 The ambitious goal to make your own wafer, well, you might be in the right talk now. In the following lecture, I couldn't resist our presenter, Nudl, will explain how to perform electron beam lithography, do it yourself. Let me switch over to Nudl, who is really down to the atomic level. Of course. So hello. Hi there. Hi. Talk us about how to make your own crappy PMMA-based resist for EBL. So it's not that it already works perfectly, but you know, it's kind of a start. Our primary interest is in kind of on-ship photonics, if you will. So by that, we mean integrating photonic components like waveguides or gratings on a ship scale level. And we're also interested in making our own MEMS devices. So yeah, I guess that sums up what we're trying to do. Of course, you could also use EBL and masks generated by it for semi-conductor stuff. But that's kind of off our scope, to be honest. But also very fascinating stuff, actually. So we want to make EBL at home. Besides an SEM, which is actually relatively cheap, between €400 and €1500 if you lurk long enough on eBay. So besides an SEM, you need the chemistry. And the thing is, the chemistry the pros are using for lithography in general is price-wise pretty hefty. If you go to those go-to vendors, it easily adds up to much more than the price for a used shitty old SEM itself. So what to do? Right, cook your own soup. And maybe it's good enough for your and our purposes. So what is EBL? EBL stands for electron beam lithography. In the end, it's simply using the rig we all know for scanning electron beam microscopy for lithography. The scanning beam changes the chemical properties like the solubility of a resist. And that's the way you can draw figures into the resist with a beam. What sucks? It's super slow compared with other lithography principles. The electrons are charging the resist and maybe interfere with a substrate or whatever you want to EBL in a bad or maybe destructive way. To prevent this, you could try to coat your stuff with a thin, hopefully not disturbing conductive layer. But sometimes that's not possible because it could undermine what you're actually trying to accomplish. It's also necessary that your workpiece is somehow vacuum compatible. If not, maybe it's not the right process for it. On the pro side, you have the fact that it needs no mass. Of course, there are other maskless processes like using DLP, DMD based projectors or laser writing, awesome stuff to be honest. The other thing is that in principle, you can achieve pretty high resolutions with EBL. Because you're using electrons, you can neglect the de Broglie wavelength above 100 nanometerish resolutions completely. There is a wide variety of resists you can use for EBL. They have different fields of application. We have tried nitrocellulose NC, but it didn't work. So we switched to one of the most common EBL resists, PMMA. PMMA comes in chains of different lengths or molecular weight. You can cut or cross link them. As far as we know, PMMA is the best documented resist for EBL. It's easily available and cheap. You can use many different solvents. Some of them aren't really toxic. If you let modern resists like ZP side, PMMA can achieve the highest resolutions and aspect ratios. We use PMMA as a positive resist. That means that the electrons induce a chain-cision process, which cuts the PMMA chains into smaller parts, which makes them easier soluble in the developers. The positive process itself is relatively easy. You apply the resist via spin coating. Then you have to pre-bake it to remove the remaining solvent from the resist layer. After that, you put it in your EBL system and expose it. Then you remove the exposed resist parts with the developer. We are using MIBK for that. After that, you can do the next process step like physical vapor deposition or electroplating to get the final structures you want. The developers' composition we are using is IPA with MIBK 3-1. There are many applications of structured resists. You can use it as a mask, as already explained, but you can also use it as a structural component for chemistry on ship applications or in MEMS or as a structural dielectric for high-frequency and optical components. As the EBL system, we are using an old SEM, a kind of retrofitted it using a red pitaya as the signal generator. To design your masks or geometry in general, you can use every CAD tool chain you want. If you use hardware like the red pitaya, you have access to a wide variety of tools and scripts doing the boring stuff like file parsing for you. We should also mention the most common mask file formats. First, there's GDS2 and the more modern OASIS. And then there is an idea we had during the last days that it could be worth it to try to define yet another file format for that task, simply because those two that we mentioned are both lacking many modern features, which you know from modern CAD formats. There are multiple strategies to scan or expose only the parts of the resist you want to be exposed. The most common ones are line-style scanning with a fast beam on the dark parts and a slow beam on the bright parts and scanning fast between the polygons and slow inside the polygons you want to expose. We are using red pitaya because they're readily available and do the job. Using simple Jupyter notebooks makes it easy to script your experiments. Electron optics is pretty similar to normal optics from the user perspective, so you also have distortions and such effects. But you can transform your exposure scan to compensate for that. For finding the so-called camera intrinsics and also for fiducial detection, if you are in need of multiple exposure passes, we recommend OpenCV. Most algorithms you need are readily available. Another annoying effect comes from the fact that the beam itself is not infinitely sharp and from the scattering of the electrons in the resist and back-scattering from the substrate. That's really the resolution killer if you go for high resolutions. There are different ways to counteract that. One is to increase the electrons energy to make the entry region tighter. An additional way is to correct the effect via deconvolution. To generate the necessary deconvolutions kernels, you can measure the effect or simulate it. PECEBL is one open source tool for doing that. And that's one form of PMMA we used. You can easily buy it on Amazon. Normally it's used to fix fingernails, I guess. DCM is one very popular solvent, but it's also not very healthy. Xylol also works, and it's not that bad. Toluol is pretty good, but also relatively toxic. Anisol could be the solvent of choice. It's not toxic relative to the other stuff. And it works good with PMMA. But you have to heat it up to 70 to 90 degrees Celsius. That's our improvised spin-coater. It's a Dremel with an SEM stub holding the substrate piece. After spin-coating, you have to pre-bake it at approximately 100 degrees Celsius for a couple of minutes to remove the solvent. OK, that's the test pattern. And our first result, not that good. The reference square is 100 micrometers. Another try. The distance between the scale lines is 10 micrometers. One example of accidental crosslinking. Most of the resist was removed, but some parts with high dosage are of decreased solubility. That's a better example. Another example of accidental cross linkage we made last night. Both squares are 10 micrometers. And thanks to Sleepy Old Joyce for providing us with the wafers. So thank you. Yeah. Thank you. Any questions? Thanks for the talk. So let's have a look. Do we have questions? No. We don't have any questions here at the moment. Guys, if you want to ask something, if you want to ask Nule, go ahead, use our communication channels, Twitter, Macedon, and IRC. Otherwise, let me quickly ask you, Nule, what's about the resolution range you can get with this technique, with yourself technique? Your question is how far we got, right? Yeah. So yeah, in the sub-micrometer scale. OK. But it's still work in progress. So yeah, if we figure out how to achieve 100 nanometers, I hope. Mm-hmm. So somebody's interested if the slides of this talk are going to be published? Yeah, I can publish them. OK. How can they be found? I put them on GitHub. OK. And link it on our Twitter account. That sounds good. The Twitter account, yeah, is available. So we got it. Thanks. So far, no more questions. Yep. Let's see if something drops in. Is there anything you want to add to this talk here at this moment? No, not much. OK. Yeah. So yeah, there's one thing. You mentioned one substance which was not as healthy as it, well, should be or could be. So what's the toxicity of this one component you mentioned? Which one? Toluol or DCM? Both are, yeah, give you cancer if you're not. Oh, OK. Yeah. So gloves and everything. OK. Well, yeah. And I would recommend a fume hood and stuff. Yeah, yeah. Well, the kitchen's possible. But you don't want to mess around with those chemicals in your kitchen, I'd say. Yeah, that's true. OK, more questions. That's good. So we have some questions coming up. What structures, actually, do you want to achieve by the end of this project? Is there anything you have in mind, something dedicated, dedicated goal you want to go for and say, OK, we got it now with this type of process? Optical gratings for 400 nanometer laser light, certain application waveguides, and gratings. Oh, OK, that's cool. To implement kind of external, small external cavities for diode lasers. Yeah, OK, OK. Are those lasers also, do it yourself type of lasers or do you buy them or do you have a lab accessible for that type of work? I buy the diode lasers, so normal laser diodes. OK, yeah. Fabry Perot, the laser diodes. I want to modify them, yeah. Yeah, sure. What's the output power of those laser diodes? In the load, I want to use them, yeah, 10 milliwatts, not very high power. Well, I think you can already destroy your eye with that. Can you? Maybe. And I shouldn't look in. Not sure. Well, we don't want to experiment that. We don't want to recommend anything like that. Well, folks, you're out there, you know it anyway. Next question is, what does the wafer cost IRL and how much is an electron beamer? So our electron beam device is an very old electron microscope, a scanning electron microscope. And those are around 500 to 1,500 euros on eBay. On eBay? Really, OK. Yeah, used ones, pretty old ones. Sure. And yeah, you only have to attach a signal generator, like a red pitaya or something like that, and you have a crappy EVL rig. So how much power does it suck? So what's the? Including the pumps, it's two kilowatts, one kilowatt. Wow. That doesn't. I haven't measured, so I'm not sure. But nothing special needed so you can actually Yeah, exactly. Plug it in more or less. Cool. Well, have you actually tried optical UV before using EBL? No. Why? Just because you're interested in electron beamers? Yeah, that's one thing. The other thing is that I'm not sure how to achieve the resolution I want to have. And I guess in all in all, an EVL system is a more simpler system. OK. OK. Micrometer resolutions. I see. All right. There's another question here. It's about frequency or so. How fast can the IC switch? How fast can you actually clock it? What? Well, yeah, somebody's asking how fast you could actually switch. I mean, if you produce an IC with that technique, how fast can you? OK. So we don't want to produce semiconductor devices like processors or something like transistors. It's more about means with micromechanical devices and optical devices. Oh, OK. That's our focus. I see. I see, yeah. Next question, of course, where can you buy the chemicals? Do you need some special license for that? For example, chemical license or so. I mean, some poisonous license, at least. In Germany, they are available. So you can order them at some online shops, like S3 chemicals, if someone is interested. So normally, if you are a larger research facility or something like that, you are able to order them at Sigma Eudrich, but they don't talk to mortals like me. So I see. OK. Next question, so they're flowing in now. That's pretty nice, actually. Do you package the IC somehow? Not yet. So it's not IC in the common sense. Currently, I'm figuring out how to mix the proper resist. So I'm in the very early stage of doing it. So no, I see packaging necessary currently. OK, yeah. How long have you been actually working on that project when we're talking about timeline now? Yeah, a couple of months. So, yeah. That's pretty short success then. Another question here, how do you bond those parts? Actually, what material do you use to pack them up? Something like epoxy? So at the current stage, there is no bonding. So but you can use epoxy glue to glue the chip or whatever on the holder. Yeah. There is another question, more like a motivational question. If you could probably name some exemplary projects that can be tinkered together, which wouldn't be possible without that type of approach you are pursuing here? Yeah, compact optical devices. So I guess it's the only way to make that. OK, goes back to your answer before. That's my motivation. So I know that there's different other stuff you can do with it, but I'm not into it. Yeah, so you're more into optics. OK, there is another question that got in. Would it be possible to do those chips with gates and everything or do you need other more exotic stuff? Yeah, for that, which isn't available. If you want to build semiconductorships with transistors, you need a diffusion or ion implanting and such stuff. So it's much more complicated than only the lithography for the masks. OK, and well, just my imagination. Where did you guys actually set that all up in your cellar, in the garage, somewhere, some rented space or headspace? Where are you actually doing that? Our laboratory is more or less in the basement, if you want to. OK, is it somehow, well, how shall I say? Is it can it be influenced by motion, not by motion, by disturbed, basically, by vibrancy? Yeah. Yes, if you go for really high resolutions, we are far away from that kind of super high resolutions. You have problems with oscillations induced by some traffic or something like that. Oh, OK. Traffic. Yeah, exactly. Something outside shakes a little bit. That would be a problem if we want to crack a hypothetical boundary like 10 or so nanometers, I guess, if that's not easily possible in a normal room. So in a normal house without the proper isolation. The theoretical limit with this technique would be achievable, 10 nanometers, that's pretty small. Yeah, theoretical, it's possible, with EBL in general. Not with our rig, but with a professional one, you can get down to five nanometers or less. So I don't know what the actual limit currently is. So yeah, my question popped up. So do you further treat the finished wafer after etching? So something you do there? My plan is to electroplate some structures on it, electrochemical plating and physical vapor deposition. That are the two things I want to do in the first place. And maybe a little bit of unisotropic and isotropic etching of substrate. So but first, I want to have a reliable and repeatable resist. OK. Yeah. And somebody is interested if you could actually use, if you could dose the material somehow to get fats. OK, yeah, that's the semiconductor stuff. So with diffusion or iron implanting, that would be possible, but I don't own the tools for that. So I don't know. I'm an optics guy. OK, yeah, back again to the optics. I completely understand, yeah. All right, let me see if there's anything popping up question-wise. We had a lot already. I think there was very, very much feedback at the moment. Don't see any new questions popping up. I thank you again. I apologize for the mega delay we had. Take off the glitches, but we made it. I'm happy about that. And thanks again. And let me quickly announce now, in a couple of seconds, we're going to switch over to the Harold News Show over there. Oh, OK. Yes.