. So this is actually with some homemade equipment that cost about £500 to build. we're actually probing a real frontier in understanding uh and this is what makes it so useful for for the research that we're doing in the lab going forward. Um so that was the last result I wanted to show you. Uh just um I'd like to just thank uh everyone in the the labs back at Lancaster uh who support the work and make it possible. Uh, UKRI provided a lot of funding and um, I'd also like to thank my brother Will who provided a lot of technical advice and help with the epoxy molding in particular. Um, and there is a a GitHub with some more information about the the instrument on there and I'm preparing an open access paper and a build guide uh, so that more labs and even individuals can start to reproduce this. Um, and yeah, with that uh, thank you all for listening. I'll probably be in the Q&A tent later if anyone has any questions. Yeah, thank you. [applause]