[EE|Times] Anayst: IBM "EUV Results Bogus"

[witeken]

Is 450mm suffering from the same kind of things, or was it simply not supported by many companies because they couldn't afford it anyway?

 

[jpiniero]

I'm convinced 450 mm is kind of dependent on EUV, one way or another.

 

[Roland00Address]

please understand that I am no expert in this field but several posters in this thread are

In the late 80's it became all the rage to spin-out your internal tool development division as a means to "unlock shareholder value". In other words the IDMs started the process of becoming less and less vertically integrated.

Adding to Idontcare's point about shareholder value. It is very expensive to create such tools, having a third party that everyone buys from, all the foundries allows allows you to pools some of the research dollars since you do not have to have 8 different ways of making the same wheel. In theory if Intel and such can keep their process lead it makes sense for Intel and other leading edge foundries to have a third party develop these tools since the tools are cheaper. It even makes sense for the trailing foundries for the tools are cheaper. In a way it is kinda sharing R&D dollars.

It can also be very dangerous and risky though if Intel can't keep its process lead with all the other tasks it takes to run a foundry. Intel makes money by having the fastest and latest foundry and translating that foundry in high prices server and consumer cpus. If this foundry advantage disappears because the competitors such as TMSC, what used to be AMD and is now GF, UMC, Samsung, etc may catch up and thus Intel can't make as much per cpu since cpus are partly sold on performance per dollar and performance per watt.

The trailing foundries make money by being cheap as chips to make the technology, that and specializing in custom designs and ease of use. That said if the big leading edge foundries open up or have excess capacity they may poach your customers and swallow up your market share.

Thus if part suppliers such as ASML fall behind or the other foundry engineers "catch up" with their fabs (catch up is a pejorative term) Intel will make less money. This is why Intel in 2012 spent 4.1 billion into AMSL. This was new capital for AMSL to play with and work on R&D and such and in exchange Intel got 15% of the company that is AMSL. Intel needs to remain the best fab and to be the best fab you need smaller geometries of transistors.

Time the market just right and you rake it in big, time the market only a few months off let alone a year plus off and you can lose it all. In some ways these companies and their investors are gambling with dollars like a bookie does on sports bets such as horse racing. It is very expensive to pay the talented engineers and pay the bills for the electronic revolution, those shareholders pay the power bill while the scientists and engineers create amazing things.

 

[Idontcare]

Is 450mm suffering from the same kind of things, or was it simply not supported by many companies because they couldn't afford it anyway?

450mm is a separate situation entirely.

The challenge with 450mm is no different than the challenge of every wafer size change that came before it - basically that you can't do anything until you have everything.

What that means is you have a real catch-22 situation with wafer transitions if you are the tool vendor. Be the first to produce a 450mm etch tool and you still can't sell them unless someone else is also selling a 450mm wet cleans tool, or 450mm litho tool, or 450mm deposition tool, etc. etc.

But the deposition company doesn't want to be first with a 450mm deposition tool, because no one will buy it until that 450mm etch tool is available, and so on.

So the challenge is really just one of coordinating roadmaps across tool suppliers (no small feat as each are beholden to different masters when it comes to R&D investments, margins, growth, etc.), combined with coordinating tool availability roadmaps with the customer's roadmaps in terms of when they might even be thinking of building another $14-20B fab.

But there aren't any what I would consider to be "technical barriers" to scaling all the existing wafer handling and process technologies that exist today at 300mm to a 450mm dimensionality. Unlike EUV which basically had no predecessor at any dimension, let alone 300mm or 450mm.

The barrier for EUV is nearly entirely technical, getting enough photons generated in a sustainable (high duty-cycle) manner so as to have throughput/meter^2 be high enough as to become commercially viable. Pour billions of dollars into that basic research all you like, isn't going to make the scientists have their brilliant breakthroughs any faster.

(analogies to fusion energy are directly applicable here, although EUV stands a better chance of becoming commercially viable at least a century sooner)

In contrast to 450mm where the only thing keeping it from being a reality today is that no one is interested in committing $20B today to build the first 450mm fab. But as soon as some company makes that decision and commitment, the tools will be available shortly thereafter. The same thing happened with 300mm, no one wanted to be first but once someone pulled the trigger then everyone scrambled to be next in line.

 

[Lepton87]

(..)

Shooting for 13.5nm made sense at the time because at the time immersion litho and multi-patterning made no technical or economic sense, and because everyone was told to have faith and believe the R&D roadmap would make it all magically happen in a timeline that would make going for something easier and less aggressive be a pointless exercise.

It turned out to be too good to be true, and here we are a decade later still waiting for EUV to deliver on the promised goods

Thanks for a well detailed answer, you should totally write some articles on the front page, even your forum posts are totally worthy of being there :thumbsup: It's very unusual for technical people to ride on faith(is that even a correct phrase?but you know what I mean) so I'm quite puzzled that something so complex and expensive was undertaken in part on faith.

 

[Idontcare]

:$ Thanks for your kind words! :$

Thanks for a well detailed answer, you should totally write some articles on the front page, even your forum posts are totally worthy of being there :thumbsup: It's very unusual for technical people to ride on faith(is that even a correct phrase?but you know what I mean) so I'm quite puzzled that something so complex and expensive was undertaken in part on faith.

The "faith" was on behalf of management, project planners and shareholders, not the "technical people" per se.

But that is no less true of all of R&D if you think about it. If the non-technical people didn't have faith in the technical people then the non-technical people would never invest in companies which do R&D of any sort, be it medical or semiconductor.

But in this case, with EUV, I would say even the technical people had little choice but to have some faith as well. The project is very wide, the technical scientist who is scrambling day in and day out for the past 7 years to get more 13.5nm photons generated in the lamp source has no idea whether or not the challenges within the "defect-free mask team" are going well or if they are facing drastically severe barriers to climb.

So in that sense, when a project has this many arms and hands, there is very much an element of faith involved in which you are hoping (if not praying) that your faceless coworkers in other pockets of the team are toiling away and successfully making progress on their piece of the puzzle because the whole thing has to come together or none of will work at all.

 

[Lepton87]

It's mind boggling how complex science projects are nowadays and how many people have to work together to get it done

 

[Roland00Address]

It's mind boggling how complex science projects are nowadays and how many people have to work together to get it done

Its always been this way. Besides the mind boggling improvements on silicon do you know how many people work on doing the human genome (90s and 00s). Before that it was the space age in the 50s to 70s. Before that the manhattan project (they literally made their own town) before that model ts, the railroads, machine replaceable parts by Eli Whitney. Everybody talks about the Great Wall of China but the Grand Canal of China (over a 1000 miles long) is in my mind more of an accomplishment. The suez was not the first canal connecting the mediterranean. Darius the great (around 500 bc) finished the "Canal of the Pharaohs" which connected the nile to the red sea and thus you can do the mediterranean to the red sea which subsequent generations improved such as the Romans (the modern version follows a very different route).

Scientists and Engineers are just continuing the great works of our predecessors, improving them and making them more grand. Same thing with our other works of progress, our artists, our literature, our teachers, etc.

Have you ever the essay I Pencil by Leonard E. Read? Link it is now almost 60 years old but it chronicles all the steps you need to do to make a simple pencil. In the very end Leonard E. Read states (as the pencil in a first person narrative) that so many hands have been involved and not one mind or one body can make a pencil, it so vast and beyond comprehension that it would take something beyond man to understand it. You have to use language such as divinity to explain it for the final result is so miraculous and yet so ordinary we take it for granted. This philosophical piece also had an intro before the essay by Nobel Economists Milton Friedman.

 

[witeken]

I understand your point, but the (technical) scale of projects seem to me a lot bigger and more ambitious than earlier times. Sure it's an iterative process, but it it's only getting more difficult. The LHC is really remarkable, and it's only going to be more difficult if they want to reach the plank scale (unification), for example.

 

[pw257008]

I understand your point, but the (technical) scale of projects seem to me a lot bigger and more ambitious than earlier times. Sure it's an iterative process, but it it's only getting more difficult. The LHC is really remarkable, and it's only going to be more difficult if they want to reach the plank scale (unification), for example.

with lower and lower payoffs...

 

[III-V]

with lower and lower payoffs...

The benefits from moving to smaller geometries are exponential in growth.

 

[pw257008]

The benefits from moving to smaller geometries are exponential in growth.

Sorry, I meant from a social standpoint. The more we know, the less we gain from every additional thing we know; the more we can do well, the lower marginal benefit to the ability to do more things better.

It doesn't have to be this way (the recent Lockheed fusion news bring hope, for instance), it just sort of is right now.

 

[Ajay]

It doesn't have to be this way (the recent Lockheed fusion news bring hope, for instance), it just sort of is right now.

It would be a huge miracle if that worked (at several levels). Without the benefit of the Sun massive gravity - it is extremely difficult to get two protons close enough together, often enough, for quantum tunneling to do the rest and start a sustainable nuclear reaction.

 

[NostaSeronx]

September 4th, 2014