How come companies like TSMC can't build their own machines? Is there really only one company in the world that has the required skills? Or is it only because of patents and exclusivity agreements?
https://en.wikipedia.org/wiki/Extreme_ultraviolet_lithograph...
I know of multiple suppliers down the supply chain which had to invest billions and had to wait 10 years before they even saw a penny.
There is a reason why everybody else essentially gave up.
ASML machines are truly a marvel of engineering. And duct tape btw, but also engineering.
There’s no fundamental reason other companies couldn’t compete. Simply invest tens, maybe hundreds of thousands of man-years and there you go, a second ASML.
ASML machines consist of at least two main parts: laser and lenses. Both has been engineered by third-party firms: TRUMPF (laser) and Carl Zeiss (lenses).
> Simply invest tens, maybe hundreds of thousands of man-years and there you go, a second ASML.
As it is often, you'd just need to know where to buy your start (remember MS DOS) and then you can invest a lot from your first revenues.
Canon might come up with a good different solution (13.10.2023):
- Canon, [], launches ASML challenge with machine to make the most advanced chips https://news.ycombinator.com/item?id=37870372
- Canon Nanoimprint lithography semiconductor manufacturing system (5nm) https://news.ycombinator.com/item?id=37894925
- Canon Prepares Nanoimprint Lithography Tool to Challenge EUV Scanners https://news.ycombinator.com/item?id=37917587
Correction: not lasers and lenses, but EUV light-source and mirrors.
The TRUMPF laser in the EUV machines is nothing too special, but the Cymer EUV light source, consisting of blasting tin microdroplets at the right angles at speed, with said lasers, is the secret sauce.
And mirrors are used instead, because lenses would absorb the EUV radiation.
But the ASML machines are substantially more than just some purchased parts put together. There’s a lot of unique engineering in bits that you can’t easily point at as a “part”, such as the entire metrology software - that’s the bit that measures how badly positioned and how crooked the wafer is and adjusts the laser and all stage positioning etc accordingly. You can’t buy stuff like that off the shelf, it’s all developed in-house. And they have a whole bunch of things like that, which are like part physics, part math, part mechanics, part software and part choosing/speccing (and sometimes designing) the exact right parts.
Canon’s new effort looks promising indeed btw, would be great if ASML gets some serious competition one day.
A fair case can be made that the twinscan concept is the key.
https://www.asml.com/en/news/press-releases/2012/asml-to-acq...
Like we’re in a book and the author needs a technobabble device that is made by just one organization, but incongruously powers entire industries that are foundational to the whole economy, but obviously coming up with an actual description of the device is a bit beyond the ability of the author. So they just call it “the companyname machine” and let the reader come up with details. “Oh it uses mirrors and lenses and concentrates light in a special way that nobody can replicate.” Reality is a hack author.
They toiled for 20 years without guarantee of success. They managed to pull it off and now they own the bleeding edge process and will do for the foreseeable future. If someone wants to do the same, they can become competitors.
It's not just a question of patents, they keep almost everything about these machines secret.
Another way to look at it, China would love to be able to copy their processes and has not yet been able to do so.
Edit: It looks like the Carl Zeiss subsidiary that does the semiconductor manufacturing tech is part owned by ASML:
https://en.wikipedia.org/wiki/Carl_Zeiss_SMT
Edit2: Looks like ASML invested a billion Euros into Carl Zeiss SMT:
https://www.asml.com/en/news/press-releases/2016/zeiss-and-a...
On one hand it’s anti-competitive because it makes it more difficult for the customer’s competitors to use leading edge technologies; on the other those leading edge technologies would be much harder and take longer to develop without the customer’s contribution.
And how is it symbiotic if Zeiss can't take other customers?
Guaranteed revenue that doesn't need sales/marketing to keep going.
A bunch, like more than a million dollars
So: they depend on each other, and they get more business and make more money by getting closer and have incentives to help the other improve (so it is not an adversarial relationship). Hence, symbiosis.
And in the imaging and litho world surprising in the level of trust. For example, Sony actually makes a number of sensors for Nikon and other camera makers, despite having often competing products.
Good news is that China has delivered their 8mm lithography machines and are promising 3mm sometime in the next two years. maybe this will introduce more competion down the road
Wait what? I've soldered SMD parts smaller than that, I guess I could do 3mm lithography free-hand with a good razor blade.
Sounds like I'd have a nice head-start in the "cutting-edge" field of sub-centimeter lithography.
I have my doubts about this.
SMIC is still struggling with 7nm, unable to deliver at scale. 3nm in two years sounds like a leap.
Do you mean 28nm?
https://www.tomshardware.com/news/chinas-first-28nm-capable-...
They’re struggling already just to copy the previous generation DUV machines, and that’s while having a bunch of machines they can study and copy.
A lot of technology you can’t really just copy it. You need the whole supply chain and expert support around it. Like with EUV they’d need someone like Zeiss to make the mirrors. Even the parts holding the mirrors are extremely niche and extremely hard to make.
SMIC could copy older TSMC tech because they could buy the same machines and poach engineers from TSMC. That’s just not viable for EUV.
Most of the West has bailed out of the electronics race so long ago, that the experience gap in between practice, and what is told in universities is like 20-30 years now.
No university in the world will teach you how to make a modern stepper. Most of research in this space been going behind closed doors from the beginning of the semiconductor industry, and nearly completely closed since around 2010.
Only places like IMEC and few others are semi-public windows to what happens behind closed doors.
In principle, physics behind the operation of a stepper is nothing secret, components not secret, manufacturing techniques not secret, but the experience of how to arrange all of above in a way it is known to work is only available to, I would say, less than 100 people worldwide.
Why should they be able to?
>Is there really only one company in the world that has the required skills?
Yes, they invested for decades into R&D for EUV lithography.
>Or is it only because of patents and exclusivity agreements?
No, they have a multi billion dollar and decade long lead.
It makes much more sense to try and pour money into a next generation manufacturing technology and work on it. Easier said then done.
The question should now be how fast can innovation happen with 2 or maybe even 3 supply lines instead of 1 global supply line for semi conductors.