698K views · Dec 26, 2021 · 17:20 · 11 HN recommenders · 20 mentions · discussed 2022–2025
Asianometry explains the EUV light source inside ASML's lithography machines, where lasers hit falling tin droplets twice to produce 13.5 nm light. It walks through the engineering of the droplet generator, the pre-pulse and main laser, and the plasma that results.
Why Hacker News points people here: Commenters repeatedly point to it whenever EUV, ASML or chip manufacturing comes up, as a clear and astonishing explanation of one of the hardest engineering problems in chipmaking.
“This video from Asianometry on EUV light sources is well worth the watch. It's pretty incredible.”HN comment, 2022
“watch this about the light source, its completely bonkers”
Shows the bonkers light source of the most complex machines.
Because the EUV machines that ASML build are the most complex machines ever built by humans and getting all the bits to work properly takes decades of R+D e.g. watch this about the light source, its completely bonkers this video
“This video from Asianometry on EUV light sources is well worth the watch. It's pretty incredible.”
Incredible account of EUV light source engineering.
EUV lithography is interesting. You drop a microscopic droplet of tin. You then aim and shoot it not once, but twice with a laser so it produces the correct wavelength. Then you do this with 100% reliability thousands of times per second all day, every day for years. The problem doing this with fusion seems to be money. Their "big" grant of 600M wouldn't even buy a handful of those machines and wouldn't come close to the many billions it has cost in the past 30+ years of research to reach the point where they are finally viable (it had been "a few months away" ever since the early 2000s). This video from Asianometry on EUV light sources is well worth the watch. It's pretty incredible. this video
“Anyone interested in the details should watch this video”
Details of the two-shot laser needed to emit EUV.
Anyone interested in the details should watch this video: this video (has the same graphics) without any hyperbole the 2-shot laser thing necessary to emit EUV is just mind bending.
“Two must watch videos about the whole process”
Two must-watch videos about the whole chipmaking process.
Two must watch videos about the whole process: this video link (Typed on a device made by this process, which I still don’t quite believe.)
“here is a great video on some of the technical/physical challenges ASML had to solve for their EUV source”
Technical and physical challenges ASML solved for its EUV source.
To add to this, here is a great video on some of the technical/physical challenges ASML had to solve for their EUV source: this video
“Can recommend [1] as the title say about the light source. Very interesting even for a non professional”
Very interesting even for non-professionals; about the EUV light source.
Cant answer the question and my knowledge is only from 'one interested' but from how long (~20 years) it took to build the current EUV source and how complicated the alternatives look this is for sure no easy feat. Can recommend [1] as the title say about the light source. Very interesting even for a non professional (and [2] as on high level overview of the machine and [3] as longer documentation inside ASML). [1] this video "The Extreme Engineering of ASML’s EUV Light Source" [2] link "The Extreme Physics Pushing Moore’s Law to the Next Level" [3] link "ASML's Secret: An exclusive view from inside the global semiconductor giant"
“Asianometry on YouTube is a good source for this kind of topic.”
Asianometry is a good source on chip manufacturing topics.
The machines are made by other companies. Lithography is the most expensive now and its made by ASML in Netherlands is the only solution for EUV lithography. But there are many steps in manufacturing and a machine for each. The major equipment manufactures are spread across US, Europe and Japan. In the US the three big equipment manufacturers are Applied Materials and Lam Research and KLA. The costs are probably driven by extreme engineering requirements (temperature, pressure, corrosive chemicals, low contamination) and R&D costs to meet these requirements. The production volumes are not high either. Its not like they are building millions of these machines. Asianometry on YouTube is a good source for this kind of topic. this video
“is pretty insane, too. See”
Shows the insane tin droplet laser process for EUV.
The etching process that uses a laser to blast a moving molten tin sphere to shape it so the next laser blast that vaporizes it produces parallel EUV light is pretty insane, too. See this video
“Shown better in a video”
Visually shows the droplet shaping and plasma generation.
The article doesn't go into much detail on this, but the UV source using the hydrogen here is incredibly cool. To get the 13.5nm light they need, they use a 20kW CO2 laser to explode tin droplets into plasma that emits the correct wavelength. 50,000 droplets per second. And they use a smaller laser to first shape the droplet into a little dish shape that explodes better. Shown better in a video - this video
“Skip to 10m50s.”
Explains the double laser hit on the tin droplet.
Twice. They have to hit the molten droplet twice . Once "gently" to deform it briefly into a concave shape, and the second, harder pulse to actually activate the droplet to emit extreme ultraviolet light Asianometry on EUV. Skip to 10m50s. this video
“Here's a good video”
Good video on vaporizing tin droplets with two laser pulses.
There's a lot of expertise with lithography around the world too (companies like Canon build lithography machines), but we're not talking the usual levels of it. The EUV lasers ASML/TRUMPF builds don't even work the same way other "conventional" lasers would[1]. You physically can't get there by incrementally improving some existing process. Now I don't know how ZEISS makes mirrors that blown up to the size of a country would have imperfections smaller than a human hair, but I'm pretty sure it's no small feat either. These companies invested decades and untold sums into this when few other companies even had incentive to attempt it themselves. Sure, other companies could eventually replace them, but you're not closing a 2-decade technology gap in an afternoon. [1] It's a bit insane, really. Vaporizing falling droplets of tin with two laser pulses 100,000 times a second to get just the right wavelength? Here's a good video: this video
“Knowing that you fire microdroplets of tin with a specific shape into a chamber at 50khz and hit it with two laser pulses is one thing”
Explains how EUV tin droplets and lasers work.
Knowing that you fire microdroplets of tin with a specific shape into a chamber at 50khz and hit it with two laser pulses is one thing, doing it is an entirely different one. How EUV works: this video
Link seemingly should be to gwern's comment? link > TSMC fabbing only the start (!), now all ASML customer support† is gone (!!) not just EUV machines, on top of which all the American-citizen employees (a nontrivial fraction due to education/birth abroad) have halted work literally overnight (!!!), and overall what sounds like the collapse of Chinese semiconductors amidst a Chinese economy already showing serious signs of distress and little capacity to keep an industry on life-support indefinitely as they scramble to survive - It keeps going & going ... !!!! I am not so doom & gloom, largely because I see China as literally the only nation in the world that has the willpower, top down comprehension/drive, and scale to sometimes buckle down & drive huge vast tasks that are necessary, and computing is obviously a top necessity. They have very nearby friends(-ish) facing very new sanctions who are having lots of problems having been cut off already, which is a very modern development & hugely emphasizes the criticalness. But they are not stupid and already understand this existential risk enormously. (And they understand that their political honor-spat for geo-control of nearby areas will probably not ever secure this asset- close though it may be- but only destabilize their access, necessitating internal development.) It's a huge challenge for China: the challenges are huge, there's so few friends, there's tons of other economic peril afoot already. It's hard enough. And co
“All of the problems you just described are solved for in EUV lightsources for lithography on chips at 100khz or more.”
Cited to argue droplet light source problems are already solved.
All of the problems you just described are solved for in EUV lightsources for lithography on chips at 100khz or more. These are less hard then you make them sound. this video Probably the hardest part is making sure the droplet is cost effective enough that we care. Again, to op's point, this is an incredibly shallow analysis. The question I would be pushing towards is: What are the hardest remaining engineering problems? How likely are we to overcome them? At the end of that process will it be a cost competitive outcome?
“ASML's crown jewels are not mere "tools", they are machines .”
Used to illustrate ASML machines are bleeding-edge, not mere tools.
ASML's crown jewels are not mere "tools", they are machines . It is bleeding edge technology and entirely irrelevant to a general discussion of economics of 'tool makers'. this video
“This Asianometry talks about the engineering”
Covers engineering of the EUV light source.
There were also issues getting the light sufficiently collated from a synchrotron. This Asianometry talks about the engineering: this video Now, China may decide that it doesn't need throughput and that will definitely help. Having a facility that can simply do it even if it can't do it at speeds that a commercial fab would accept would be fine for strategic reasons. However, it will be 20+ years of engineering. And that's just the light source . The optics, for example, are a whole other engineering project.
“Current EUV light source requires a 20kw carbon laser.”
Linked as a source on current EUV light source requirements.
The graphic from the OIST in the link makes reference to both the reduced power needed and reduced capital costs, less mirrors & cheaper EUV light source. Also, less power -> less heat -> less cooling. I scoured for some other info as well: A single chip would go through a photo-lithography tool about 80 times. A photo-lithography tool can work on approx. 125 wafers per hour (5 FOUPS). A wafer contains ~230 chips. Current EUV light source requires a 20kw carbon laser. link this video
“There are actually two lasers; the first one to reshape the spherical tin droplet to concave shape”
Cited for the two-laser tin droplet process.
> the laser tin-zapping thing ASML uses. The Science/Engineering to make this work is nothing short of magic! There are actually two lasers; the first one to reshape the spherical tin droplet to concave shape and the second one to vaporize it to plasma generating EUV range light. The Extreme Engineering of ASML’s EUV Light Source by Asianometry - this video
“I didn't realize that the droplets are hit twice by different lasers.”
Revealed that droplets are hit twice by different lasers.
I just found this one. link Edit: I didn't realize that the droplets are hit twice by different lasers. this video
“Video about the light souce”
Video about the EUV light source for ASML machines.
ASML is so interesting. Its like sci-fi that one firm in the Netherlands knows how to make the most complicated machine ever made, and no one else can do it. And it is arguably the most complicated machine ever made. 50,000 times a second, the EUV lithography machine hits a 25 micron drop of molten tin that is moving at 70 meters per second with two co-ordinated lasers, the first hit to change the shape of the drop of tin in exactly the right way, the second hit to vaporise it, creating Extreme Ultraviolet Light at the right wavelength to etch chip designs onto silicon at "5nm process" sizes. Some labs can cobble together something similar as a proof of concept, but not well enough to make it feasible for mass production of chips. Video about the light souce - this video No one else in the world is able to make these machines. If you buy one it costs $150m and gets shipped to you in forty containers on specially adapted planes. Very few firms have the resources/know how to even run the machines - which is what makes TSMC so important.
