Intel splits up manufacturing group amid production delays

[drunkenmaster]

I will repeat my question to you @drunkenmaster . Is tin the requirement or any molecule, be it oxygen, hydrogen, nitrogen, neon, etc gas molecule, can form the plasma?

As I understand it Tin is simply the most efficient in terms of power output, ie you get a much stronger 13.5nm EUV beam from a given power input from tin based plasma than anything else by quite some margin, Xenon is the next most efficient commonly used molecule to be used and it's drastically less efficient (~30% or so less at a glance), and a couple other less common more problematic molecules are both more expensive and still no where near as efficient as tin. The reason for 13.5nm is again as I understand it that EUV is extremely heavily absorbed by pretty much everything and the best reflective surfaces they can make to be used to aim the light reflects the more light in the 10-14nm range. Meaning 13.5nm and tin produce the most power efficient EUV output possible (currently) and switching between materials and settling on the 10-14nm range is what has allowed them to scale up power towards levels considered commercially applicable.

Usually with these it isn't as simple as a binary can or can't - generally you will get some sucesses but with too low chance to be commercially viable.

I mean it's binary in that either you can get quad patterning working or they can't get 10nm out because they can't make feature sizes significantly smaller than their 14nm node without it, and without it they can't make competitive 7nm chips either.

 

[4K8KW10]

As I understand it Tin is simply the most efficient in terms of power output, ie you get a much stronger 13.5nm EUV beam from a given power input from tin based plasma than anything else by quite some margin, Xenon is the next most efficient commonly used molecule to be used and it's drastically less efficient (~30% or so less at a glance), and a couple other less common more problematic molecules are both more expensive and still no where near as efficient as tin. The reason for 13.5nm is again as I understand it that EUV is extremely heavily absorbed by pretty much everything and the best reflective surfaces they can make to be used to aim the light reflects the more light in the 10-14nm range. Meaning 13.5nm and tin produce the most power efficient EUV output possible (currently) and switching between materials and settling on the 10-14nm range is what has allowed them to scale up power towards levels considered commercially applicable.

I was thinking of what's more efficient in terms of power input - because one can use relatively low-power magnetrons to get plasma from Xenon/Neon, etc.
I don't know whether the magnetrons can produce plasma from tin, though.

 

[Rroff]

I mean it's binary in that either you can get quad patterning working or they can't get 10nm out because they can't make feature sizes significantly smaller than their 14nm node without it, and without it they can't make competitive 7nm chips either.

Grossly simplified it is like running a non-native resolution on a monitor - some instances of the same character in a block of text render fine others are a mess with lines too thick or too thin, etc. with enough iterations you mihht get some sucesses.

 

[drunkenmaster]

I was thinking of what's more efficient in terms of power input - because one can use relatively low-power magnetrons to get plasma from Xenon/Neon, etc.
I don't know whether the magnetrons can produce plasma from tin, though.

It's about the specific output of wavelength. Only one state of Xenon pumps out 13.5nm wavelength from plasma, most is at 10.8nm, which is actually many times more powerful than it's 13.5nm output. The more specific reason is reflectivity. Mo/Si mirrors basically work at 13.5nm while potentially Xenon putting out 10.8nm and using Beryllium based mirrors could work but they are toxic and difficult to make.

THere was definitely potentially and I believe they were even slightly more reflective which would have made a ~11nm wavelength potentially more efficient, but beryllium is just so toxic it seems like everyone basically said **** that, we'll go with 13.5nm. It was also only a very small amount more efficient in reflectivity. Lots of different tin states output 13.5nm wavelengths, and others, but the peak for tin is at 13.5nm and works well in the range for Mo/Si based mirrors so that just fit together better. It seems like the primary consideration was that Mo/Si was the only reasonable choice so they needed something to output in the wavelength's that combination reflected best which led to 13.5nm (or close) and tin was found to be by a good margin the most efficient. The peak wavelength goes down as atomic number goes down which is why Xenon's peak is lower.

Presumably if they found a way to work with beryllium safely then they'd have picked well, definitely not tin as it barely outputs anything below 12.5nm, but quite likely xenon which also produces less debris than tin so would be an even bigger bonus, though it's possible a different molecule would be more efficient than xenon.

 

[Mercutio]

I like cheesecake.
(You boys have fried my brain beyond any reasonable OT discussion and hat's off to you both).