Every Government wants a fab, Does the market?

[Mad2]

....

The publicity around the formation of Rapidus indicates they are interested in small batch and short cycle but it is hard to know exactly what they are thinking. I know Japan has funded research on small lines but what was published has not seemed useful outside of a research lab.

It's been done 30 yrs ago: https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=670305 . Impractical for manufacturing/cost then and now...but a nice dream.

 

[Tanj]

Even assuming 5 wph, the E-beam tools would take up ~30x more space than a comparable DUV tool and ~5x the size of the new hotness of low NA EUV tools.

True, it would be madness to charge straight in front of the EUV juggernaut. Hence my interest in the overal process flow for maskless while keeping the machines simple, but it seems the rest of the fab will be a severe drag on getting much benefit from that. But it might keep hot lots from interrupting the EUV machines, for example. Still learning about fab flow.

Also, EBL has much finer resolution potential than EUV.

Using DUV would also defeat the stated purpose of it being maskless flow so anybody in their garage can tape out on this thing.

Which is why I mentioned needing maskless DUV as well, so the whole stack can be maskless. I doubt it will simplify design rules enough for everyone to run EDA in their garage, or avoid deep investments in verification. There will still be process margins, stochastics, variations in registration, etc. and good EDA tools will be needed for DFM. Plus, in hot lots for engineering samples you may want the chips to be as close to the same as the HVM version that will later be running using pattern masks, so the EDA would need to apply a complete set of DFM rules, then just tape out for the device pattern not the mask pattern.

Maskless is not magic. Which is why I am interested to understand how it could be of use in the real fabs.

 

[Tanj]

It's been done 30 yrs ago: https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=670305 . Impractical for manufacturing/cost then and now...but a nice dream.

Thanks for that reference. In some ways back then it was more advanced than the recent Japanese effort, which is clearly not aimed at commercial production: https://www.asia-anf.org/wp-content/uploads/2019/02/Minimal-Fab.pdf But the Japanese minimal fab is aiming at a very different price point. They sold one system into a university in Saudi Arabia, for example, it seems for use in postgrad training for semiconductor design.

Of course a line such as the MMST would be absurdly expensive per wafer if using a modern stepper. But what about a slower, cheaper maskless litho machine? I wonder what the cost and throughput would be needed to make it commercially attractive. After all, even if cycle time takes a week or two with the number of process steps on a modern chip those hot lots would be in demand at very premium price. And you could argue there is a huge cost benefit in removing the hot lot problem from the high volume fabs, while keeping customers happy.

 

[Barnsley]

Thanks for that reference. In some ways back then it was more advanced than the recent Japanese effort, which is clearly not aimed at commercial production: https://www.asia-anf.org/wp-content/uploads/2019/02/Minimal-Fab.pdf But the Japanese minimal fab is aiming at a very different price point. They sold one system into a university in Saudi Arabia, for example, it seems for use in postgrad training for semiconductor design.

Of course a line such as the MMST would be absurdly expensive per wafer if using a modern stepper. But what about a slower, cheaper maskless litho machine? I wonder what the cost and throughput would be needed to make it commercially attractive. After all, even if cycle time takes a week or two with the number of process steps on a modern chip those hot lots would be in demand at very premium price. And you could argue there is a huge cost benefit in removing the hot lot problem from the high volume fabs, while keeping customers happy.

If maskless come to the fore , where does this leave ASML?

Is it not possible for China/India for example to go down this route as nobody seems to be doing it seriously

 

[cliff]

Hi Cliff,
Yes, I agree that legacy ME should be a priority. For instance, I think that the world is incredibly vulnerable if China decides to disrupt the supply of 28nm ME. A global agenda that addresses the production of legacy ME is just as important as SOTA ME. Unfortunately, it’s not a sexy endeavor, doesn’t drive job creation, and wont get anyone re-elected.

What's ME and SOTA ME?

 

[cheesehead]

Hi Cliff,
Yes, I agree that legacy ME should be a priority. For instance, I think that the world is incredibly vulnerable if China decides to disrupt the supply of 28nm ME.

How could China do so? Would they restrict production capacitiy and/or export of wafers, or do these mature nodes require specific raw materials that can (currently) only be sourced from China?

 

[Tanj]

If maskless come to the fore , where does this leave ASML?

ASML is manufacturing 300 WPH on DUV and 150 on EUV, with goals to go much higher. As I pointed out above, that is a juggernaut. Even at, say, 5 WPH a maskless machine would just fill a niche like hot lots or low-volume specialty chips, not displacing the kind of HVM engineering in a modern ASML (or even Nikon) scanner.

Is it not possible for China/India for example to go down this route as nobody seems to be doing it seriously

Well yes but their ambition is high volume manufacturing, so why would they? A state of the art process is a lot more than just the litho, and you need volume for the CAPEX to be paid off. I suspect you could make a viable business with low volume versions of the mainstream equipment, but that is still a lot of R&D for what will be just a small business compared to regular fabs. Indeed, my prediction is that it would best work as an adjunct to ta big fab, allowing hot lots to be processed without interrupting the valuable orders, but mimicking the same process so that engineering samples can graduate smoothly to the main fab after they have been verified.