Hopefully the complexity of fabs and their operations makes it obvious why chip shortages can easily take 3-5 years to address if creating new fab capacity is the problem. It also costs between $30B-$50B to create a new "mega/ultra/super fab" operating at volumes typical of TSMC, Intel, Samsung, etc.
Even then, the small corner of a fab my business operates in isn't even mentioned in the article because it's far "too small" - all the businesses combined into our segment has total market revenue ranges in the $5B-$10B/year range. We are essential but small money-wise for most people to care.
The other part of the story with water - some of the most advanced water purification and recycling development is happening in this realm as well. Another company product line unrelated to ours is able to detect 10-nanometer-sized contaminants in water at volume both to assure incoming water purity as well as outgoing. Much of the total volume of water is recycled.
There are also surprising ways seemingly unrelated parts contribute. For example our product operates on clean-room floors but we put a ton of effort to reduce the floor footprint (we're 30% smaller than the de facto competitor). That directly reduces the amount of water consumed as a multiplier effect because it means you get more productive output per square foot of HVAC air processing capacity. That HVAC is primarily chilled water based. It also reduced energy consumption or increases productive output per unit kW.
Whenever you have high capital investment, you get the most innovation in basic technologies required for the business and the most spin-offs as well. Whenever you have low investment, you get zero innovation.
If you want to find out how chips are actually made, this documentary filmed inside an Intel fab in 1978 is excellent: https://en.wikipedia.org/wiki/Now_the_Chips_Are_Down You can find the full thing on iplayer (in the UK) or various video sites.
Also "Indistinguishable from Magic: Manufacturing Modern Computer Chips" https://www.youtube.com/watch?v=NGFhc8R_uO4 is a presentation from a manufacturing engineer which goes into some nitty-gritty
Seconded. Such a well done presentation absolutely packed with info and lots of inside-baseball comments which were interesting as well. I'll never be within 10 miles of a fab, but still totally worthwhile knowledge to have.
It would be disastrous. While TSMC does have a few foundries outside of Taiwan, the company accounts for about a third of total global semiconductor production. Maybe more if you consider just the most advanced process nodes. It's not that we'd lose their particular capacity, since Samsung is a fairly equal competitor, but the overall loss would be catastrophic.
Everyone would be in a dire situation, but computers and smartphones would especially be hit hard because they are using the newest nodes. As an example, Apple's SoCs (at least the recent ones) all use TSMC 5nm and that production would just disappear.
It's not correct that all our eggs are in one basket. There just aren't enough baskets to accommodate the loss of an entire one. Other players include SMIC, UMC, Global Foundries, and now Intel, but their capacity isn't significant enough to make up for the loss.
This is why there is a push both to "second source" production, and to increase production capacity overall, especially in more politically stable geographies. The latter is a huge challenge because it requires a lot of capital and a lot of time.
I do hope more fab's are set up around the world, it would be horrible to have to build fabs after just losing 1/3 of the worlds manufacturing capacity, especially as the new fab itself would consume so much semiconductor in its own right.
There's a reason the US has a symbolic military presence in Taiwan, most likely even on the Pratas island to train Taiwan defence troops there. It would be the first and obvious strategic target, and taking the otherwise uninhabited island without hitting the potentially present US military personnel would be hard - which should make China at least think twice about making a move against it.
Quora: How does the NYT determine which articles have comments?
Bassey Etim, Community Manager for the New York Times:
The vast majority of NYT comments are handled by a human moderator. This means that we have to make an editorial decision about which comment threads we will open for comments each day. Also, we adhere to a sort-of "slow moderation" theory, which posits that the best way to respect the commenting efforts of our readership is to ensure that their comments exist in an urbane, literate environment. (Definitely not an approach that is good for everyone, but it works fabulously for us, for reasons you can probably divine.) Our goal is to have every NYT comment thread offer tangible added value to each article for our readership.
The costs to this approach are obvious -- it takes a long time, and many stories do not get comments. But NYT readers expect the highest-quality everything from us, so that's what we deliver.
To answer the root of your question, these are our general criteria for opening a story for comment, generally in order of importance:
- News value of the story
- Projected reader interest in the story
- Have we recently had comments on this issue?
- Whether we can moderate the projected # of comments in a timely fashion.
Any idea why the colour on the fab floor is an odd yellow hue? Is the standard colour spectrum for interior lighting not compatible with chipmaking? I'm imagining a situation akin to the darkroom requirements for processing photographs.
Anything UV sensitive is usually done in a yellow room, so that's probably it.
I know this is exactly the case for PCB manufacture and printing press plate making to lessen accidental exposure, so it's exactly like a red safelight in a darkroom.
My understanding is that was the original purpose, but probably unnecessary given that all the relevant process steps are encased in tools that are not exposed to the general clean room environment in operation (they have to maintain a higher grade of clean room internally).
However, I suspect it's probably going to continue being propagated forward due to Copy Exactly which has historically meant that Intel documents and copies the smallest details in all their fabs just in case they are necessary. In the end, if it ain't broke...
Chesterton's Fence [0] should overrule fear of cargo culting, especially when the fence (or in this case color) was clearly put up by people who really knew what they were doing at the time.
Unless there is a significant cost to it, and unless the benefit is understood, there is no reason that they should be faulted as ignorant cargo cultists simply for keeping it in place.
> For decades, Intel’s cleanrooms have been lit like darkrooms, bathed in a deep, low yellow. “That’s an anachronism,” says Mark Bohr, a small, serious man who has spent his entire 38-year career making chips, and who’s now Intel’s top manufacturing scientist. “Nobody’s had the courage to change it.”
It's possible that some level of unresolved empirical "magic" be involved in the process, which does not diminish the value of the end product. Why deprive yourself of hard learned lessons even if you don't understand it all?
This remembers me, when many years ago I worked with old chemical photography - used special light to limit parasitic exposure of film and photopaper.
Usually used dark red light, or yellow-green.
And I lived in absolutely typical apartment, and used for photography ordinary bathroom, but switched off normal lights and used special lights, and apartments become looking like them from fantastic films.
So, yes, my first education electronics and I've few times been at real fabs.
At fabs there are two things which don't appear in typical homes - many devices emit ultraviolet and it is extremely powerful, so appear even through covers, or use open electric heaters, which emit red.
Main lights I have seen there, where just ordinary fluorescent lamps, but I sure, on more modern fabs, used special lamps or color filters, to filter out part of spectrum which could cause unnecessary additional dose for photo-chemistry.
Not all lithography is Deep/Xtreme UV even in the newest fabs, because the equipment is so expensive and relatively slow. A lot of lithography is done the old way and is sensitive to UV from any lights. Box to box processing is almost complete, but there are times when a box may need to be opened or exposed to ambient light.
So there are some pretty good reasons to not to increase uncontrolled UV light exposure.
It's because photoresist is sensitive to deep blue out to UV so you want to avoid that kind of light as wafers with photoresist applied but not-yet-exposed are running around in their robotic FOUPs through the facility.
It's exactly analogous to red safe lights used for film print photography.
Hopefully the complexity of fabs and their operations makes it obvious why chip shortages can easily take 3-5 years to address if creating new fab capacity is the problem. It also costs between $30B-$50B to create a new "mega/ultra/super fab" operating at volumes typical of TSMC, Intel, Samsung, etc.
Even then, the small corner of a fab my business operates in isn't even mentioned in the article because it's far "too small" - all the businesses combined into our segment has total market revenue ranges in the $5B-$10B/year range. We are essential but small money-wise for most people to care.
The other part of the story with water - some of the most advanced water purification and recycling development is happening in this realm as well. Another company product line unrelated to ours is able to detect 10-nanometer-sized contaminants in water at volume both to assure incoming water purity as well as outgoing. Much of the total volume of water is recycled.
There are also surprising ways seemingly unrelated parts contribute. For example our product operates on clean-room floors but we put a ton of effort to reduce the floor footprint (we're 30% smaller than the de facto competitor). That directly reduces the amount of water consumed as a multiplier effect because it means you get more productive output per square foot of HVAC air processing capacity. That HVAC is primarily chilled water based. It also reduced energy consumption or increases productive output per unit kW.
Whenever you have high capital investment, you get the most innovation in basic technologies required for the business and the most spin-offs as well. Whenever you have low investment, you get zero innovation.