The main idea of making masks and making transistors smaller came about because this guy, I think he was working for TI, came about and he was looking at a microscope and he was like, wait a minute, if I can flip the microscope over and shine light from the other side, it gets smaller.
Hello everyone and welcome to another Semi Doped podcast. I'm Austin Lyons with Chipstrat, and with me is Vik Sekar from Vik's newsletter. Hey Vik, what's going on, man?
Yeah, not that much other than like everybody's panicked about the 13F from that Situational Awareness hedge fund, and there was this selloff in optics and everybody freaked out. I'm like, what's going on? First of all, I still don't really know what a 13F is. I guess it's like a hedge fund has to report some holdings that, you know, they hold and what they sell and what they bought. I think something like that.
And Leopold Aschenbrenner, is that how you pronounce his name? Sorry if I got that wrong. I have no idea. It's too many syllables. Yeah, I mean, he, I guess he made a lot of money and converted like a few million into a billion or something. And then now everybody's following him for the latest stock tips, and when everybody looks at his 13F, they panic and there's this whole selloff. Apparently he shorted optics and shorted AMD, Intel, and he sold in, I don't know.
So it's like everything is down in the semi market because of some hedge fund thingy. This is what's happening. It's funny. I mean, so if I'm him, I'd wake up and look and go, "Wow, I can buy a few puts because on these 13F, it doesn't talk, I don't think it talked about like the size of his puts or anything." So he can just buy a few puts at some point in time, drive the price down. Could he use that to then buy the price? I mean, that's market manipulation, but on the other hand, it's like people are trading on public data and on vibes, you know? So just like, and he could throw people off his trail, right? Like just buy a tiny amount of puts in like everything and you wouldn't know what he's investing in.
Yeah. I don't know. I think Leopold listens to this podcast. If you do, just let us know. Maybe you can come explain to us what a 13F is because I don't know about you, man. I have no idea what all this stuff is. So I'm happy to learn from the best.
Yes, Leopold. You're welcome anytime. All right. So today we're going to talk lithography. So I thought it would be really interesting to talk about the economic challenges of lithography, modern EUV lithography especially, because, you know, ultimately incentives drive outcomes, and there are challenges with the increasing cost of lithography, the increasing costs of fabs, and you start to see, you know, TSMC can afford the next, you know, process node, and Intel and Samsung are trying to stay in the race or be in the race. There aren't that many other competitors, and from afar, if you're like a semi-tourist, as they like to say on X, you might say like, hey, TSMC is crushing it, you know, why aren't more people in this game? And of course, as you start to unpack it, you realize that costs are a barrier to let people enter, and so then you start pulling on the thread and you ask, well, what cost? And one of those major contributors to the capex needed to participate is modern EUV. And so, yeah, what do you think? Should we talk EUV?
Yeah, that's a great topic for today. I think it ties in tangentially to what is happening to the market today and why people think that we are not in an unconstrained trajectory upwards. And this basically stems from the recent Gavin Baker interview that I put in one of our daily Semi Doped daily updates on Substack. So if anybody is not subscribed to the Substack, I recommend it. It's free. You know, you can go get subscribed on that semidope.com. But the idea that Gavin said, you know, floated in that interview was we are not in a bubble because TSMC is basically holding back the entire industry by not creating enough chips for everybody, by controlling their level of capex spend on tools that require, you know, DUV and EUV. And most famously, TSMC is not pro EUV. They think the tools are too expensive and that they want to stay on DUV with multi-patterning as far as they can manage to do so. And we'll talk about what DUV with multi-patterning means in this episode. But this is where we are right now. Tools cost a lot. I think an EUV machine costs something to the tune of $400 million. And that's just one machine, and you need to run many of them to produce the chips at scale that we need, and that scale is continuously increasing. So it really comes down to is EUV the only way forward? And then there is Hyper-NA EUV, which is like extremely, I don't even know if those are in production yet. I don't think so, but those are going to cost close to a billion dollars. That's insane.
Yes. Totally. So we, yes, we'll unpack all this for everyone. When you're like, "What's DUV? What's EUV? Why does it cost so much?" That's what this episode's all about. Lithography Masterclass, hopefully to set the stage for hopefully even future conversations around lithography, some startups that are out there. And actually, you know, I just saw ASML shared a roadmap. I think it was IMEC had a conference. It's going on like right now, the ITF World conference. And ASML showed some of their roadmap and how they think they're going to be able to bring the cost per exposure down. And it would be nice to unpack this, but before we get to all of those nitty-gritty details, we want to educate our listeners on some of the fundamentals. So let me throw out some numbers. So I want to, one thing I want to set one thing for Austin before Austin. So most of this stuff, Austin is actually the expert because he actually spent time working on this kind of lithography stuff in a clean room, which is something I can't say I have done. I've measured wafers and measured transistors, that kind of stuff. So I've handled completed silicon wafers, but I've never actually done lithography myself, and you have. So this is a cool thing. You can tell us a little bit more rather than just what you can read on the internet perhaps. It's interesting what it looks like inside a clean room. I'm excited to hear.
Yes. Yes. So, well, thank you for mentioning that. So, yes, when I was in grad school, I worked at the University of Illinois Urbana-Champaign doing research as a research assistant making graphene-based transistors. So I was in a lab with Professor Eric Pop. He's at Stanford now and he studies 2D materials. So we were studying, our people on our team were studying carbon nanotubes, and then I got to work in graphene. And so, yes, I got to be in a clean room with, you know, silicon wafers, etching, you know, depositing, doing lithography. And in fact, I even got to do e-beam lithography because we were trying to make, you know, very precise kind of one-off little nanoelectronic systems. And it didn't matter that the throughput of e-beam is really low, which we're not going to talk about e-beam lithography much here, but it's just a cool thing that I got to experience. And in fact, when I first started, we were mechanically exfoliating graphene, which means we were taking basically tape. And mechanically, we'd get graphite, like layers of pencil lead essentially, and then we'd take like literally like clear tape and then stick it on top of the graphite and then pull it off and then take it under a microscope and look, and you can kind of like figure out how many layers of graphene you have there. And so I just scan around, you know, I'm like doing some of this stuff in a bunny suit and then going back into the lab and moving this thing around, you know, and just thinking like, oh, what am I doing with my life? I'm trying to find single-layer graphene. That's what I'm doing. But of course, once you find single-layer graphene, then we would make trans, we pattern the transistors right there on it and then take measurements and publish it because it was a hard thing to do. And so we kind of had an advantage that we knew how to do it. I won't talk too much more. Or we eventually did chemical vapor deposition to create graphene using like copper foil essentially, and then we could grow graphene, and it wasn't as pristine and pure, but it was more scalable. And anyway, good memories. I spent a lot of time.
Yeah, amazing stuff. I bet it was a great experience doing all this stuff.
At the time, it was probably frustrating as all hell, as all research is, but then looking back at it, it's like, ah, I don't mind doing that again now. At least like going into a clean room and like pottering around would be nice.
Totally, man. It was like, it was challenging in that you had to, you couldn't just do the research, you had to do a lot of exploration to figure out how to even make the thing before then you could actually measure it and do the research. And so, yes, it was frustrating at the time because you'd spend all this time, a whole day, and then you get to the end and be like, "Oh, that didn't work." You know, but yes, looking back now, it's like, oh, how fun was that? How much intellectual freedom there to just be like, you know, your research professor says, "Hey, go make this thing." And you're like, "Well, I guess I got to read and experiment and try to figure out how to make it and then measure it and then see if our hypothesis was correct," you know.
Yes. Yes. Yeah. It's nice. Okay, now we're good. Now, you know, we hit carbon nanotubes already. Like research that's never made it out of the lab. You spent time doing that. Awesome. Now, let's talk about stuff we can actually make.
Yes. Okay. So, let's talk about... Okay. So, have you heard of Rock's law, by the way?
Okay. So, this is a, you know, a quote-unquote law just like Moore's law. It's really an observation, named after Arthur Rock, and he was an early investor in Intel, maybe, kind of one of the founders of the venture capital industry, if you will. But the observation was that the cost of a semiconductor fab doubles every four years. So, you know, it gets more and more expensive. So now this is interesting. This is related to Moore's law, where Moore's law said that the number of transistors in an integrated circuit doubles every two years. And so if the cost of a fab doubles every four years, that's slower than the number of transistors doubling every two years. And so what that means is that Moore's law is really like an economic statement that says for roughly the same price we can get more transistors per die area, per chip, per unit area over time. And that's really what drove the industry, was that economic realization that for the same dollars you could get more compute over time. We are getting to the point where the cost of fabs is getting so high that, you know, we've seen decades ago really Moore's law, the number of transistors doubling shrinking for physical reasons, but also from an economic perspective actually the cost of transistors flatlining and potentially even coming up, the cost per transistor. And then again the question is what's driving that fundamental change to some of these important economic scaling laws that we've seen for decades, and lithography is a big contributor. Like you said, low-NA EUV tools used to cost around $250 million, now high-NA is coming out around $400 million, and give or take, you know, these numbers change over time and it probably depends per customer, but you can back some of this stuff up out of ASML's reports because they only sell, you know, they don't sell many of these per year. But anyway, yes, the rumored Hyper-NA, which would be a future tool, could cost anywhere from $600 to $800 million potentially. If the Strait of Hormuz stays closed, maybe a billion dollars because that's driving everything up.
So, which means lithography, all of these tools, like you said, and we'll get into it, you know, you might need 15 tools to open up a new fab. So I think I'd seen some CNBC coverage where it said that Intel's 18A fab in Arizona, Fab 52, needed 15 EUV machines. So imagine that, you know, half a billion dollars a pop and you need 15 of them. That's no joke. And that's why a brand new fab costs on the order of 20, 30 billion. So I just wanted to quick paint the picture of the enormous capex cost to just build one new fab to try to stay in this race. And of course that ultimately means the cost per wafer will have to go up too because you only can do so many wafer starts per month, say tens of thousands of wafer starts per month or a hundred thousand, and you need to amortize the cost of all those tools over that fixed number of wafers.
Okay, cool. So let me back up a little bit and, you know, quickly frame it in a way that, you know, I usually process things. So one of the biggest problems of modern lithography is cost, and that primarily stems from having to make smaller and smaller transistors. And we have gone from deep ultraviolet, which is what we'll refer to as DUV, to eventually extreme ultraviolet, which is what EUV stands for. And then we have within EUV, we have several levels of numerical aperture. I think we should define what that is, you know, going forward next so that we actually have all these basic terms in place. So we have low numerical aperture or low-NA EUV and high-NA EUV, and then Hyper-NA EUV. And the higher you go from low to high to Hyper, you can make smaller and smaller transistors. Right.
So basically, higher the numerical aperture, the smaller the transistors, the smaller the feature sizes you can make. And now what you need is regardless of whether you choose a deep ultraviolet or an extreme ultraviolet machine, you're going to have to need like at least 10 or 20 of them in a fab. And if each of them costs like, I don't know, half a billion dollars and you want to put in 10 of them, you're spending $5 billion in just these EUV machines. And then not only that, you have to put a whole lot of other infrastructure like cooling, like clean rooms are difficult because you have to have the vac, the HVAC systems that pulls out all of the dust particles from the air. And so based on how many dust particles you can find per unit volume, these clean rooms have different classifications, right? So you've got class 1, class 10, class 100. And I think the higher the number, you got more particles per cubic, you know, volume of air. So all of this stuff takes like an enormous amount of money and time to build. And if you actually, this is like another point, if you look at like TSMC's construction of their fabs, it's actually critical machines are suspended on pistons, like entire factory floors are suspended on pistons so that it's immune to like earthquakes and stuff. So it is very expensive to build a fab and takes years. It takes years. It could take three to five years, you know. So this is why we can't simply add chip capacity willy-nilly, right?
Totally. Totally. And not only that, if you haven't seen a fab, I'd encourage people to figure out how to get on a fab tour if they can somehow. I know it's very difficult, but...
I want to go. I've never been in one.
Vik wants to go. I actually had the good fortune recently of getting to go to Intel's Fab 52 and tour it. But a fab, you also need a... So not only do you not want the machines to move around even a tiny bit from earthquakes, but even from like passing traffic and stuff because of course we're making transistors on like the atomic scale, nanometer scale. And so you just, you don't want all that, any sort of mechanical wiggling and movement. But you also, these EUV machines, they have like big power sources and those to the, in the light source, like a lot of that actually goes in the subfab floor. So there's a floor beneath like the main floor where the tool sits and they have all this equipment underneath it. And then of course, you know, there's like a floor above it and you've got, you got to flow all that air through. And so I just also wanted to illustrate that it's not just like, like data centers where you're just like, "Oh, find some real estate, slap a building up, throw some racks in, and you're good." But yes, it takes, even from a construction and HVAC perspective, building a fab is no joke. And it goes beyond that because if you look at Intel fabs and the way they've built it in the past, they had this copy-exactly method, which means they copy exactly. This is not something that they mess around with. They use like similar plumbing. I've heard that they even use the same brand of paint because they do not want anything to go wrong. Because...
If small things happen and change the way a fab functions, you can't get the yield up. And if you can't get the yield up after spending $20 billion, you can't make enough wafers, which means you can't sell them and make a profit. So Intel decided to copy exactly. And that actually slowed a lot of stuff down for them. That's a different story. But really, that's how difficult it is to build a fab. Really? I mean, that's insane.
Totally. Totally. So, all right, going back, you mentioned DUV and EUV. And so let's tell listeners a little bit. So back in the DUV days, the light source that was used had a wavelength that eventually made its way to 193 nanometers. And I guess even zooming out even further. So lithography at the end of the day, for those who don't know, I think everyone probably does these days because ASML is an awesome company and everyone wants to invest or has invested and so at a high level understands what lithography is. But we're talking about ultimately being able to expose light to sort of, and I put in quotes, draw, you know, the shape of transistors or the shape of areas that you want to etch away that you will leave parts of the transistor but etch away other parts of the surface of the chip. And so ultimately to make transistors smaller and smaller, you can either make the wavelength of, ultimately you need to make the wavelength of light smaller and smaller. But there's also something which we can get into, which is the numerical aperture, the mirrors that you talked about making, changing the numerical aperture. But just focusing on making the wavelength of the light smaller. You know, the canonical example here is like writing with a Sharpie. You're going to draw like fat lines. And if you can write with a fine tip marker, a fine tip pen, you can ultimately make a lot thinner lines and you could draw smaller precision features. So that's what the industry was trying to go to from DUV, deep ultraviolet lithography, to EUV, which uses 13.5 nanometer light. So ultimately you're going an order of magnitude smaller from the fat marker down to the fine tip pen.
Yeah. So this whole relationship that you mentioned here, where you want a smaller wavelength of light but you want a higher numerical aperture, this is governed by what is known as the Rayleigh criterion, which means that the smallest dimension you can make on a wafer is literally proportional to the, you know, wavelength but inversely proportional to the numerical aperture. There is also a constant factor here that's often called K1, which we won't get into here, but think of it as another knob which you can use by designing the masks that, you know, selectively allow light or don't allow light in regions. You know, they do all kinds of tricks on those masks to improve this proportionality factor K1. We won't get into it, but these are the factors. So there are some bunch of tricks. Then there's the wavelength and then there's a numerical aperture. So the smaller the wavelength you go, the better. And to Austin's point here, deep ultraviolet lithography was most famously, you know, ended at what is called argon fluoride lithography. Is that right?
Mm-hmm. At 193 nanometers, and then there was a, like, a quantum leap down to 13.5 nanometers with EUV, with extreme ultraviolet. So that's a big change. That's like more than a 10x change, right? And going down another 10x hasn't happened yet, but we will get to how that can happen at the end of this episode. But yes, so keep going. Let's go with...
Let's talk about DUV for a second. So do you want to explain, like, where the 193 nanometer light comes from with argon fluoride?
Yeah, there was a whole evolution to that as well. It's not like we just landed up there, right? You know, when we started lithography, like in the 1980s, it was mostly like what was called i-line lithography, which had a wavelength of like 365 nanometers. Then over the years, people realized, like, wait, we've got to make this better. And then they came up with, you know, krypton fluoride lithography, KrF lithography, that went to, you know, 248 nanometers. So just by changing the kind of light source that you're shining through and the wavelength of the light source, you could get better features. So this was like going through the '90s, you could have like 248 nanometers that evolved to like argon fluoride lithography where they went to, you know, 193 nanometer. And that was pretty cool, but then that lasted all the way through the 2000s, let's say, and kind of they kind of ran out of light sources. They did try some other light sources along the way, but they didn't really, like, work out for various reasons. And then they were kind of stuck with, like, argon fluoride for a while, but then, like, they thought about it, like, how do we improve numerical aperture? Somehow we have to improve it. And the answer was extreme... Actually, it's amazing if you come to think of the history of lithography, it's insane. Some smart guy came up with the idea and said, how about we put water on the wafer? Like, let's just put water on it. Like, what, you're going to put water? So, yeah, that's literally what they did. They put extremely pure water on top of the mask and then put the light through the water onto the mask, and that came to be called, like, immersion lithography. So that actually helped scale transistors further by just, like, putting water on the wafers. It's, like, insane, right? So...
The history of lithography is amazing. I wanted to tell you the way it all started. I don't know if you know this, but the way it all started was in the early days. I forget the name of this guy, but, you know, look out on Semi Doped. We'll have a whole post on this thing. The main idea of making masks and making transistors smaller came about because this guy, I think he was working for TI, came about and he was looking at a microscope and he was like, wait a minute, if I can flip the microscope over and shine light from the other side, it gets smaller, right? Everybody knows you look the wrong way at these things, like, stuff gets smaller. I'm like, this is it. Like, I'm going to turn the microscope upside down and shine light through the wrong end and everything gets smaller. And that's how all of lithography came about, by turning the optics in lithography came about because this one guy had the idea to turn the microscope upside down. So that's how it all started, right? And then we've been continuously going down the path of these various laser materials down to putting immersion lithography with water and ultimately coming down to, like, extreme ultraviolet lithography, which is an engineering feat that is an achievement for, like, humankind. That's how big it is. We'll talk about it too, but, yeah.
