Wow this is a nice read. I never thought injection moulding precision, relative to the dimensions of the object, to be in the same ballpark of what you can achieve with photolithography in chip manufacturing. This of course makes sense because we are at the end limited by the same principles of mechanical inaccuracy.
This is really inaccurate. The real reason is similar to why America was at the forefront of the other high tech sectors like aviation etc too: massive defense spending, a lot of business people (like Fairchild) willing to invest in a sector where they see the potential procurement from Pentagon, while starting to serve the civilian sector.
The origin of the semiconductor industry of USA is in the WWII military research for microwave detectors used in radars.
The vacuum diodes were no longer useful at such high frequencies, so it was attempted to make better point-contact semiconductor diodes using germanium and silicon pure crystals, instead of using natural minerals, like galena, which had poor performance and were not reproducible.
Much of this WWII research and development effort has been done at the Bell labs. The most important results were the development of technologies for purifying germanium and silicon at levels never succeeded before for other chemical substances, for growing single crystals of Ge and Si, and for doping them in a controlled way with impurities.
Before WWII, all attempts to make semiconductor devices with good performances were unsuccessful, with the exception of a few applications that had very low performance requirements, like the AC rectifiers with selenium or copper oxide. The reason is that the properties of semiconductors are hugely influenced by even very small amounts of impurities or crystal defects.
Only by the end of WWII, with the availability of pure single crystals of germanium and silicon, which were the result of radar development during the war, the research on semiconductor devices could really start.
The experimental discovery of the point-contact transistor by Bardeen and Brattain was a direct product of the Bell team trying to find other applications for the technology of making point-contact diodes that was developed at Bell during the war. Then, stimulated by the experimental results, Shockley, who was an excellent theoretical physicist, developed a theory of the electrical conduction in semiconductors that has been the basis for the invention of the other semiconductor devices during the following decades. Shockley himself has invented several semiconductor devices using his theory, the very important BJT (bipolar junction transistor) and JFET (junction field-effect transistor) and the less important PNPN diode (a.k.a. Shockley diode).
The announcement of the transistor, which was "open-sourced" by the Bell Labs triggered intense efforts of R&D in semiconductor devices at many companies in USA and all over the world.
The very quick evolution of the semiconductor industry in USA and abroad during the first decades was determined by a complete disregard for what nowadays is called "IP".
ATT and the Bell Labs licensed the semiconductor technology cheaply to anyone and even gave it for free for certain purposes (e.g. for the purpose of making hearing aids, respecting the wishes of Alexander Graham Bell, the founder of ATT).
Then, in the following years, all advances in semiconductor devices and semiconductor technology were published with complete recipes of how to reproduce them. This ensured that all innovations spread immediately to all companies active in this domain. While significant inventions were patented, at that time patents were typically still licensed fairly and non-discriminatory, instead of being used as weapons against competitors.
The semiconductor and computer industries would have never flourished and something like the Silicon Valley would have never been created in the current environment of secrecy and paranoia about "protecting IP" and of abusing the patent and copyright laws to prevent competition and reach monopoly status.
To be fair, not protecting "IP" was the right strategy when the market for the semiconductor industry was growing, because the sharing of all knowledge ensured a much faster growth of the market, which was achieved both by replacing older technologies and by creating new applications enabled by the properties of the new devices. The growth of the market ensured that sharing was a win for each company.
In a stagnant market, a company can grow only if another shrinks, so a much more adversarial attitude is needed if growth is the goal, like weaponizing the "IP".
Even in late 90's the IP wasn't waponized. Most start-ups were just a bunch of employees leaving a company to compete with them while improving upon the previous IP.
Nowadays even start-ups are paranoid and starting a semicon company is orders of magnitude harder..
This is an interesting point of view. I had read elsewhere the idea that the technology was moving so fast, copying was almost useless (this was in regards to why the SSR's did NOT get very far in semiconductor design . . . their strategy was to copy everything, similar to their nuclear and rocket strategy): by the time you were reading the paper the industry had moved so far beyond, it wasn't very useful.
Bell Labs was primarily funded by AT&T, a private company. Like I said, an environment with low taxes and real freedom is the source of prosperity, creativity, and innovation.
Why? Rice is what you eat if you can't afford anything better. This parallels every other culture - the staple food will keep you alive, but if you have any money, you'll eat something better than that.
You know how "bread and water" is considered a terrible diet that only prisoners eat, and then only because they're not given a choice?
(And how modern prisoners get a much better diet?)
Bread and water is prisoner food, but avocado toast and cream-cheese bagels at the corner bodega are considered mid-to-upper-class fare. Pasta (also wheat) can range from kraft mac-and-cheese (poor-coded) to hand-made pasta with pesto sauce.
Rice and tea (ochazuke) is historically the "bread and water" equivalent in Japan, but people of every socioeconomic class still eat rice and miso soup for breakfast, eat rice balls (onigiri) regularly, and generally eat a diet with a lot of rice.
Even though rice is the staple food of Japan, I'd actually argue that instant ramen is much more poor-coded these days than even ochazuke.
I wouldn't be surprised if the middle class and lower class eat more-or-less identical quantities of rice.
> Bread and water is prisoner food, but avocado toast and cream-cheese bagels at the corner bodega are considered mid-to-upper-class fare.
That's not an example of nuance. An expensive fruit and a heavily-processed cheese are much higher-grade food than bread is.
> Pasta (also wheat) can range from kraft mac-and-cheese (poor-coded) to hand-made pasta with pesto sauce.
Same thing; cheese is a high-grade food, and even pesto is chock full of fat.
> Even though rice is the staple food of Japan, I'd actually argue that instant ramen is much more poor-coded these days than even ochazuke.
And this is a statement that even the poorest people in Japan aren't so poor that they have to subsist on rice. There's no question about which of instant ramen or ochazuke is a better meal. Instant ramen comes with tons of spices, fats, salt, some vegetables, and even a little meat.
Complex processors like AMD Athlon and Intel Pentium 4, which were made in 180 nm a quarter of century ago, had clock frequencies between 1 GHz and 2 GHz. Pentium 4 used internally a double frequency clock for the simpler 32-bit arithmetic-logic units, i.e. up to around 4 GHz.
Today the manufacturing process could be better optimized than 25 years ago, so some logic circuits much simpler than a 64-bit CPU (the previous were 32-bit CPUs for integers, but they had 64-bit/80-bit FPUs working at full speed), i.e. with much less gate delays per pipeline stage, might be able to reach 12 GHz.
However, something like a 64-bit ALU will certainly not reach 12 GHz. Even a 32-bit ALU is very unlikely to reach 12 GHz. Simple things, like shift registers and Galois-field counters, might reach such speeds, or even higher.
The next CMOS process generation, i.e. 130 nm, already allows making complex processors with more than a half of the maximum clock frequency of the fastest processors of today. It also allows making analog amplifiers and mixers for the 5 GHz WiFi frequency bands.
This is plainly wrong. I'm a chip designer. There's no way to implement a DFF operating at 12 GHz in 180nm, period. This isn't an optimization problem, it is physics.
[2] "If an elderly but distinguished scientist says that something is possible, he is almost certainly right; but if he says that it is impossible, he is very probably wrong." - Arthur C. Clarke
Because you can just look into it and see if it's what you sent fof production, and if not and the word gets out you are done as a fab. Fab business is about trust. You also should trust that your design isn't leaked to the competition.
It's very common to xray the dies, especially for debugging. Also common is to etch it layer by layer, take photos and rebuild the circuit schematic, mainly for reverse engineering but I've seen companies doing it to their own dies too.
Things get more blurry at the board level, the combinations of suppliers and service providers are endless.
That's one way to make sure people living under aerial bombing firmly support a regime defending their sovereignty, hence legitimizing the islamic republic. Example: Taliban, with boots on the ground, didn't get any weaker at the end.
Instead of this we have anti dual-use policies, especially in semiconductor. Any chip a fab produces need hefty paper work to prove it cannot be used for military. This is due to the military-industrial complex lobby. They don't want cheap competition.
Apparently 14 K cooling is not used even up to 5N or 6N purity, commercial large-scale sources use various other tricks to remove the other gases. They do cool the input gas down to liquid nitrogen temperatures as one of the first steps.
My point is that there's "maximally efficient / profitable" versus "can be made available as an emergency alternative".
Cooling to 14 K isn't the cheapest option, but it has very low complexity. You can "simply" pressurise the source gas, cool it to room temperature through an ordinary heat exchanger, then allow it to expand. The only issue is that if you do this naively, the expansion nozzle will get clogged with ice.
Obviously, this wastes a lot of Helium, but we have lots of it. If what's needed is high purity Helium, then throwing away even 90% to get 10% that's 6N pure should be no problem for an industrial nation.
You can't just spin up such a facility in a few days or weeks though, surely? Even if the core of a process is relatively simple physically, you still need all the supporting infrastructure to make it happen.
Starting from an empty lot, no, it would take nearly a year.
However, any air (or gas) liquefaction / separation plant that is already making purified industrial gases from air or other sources could be adapted in a matter of weeks or at most a couple of months.
