4 ms·

China Aims to Use Particle Accelerators to Build Chips and Evade EUV Sanctions

HalloDan 3y ago
Like this? https://www.youtube.com/watch?v=0igQuerc3J0 https://www.youtube.com/watch?v=0igQuerc3J0
- jschlesser 3y ago
  The article says its SSMB which is a hybrid of synchrotron and a linear free electron laser. https://www.quora.com/What-is-a-Synchrotron-SSMB-generated-EUV-light-source-and-how-soon-can-it-be-used-in-the-high-volume-manufacture-of-semiconductors https://www.quora.com/What-is-a-Synchrotron-SSMB-generated-E...
mikhailfranco 3y ago
Is this something that could be researched with the UK's 20-year old Diamond Light synchroton facility? https://en.m.wikipedia.org/wiki/Diamond_Light_Source https://en.m.wikipedia.org/wiki/Diamond_Light_Source I notice a major £500m upgrade was recently announced: https://www.diamond.ac.uk/Home/News/LatestNews/2023/060923.html https://www.diamond.ac.uk/Home/News/LatestNews/2023/060923.h...
- bsder 3y ago
  Doubtful. ASML didn't choose to laser vaporize rapidly moving miniscule tin droplets because it was easy. It was simply easier than making a synchrotron work.
  - MichaelRazum 3y ago
    I think this is a bad argument. It's like saying only because someone clever choose path A it is wrong to choose path B.
    - rbanffy 3y ago
      Free electron lasers are not without their difficulties either, so I guess China assessed how well they could follow ASML and, given their difficulties, decided to learn from someone else's mistakes instead of making them themselves. They still might learn from their own mistakes, but at least they are trying something different. I'm not sure vaporizing tin-droplets can be improved enough to solve all of ASML's problems.
  - rbanffy 3y ago
    IIRC, they chose it because it's more compact and easier to install in a factory than a synchrotron, and also are a continuation of their current product evolution. I don't think they'd be able to make a synchrotron suitable for this that could be shipped in a standard container (or loaded onto a single truck).
    - bsder 3y ago
      There were also issues getting the light sufficiently collated from a synchrotron. This Asianometry talks about the engineering: https://www.youtube.com/watch?v=5Ge2RcvDlgw https://www.youtube.com/watch?v=5Ge2RcvDlgw 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.
  - dirtyid 3y ago
    IIRC EU/EUCLIDES exploration of EUV light source concluded synchrotron easier/cheaper to _technically_ operate long term, but exploration limited to 8W, compact designs. As far as I can tell, getting to higher power = scaling up size to 100s of m = made it logistically infeasible to build/plan fabs around. The Tsinghua SSMB design is 40m diameter, which is compact enough to be useful. Also ASML chose laser plasma because they thought synchrotron was too expensive to build, but years of technical challenges and a machine costs 200m, next gen will be 400m... that's already the price of multiple small synchrotrons. TLDR is laser plasma, works, is very expensive to operate, and ASML is probably locked in due to sunk cost.
ooterness 3y ago
This is just an alternate EUV light source. I don't think the light source is the hard part of building systems for 2 nm lithography.
- rbanffy 3y ago
  ASML does that by vaporizing tiny droplets of tin while they are in free-fall in a vacuum by hitting them with very precisely timed laser pulses. This light source took years to develop and ASML opted to do it this way because the development risks were lower considering their previous experience with similar less extreme UV light sources.
rbanffy 3y ago
What I really like about this is that it's a completely different approach from what the other big player is doing. Evolution in nature is based on small iterative variations to locate local maxima without ever regressing to a worse fit on its way to an even better maximum. Very few times we see someone doing something wild in hopes they'll end with something better in the end.
simne 3y ago
This remembers me USSR, when they hit by sanctions, they also try to create some extraordinary solution, centralized way, because of limited total resources, but better concentrated than in free market economy. Sometimes, centralized way could even get ahead decentralized free market way. Will see.

Like this? https://www.youtube.com/watch?v=0igQuerc3J0 https://www.youtube.com/watch?v=0igQuerc3J0

jschlesser 3y ago
The article says its SSMB which is a hybrid of synchrotron and a linear free electron laser. https://www.quora.com/What-is-a-Synchrotron-SSMB-generated-EUV-light-source-and-how-soon-can-it-be-used-in-the-high-volume-manufacture-of-semiconductors https://www.quora.com/What-is-a-Synchrotron-SSMB-generated-E...

Is this something that could be researched with the UK's 20-year old Diamond Light synchroton facility? https://en.m.wikipedia.org/wiki/Diamond_Light_Source https://en.m.wikipedia.org/wiki/Diamond_Light_Source I notice a major £500m upgrade was recently announced: https://www.diamond.ac.uk/Home/News/LatestNews/2023/060923.html https://www.diamond.ac.uk/Home/News/LatestNews/2023/060923.h...

bsder 3y ago
Doubtful. ASML didn't choose to laser vaporize rapidly moving miniscule tin droplets because it was easy. It was simply easier than making a synchrotron work.
- MichaelRazum 3y ago
  I think this is a bad argument. It's like saying only because someone clever choose path A it is wrong to choose path B.
  - rbanffy 3y ago
    Free electron lasers are not without their difficulties either, so I guess China assessed how well they could follow ASML and, given their difficulties, decided to learn from someone else's mistakes instead of making them themselves. They still might learn from their own mistakes, but at least they are trying something different. I'm not sure vaporizing tin-droplets can be improved enough to solve all of ASML's problems.
- rbanffy 3y ago
  IIRC, they chose it because it's more compact and easier to install in a factory than a synchrotron, and also are a continuation of their current product evolution. I don't think they'd be able to make a synchrotron suitable for this that could be shipped in a standard container (or loaded onto a single truck).
  - bsder 3y ago
    There were also issues getting the light sufficiently collated from a synchrotron. This Asianometry talks about the engineering: https://www.youtube.com/watch?v=5Ge2RcvDlgw https://www.youtube.com/watch?v=5Ge2RcvDlgw 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.
- dirtyid 3y ago
  IIRC EU/EUCLIDES exploration of EUV light source concluded synchrotron easier/cheaper to _technically_ operate long term, but exploration limited to 8W, compact designs. As far as I can tell, getting to higher power = scaling up size to 100s of m = made it logistically infeasible to build/plan fabs around. The Tsinghua SSMB design is 40m diameter, which is compact enough to be useful. Also ASML chose laser plasma because they thought synchrotron was too expensive to build, but years of technical challenges and a machine costs 200m, next gen will be 400m... that's already the price of multiple small synchrotrons. TLDR is laser plasma, works, is very expensive to operate, and ASML is probably locked in due to sunk cost.

ooterness 3y ago

This is just an alternate EUV light source. I don't think the light source is the hard part of building systems for 2 nm lithography.

rbanffy 3y ago
ASML does that by vaporizing tiny droplets of tin while they are in free-fall in a vacuum by hitting them with very precisely timed laser pulses. This light source took years to develop and ASML opted to do it this way because the development risks were lower considering their previous experience with similar less extreme UV light sources.

rbanffy 3y ago

What I really like about this is that it's a completely different approach from what the other big player is doing. Evolution in nature is based on small iterative variations to locate local maxima without ever regressing to a worse fit on its way to an even better maximum. Very few times we see someone doing something wild in hopes they'll end with something better in the end.

simne 3y ago

This remembers me USSR, when they hit by sanctions, they also try to create some extraordinary solution, centralized way, because of limited total resources, but better concentrated than in free market economy. Sometimes, centralized way could even get ahead decentralized free market way. Will see.