ASML tips roadmap for EUV litho production platform

[Idontcare]

I can't see how to fix the problem you're talking about, that you need to employ two sets of people to do the work of one just so there's 'competition' and the cash is actually directed to R&D by those companies.

The optimal way would just be to have the first team's engineers do the work required. Then the world itself would be better off by a large amount from the saved resources.

I'm not trying to be glib, and your post is terse enough that it is left open to interpretation, but are you basically saying "why don't people motivate themselves to get the job done without operating in an environment of competition?"

(cuz if you are then yeah, that doesn't work, see USSR circa 1985)

You really do need competition to motivate people to be inspired to innovate and create. There's no shortcuts to human nature.

And the model isn't broken per se, look no farther than the history of Moore's Law for proof, but it just goes towards the conundrum posed by industrial collaboration in R&D.

It is very much a cold war in the industry between the varying tiers of the supply chain.

 

[Martimus]

It's a footprint issue. If you just want higher aggregate throughput you buy more tools and run them in parallel. A typical fab will have 10 or so 193nm scanners for example.

But the challenge is the square acreage allocated to multiple EUV steppers, in addition to the cost of buying multiple steppers.

There is a also a conflict of interest in the equation. The customer wants to buy as few tools as possible, and buying fewer tools is made possible by having the throughput of the tools themselves become higher.

Q: Who is going to make the throughput higher? A: The very same company that wants to sell them, lots of them, the more of them they can sell the better.

So it is not as simple as "please triple the throughput so I only have to buy one of them instead of three of them" because the supplier is going to say "well my revenue model is unchanged, so if you only buy one tool then I need to sell it to you for 3x as much".

The motivation for the tool supplier to bend a little on price is the threat/fear of losing the sale to competition...which is why GloFo and Intel are keen to keep Nikon in the EUV race to balance ASML's obvious need to do what is in the best interests of their shareholders.

I've worked on all sides of that dynamic business relationship and it can be interesting from many perspectives, but loads of fun. (until the jdp comes to an end, then it ain't so much fun as people usually get laid off at whatever company did not get the contract )

The issue is that current tools run at a rate of around 5 wafers per hour, and to meet the 50 wafers per hour goal they had set they would need to boost the power of the laser from 11W to 100W. At least that is what the article is claiming.

The problem with boosting the laser that much is that it greatly increases the downtime of the laser between masks.

I was asking if they could add lasers to mitigate this, even design a quick change. The tooling already costs $120M, so adding an additional $20M to the cost with these additional components might make the whole setup worthwhile.

As it is, it isn't really worthwhile at all at a rate of 5 wafers per hour, and I doubt there will be many costumers at all at that rate. It would likley be in ASML's best interest to increase throughput on their tools, since the current throughput is nowhere near cost effective enough for customers to consider purchasing. (Or at least that is what the article was trying to convey. I am not an expert or even experienced in this work so I cannot verify this claim.)

 

[gorobei]

clearly we are reaching the limits of hard science, time for us to follow tony stark's iron manual and start making metal eating bacteria stuffed with InGaAs that die and deposit the semiconductor materials on the relevant traces.

at what point are we going to need nano machines to start making our ic?

 

[khon]

Would it be feasible to add additional lasers to the tool, and run them at higher power to minimize downtime? (In other words, have 3 lasers, with each one running at 100W for 1/3 of the time, while the other two are off and cooling down). I could see that being difficult to keep alignments proper, but it is the simplest solution to the problem I can think of right now.

Short answer: No.

Even if the laser could run at 100W for a short time, there is no way you would be able to swap back and forth while keeping it sufficiently calibrated.