semiconductor folks!
-
neither me nor anybody i've talked to so far (including some industry people) could point at any process steps they know involves helium, and i've no idea where this claim originated
from some discussion and research, it seems like one of the major uses is in ESC/BSG wafer clamp devices (https://csmantech.org/wp-content/acfrcwduploads/field_5e8cddf5ddd10/post_6573/15.3_Characterization_of_Electrostatic_Chuck.pdf) where flow of helium is less than 20 sccm per chuck.
this represents about 2 micrograms of helium per second per chuck, which i'm not too concerned about. there are of course some other applications that may consume more helium
-
@whitequark FWIW, my hunch is, if chipmaking is affected by helium deficit, the troubles would be more likely to hit R&D and possibly the construction of new foundries than ongoing manufacturing processes. Cryocrystallography is a thing in materials science research, and while it used to largely get by with liquid nitrogen, the bleeding^W chattering edge of new research has been shifting steadily coldwards.
@riley yea that's reasonable
-
@whitequark That's a good question ye.
Maybe some of the vapor deposition processes?
@Hemera As far as I know, all currently used cold vapour deposition techniques of relevance to (mainstream) chip-making can be done at liquid nitrogen temperature or higher. Quantum computer research might have uses for colder ones. It's not impossible that NSA could be using helium-requiring chips in large numbers, and their supplier(s) would have trouble without helium, but I don't know of solid leaks positively affirming such, so far.
-
@Hemera As far as I know, all currently used cold vapour deposition techniques of relevance to (mainstream) chip-making can be done at liquid nitrogen temperature or higher. Quantum computer research might have uses for colder ones. It's not impossible that NSA could be using helium-requiring chips in large numbers, and their supplier(s) would have trouble without helium, but I don't know of solid leaks positively affirming such, so far.
-
@be_far thanks. I have some background in vacuum systems so this makes a lot of sense. do you know the ballpark number for how much helium is used, after accounting for reclamation?
@whitequark the one number I’ve ever seen (couldn’t ever find a source, everyone just repeats it) is that one advanced TSMC fab uses 500,000 cubic meters of helium a year. If it’s used as a purge gas like you mentioned in another comment I imagine it’s very low reclamation but I don’t know if that number takes reclamation into account (even if it’s not made up).
-
@whitequark Oh. Yeah. That could be a use.
-
@whitequark @riley @Hemera Purge gas is the explanation I had seen, and I had heard also that EUV machines needed it (though apparently, people in this thread are suggesting hydrogen is used ?)
I think it would make sense to have a non-reactive gas atmosphere, though.
-
@whitequark the one number I’ve ever seen (couldn’t ever find a source, everyone just repeats it) is that one advanced TSMC fab uses 500,000 cubic meters of helium a year. If it’s used as a purge gas like you mentioned in another comment I imagine it’s very low reclamation but I don’t know if that number takes reclamation into account (even if it’s not made up).
@be_far yeah it's infuriating how often someone just pulls a figure out of who knows where and then everyone else repeats it
-
@be_far yeah it's infuriating how often someone just pulls a figure out of who knows where and then everyone else repeats it
@whitequark it has to be AI generated at this point, you’d think there would be some interview somewhere saying it but I’ve never found one. TSMC’s own periodic sustainability reports are in dollars, maybe you could convert using price data?
-
@whitequark@social.treehouse.systems @riley@toot.cat @Hemera@meow.social I think some lasers also use helium, either to get the right frequency or temperature. Might be required to etch the gates at the required size?
-
semiconductor folks! I've seen a lot of talking heads repeat the claim that "a helium shortage is bad for chip production", never substantiated with useful information. do any of you know:
- what is helium actually used in the processes?
- which specific processes would be affected?
- how much helium (ballpark) is needed per year?
- where, if anywhere, a closed cycle is used?
- what happened to the strategic helium reserve in the US?
@whitequark I think helium makes the bubbles smaller giving the chips a more premium mouthfeel. -
semiconductor folks! I've seen a lot of talking heads repeat the claim that "a helium shortage is bad for chip production", never substantiated with useful information. do any of you know:
- what is helium actually used in the processes?
- which specific processes would be affected?
- how much helium (ballpark) is needed per year?
- where, if anywhere, a closed cycle is used?
- what happened to the strategic helium reserve in the US?
@whitequark@social.treehouse.systems it’s used for loads of things, as a carrier gas, for cooling, and the main way to find leaks in an ultra high vacuum system (of which fabs have loads) is by spraying helium around the outside and looking for helium that makes it inside
there’s probably plenty of other uses I’m just ignorant of -
@whitequark@social.treehouse.systems it’s used for loads of things, as a carrier gas, for cooling, and the main way to find leaks in an ultra high vacuum system (of which fabs have loads) is by spraying helium around the outside and looking for helium that makes it inside
there’s probably plenty of other uses I’m just ignorant of@xeno do you work in the industry?
-
@whitequark@social.treehouse.systems it’s used for loads of things, as a carrier gas, for cooling, and the main way to find leaks in an ultra high vacuum system (of which fabs have loads) is by spraying helium around the outside and looking for helium that makes it inside
there’s probably plenty of other uses I’m just ignorant of@xeno @whitequark yeah but leak detection can't be a high volume use compared to actual production work. Like how often do tools develop leaks if you're not actively working on them?
-
@xeno do you work in the industry?
@whitequark@social.treehouse.systems oh nope sorry I didn’t parse “semiconductor folks” as people in the industry
I have been doing smal-scale semiconductor research work and I’ve learned from some actual semiconductor people but I am def not one of them -
@xeno @whitequark yeah but leak detection can't be a high volume use compared to actual production work. Like how often do tools develop leaks if you're not actively working on them?
@xeno @whitequark i would expect most leaks to be after maintenance outages then the tool runs leak free for a comparatively long time
-
@xeno do you work in the industry?
@whitequark@social.treehouse.systems oh nope sorry I didn’t parse “semiconductor folks” as people in the industry
I have been doing small-scale semiconductor research work and I’ve learned from some actual semiconductor people but I am def not one of them -
@whitequark@social.treehouse.systems oh nope sorry I didn’t parse “semiconductor folks” as people in the industry
I have been doing small-scale semiconductor research work and I’ve learned from some actual semiconductor people but I am def not one of them@xeno oh no research definitely counts
do you know how much helium is used as a carrier gas? cooling in ESC/BSG systems seems to be "few sccm" which is not a geopolitically relevant amount, neither is leak checking
-
@xeno @whitequark yeah but leak detection can't be a high volume use compared to actual production work. Like how often do tools develop leaks if you're not actively working on them?
@whitequark@social.treehouse.systems @azonenberg@ioc.exchange that’s true,
I know people in other labs with UHV stuff who have had to install He recycling systems and stuff over the past couple years to deal with the shortage but they might have some niche use for lots of He they never told me about, I was just thinking of “things that use He” -
semiconductor folks! I've seen a lot of talking heads repeat the claim that "a helium shortage is bad for chip production", never substantiated with useful information. do any of you know:
- what is helium actually used in the processes?
- which specific processes would be affected?
- how much helium (ballpark) is needed per year?
- where, if anywhere, a closed cycle is used?
- what happened to the strategic helium reserve in the US?
@whitequark I think it may be used in some plasma etch processes. Dry etching is basically sorcery
