A crucial idea for silicon PV cells was excluded by a patent for 20 years
theconversation.com
theconversation.com
*and the environment.
The article itself doesn’t provide much detail to back up its claims. And specifically it would be interesting to understand the patent issue better.
The patent they cite says claims “ A silicon single crystal produced according to Czochralski method using a melt in contact with a quartz crucible, to which Ga (gallium) is added as a dopant that controls resistivity of the crystal in a range of 5Ω.cm to 0.1Ω.cm, wherein a diameter of the single crystal is 4 inches or more, and the single crystal is used for a solar cell.”
Which seems pretty broad. Would that really have held up in court if it was tested? Was this patent really blocking fabrication of gallium doped silicon?
A text from 2015 suggests there were open issues around fabricating gallium doped silicon:
https://books.google.co.jp/books?id=S43SBQAAQBAJ&pg=PA253&re...
So it seems likely that there were open issues around fabrication until at least 2015. Have there been other process developments that have made Gallium doping viable in recent years?
The patent largely covers the fabrication process, and was recently licensed to a Chinese fabrication company. Is it likely that this IP was really blocking them? I.e. given that they may not publicly disclose their fabrication process, how would you know they were infringing.
If this IP was of such fundamental importance why was it not challenged? This is rare in my experience (outside of semiconductors at least).
So, it doesn’t seem clear cut to me. And it would be interesting to understand the issues better.
How often will that happen? The existing system seems to massively favour keeping existing big companies big. A small inventor can easily ignored. Meaning, there's quite a difference between being right and ensuring a company does the right thing.
Some products (e.g. mobile phones) are covered by hundreds of patents. A small inventor having a huge benefit from patents? It maybe happens, but I really doubt the benefits outweighs the clear damage that patents are having.
What this means is that any practical implementation of an idea needs to be a multi-million dollar idea in order to be worth patenting. Otherwise you're just wasting money.
I.e., a patent that protects you against a later filer, but that’s it.
Patents are meant as a reward for making an invention, the deal being that you get a monopoly on this new invention for a while and in exchange you have to tell the public how to implement your invention. If you could patent known things that would defeat the whole point.
What changed is that until recently the US operated under a first-to-invent principle, so if two people invented the same thing and they both wanted a patent on it, the one who invented it first would get the patent. As you can imagine, proving when you invented something can be quite difficult at times, and the whole process can be messy. First-to-file just means that in such a situation, the first person to file wins, which puts the US rules in line with other countries and makes things much simpler.
The point is that with a patent, the small inventor can go on to raise further funds to exploit and defend the patent if necessary.
I’m not defending the patent system, but saying it’s too expensive doesn’t make much sense.
Yet AFAIK due to some patent being held by Stratasis that made only professional level horrendously expensive and proprietary 3D printers that never got any wide adoption, the 3D printing revolution was held a decade+ until the damned patent finally expired. And the rest is history.
https://en.wikipedia.org/wiki/Fractal_compression
Slow to encode, and there's no reason it should perform better than DCTs or any other modern image codec that's built around being machine-friendly.
The reason fractal compression sometimes got much better compression ratios than DCTs is apparently that it did a better job of capturing the structure of the world (a prior probability distribution), more than anything about the quirks of visual perception. We know lots of basis functions that are a little better at giving us sparse bases than DCTs for low absolute error on real-world images, even before any kind of perceptual weighting. IFSs aren't quite linear basis functions (I mean they're normally "linear", but what's being transformed linearly is the (x, y) vector and not the (r,g,b) vector), and it wouldn't be terribly surprising if they could do better still. Particularly given past examples of them doing amazing.
He used his political connections to extend his patent and only after that expired could the various improvements made by users (paricularly the cornish tin miners) be used more widely.
And as with anything that threatens the establishment, there's some right wing economist writing a paper about how it's all a myth:
https://www.jstor.org/stable/10.1086/658495
Looks like he works for the Cato Institute now so we should probably consider this alongside their climate change commentary in terms of baysian likelihood of truth.
But interestingly, he's mostly responding to libertarian economist who argue that IP in general is a government monopoly with all the stuff you'd associate with that:
Creditors/investors don't like backing something if the asset disappears if the business itself doesn't work out.
Answer: definitely less than 20 years. So why then should the first group to discover this be given 20 years of exclusivity?
People claim that this promise of exclusivity drives the research. On the other hand, why pour money into research if there’s a strong chance that my competitor will beat me to the punch and then forbid me from making use of the equivalent outcome that my own research yields (and of the in-house talent I developed along the way)?
>Answer: definitely less than 20 years. So why then should the first group to discover this be given 20 years of exclusivity?
What is absolutely hilarious about your musings here is that you are completely, 100% wrong. You make un-informed guesses and manage to get just about every aspect of the issue completely wrong. Well done.
Would you like to know when it was known that gallium-doped silicon substrate was going to perform well and possibly better than boron-doped substrate? At least the mid-70s. Would you like to know when this patent was issued? 2000. So for almost 25 years everyone KNEW gallium was better. It was literally sitting out there on every periodic table on the planet and for some reason no one produced gallium-doped PV cells at scale or cost. Why is that?
Maybe because the trick was not knowing that gallium is what you wanted to dope the silicon with, but in knowing HOW TO ACTUALLY PULL IT OFF. For more than two decades it was staring everyone in the face. For more than two decades everyone knew what the target was. For more than two decades NO ONE DID IT. That is why the first group to figure out how to manufacture gallium-doped silicon with the proper amount of other components were given exclusivity, because if they had not managed to do it then maybe we would all still be waiting for gallium-doped PV cells.
Just looking at the field of software process, there are things I've been doing for 20 years that sure seem obvious to me and that I thought would be obvious to everybody else in short order. But here we are and the dominant process approach has gone from "chaotic waterfall" to "chaotic waterfall with Scrum jargon and modestly shorter delivery cycles".
Turns out that it is incredibly fucking hard to manufacture silicon ingots with the correct doping but without too much oxygen in them that make the gallium-doped wafers perform worse than boron-doped ones. Everyone knew gallium was a better target, no one had a fucking clue how to make them at scale or at an acceptable cost. Figuring this out is ENTIRELY what this patent is about.
"Yeah we violated your 'something but on the internet patent' Here's a $50 Home Depot card."
But there is a world of difference between, 100's of people working for a decade on a problem kind of patents. And someone sitting on their couch for an afternoon patents. You seriously want to protect the former. And not really the latter.
Contractor says: we can do this for $X.
Government says yes or no to the project.
You do the same for patent applications.
It's basically answering the question "how much is this technology worth to society?" If they don't pay for it, then the technology will end up as a trade secret.
The idea is that patents are only supposed to be granted for novel, nonobvious things. If nobody had ever realized that turguts would be advantageous for frotzing quibbins, then you're an inventor if you figured out how to do it, and you can get a government-granted monopoly. But presumably if you're offering to sell someone a turgut-based quibbin-frotzing product, even if you haven't actually built it yet, you must be pretty sure it's workable and worthwhile. And now anybody else can go ahead and start frotzing the quibbins with a turgut now that you've pointed out how, and although the patent law gives you a bit of a grace period to patent it, it doesn't give you unlimited leeway.
If it turns out that the turgut has to be freebled first, then you (or anybody else) might be able to patent the freebling part at any later date, since that wasn't in the original offer to sell. But only if the need for freebling wouldn't be obvious to one skilled in the art.
So, basically, patents are intended for things that are not known to be possible until someone figures out how to do them. Lemelson got away with some pretty shameless exploitation of the loopholes in these criteria, a lot of which are now closed.
Man, I hate this argument. A secret is a secret because it's not obvious. If something is naturally likely to be invented by the first person who happens to encounter the problem it solves, it's not a "secret."
That seems to have been what happened here. Society does not benefit from granting a government-enforced 20-year monopoly on things like this. In fact, the cost in terms of wasted energy is incalculable. Patents wall off entire areas of R&D in many instances; we literally don't know what they cost us to grant.
Gallium (as GaAs) was being used in solar cells in the mid 60s. A quick search finds papers from the late 70s describing different effects of boron and gallium doping of silicon for solar cells to extend the life of satellites (the effect of the patent in question is to extend the working lifetime of panels so that solar radiation does not damage the panel.) So apparently this was so obvious that the first person who happened to encounter it missed it. It was then missed over and over and over again for at least 20 years and possibly 35.
Yeah, definitely something "likely to be invented by the first person who happens to encounter the problem it solves"...
More cynically, they may have preferred the fact that the panels would degrade over time, driving sales. Now that the patent is expiring, they either dope with gallium or lose customers to their competitors.
The article somehow doesn't mention that.
JA mentions gallium, but they seem to consider the weather-tightness of the surface film to be a bigger factor in lifespan.
[1] http://taiyangnews.info/technology/taiyangnews-500w-conferen...
https://www.jasolar.com.cn/index.php?m=content&c=index&a=sho...
If the title read "silicon solar cells" it would be clearer to the layperson (like me).
Many of us read the comments to decide if the article is worth reading, so it helps.
Strictly speaking, it ought to be "PhV" (compare "PhD"), but no one does that for some reason.
English has this stupid policy of keeping the original spelling of any loanword it picks up, gradually drifting closer and closer to being an ideographic script. In this case English got its Greek morphemes by way of Latin, where they'd mostly already acquired standard spellings based on the older pronunciation. In Spanish, which has a phonetic script, it's much saner: "panel fotovoltáico".
Maybe we should start calling them ΦV panels.
https://en.longi-solar.com/home/events/press_detail/id/228_G...
There are modest but real improvements in efficiency and resistance to multiple cell degradation mechanisms. Cell lifetime does not have a hard cutoff. Typically a panel's lifetime is limited by failed backsheets or wiring/soldering rather than by the more gradual optical-electronic degradation mechanisms that gallium doping helps prevent.
The key for both is that the government is providing exclusive security of tenure over some natural opportunity in order to facilitate production. But if people take the opportunity but don't do anything with it, whether for speculation or laziness, they are excluding others. That would only be acceptable if they have paid the community for that right to do so.
Otherwise, patents are to protect profits companies must recoup for their R&D spending.
This system has failed in a number of ways.
1. The rate of technological evolution was significantly slower, and a time-limited monopoly was less impactful, in previous history. The industrial revolution covers some 100 year period. The digital revolution covers about 40. Things get obsoleted much faster these days, to the point that a 20-year monopoly is literally the whole lifetime of the technology
2. Patents have been granted too easily for too little (largely because there’s no repercussion for filing, and re-filing, dumb patents) allowing for extremely broad interpretations, and a single technology incorporating hundreds of different patents (eg h265)
3. Because there’s so many of them, and they’re often so vaguely defined, I’m fairly certain almost no one actually reads them to learn how to implement something, or improve on the design. I’m also fairly certain that reading patents is a great way to “poison” yourself — if it can be shown you read the parent at some point, and then violated it, it’s a trivial lawsuit.
This is done by either refusing to license the patent, or more likely by requesting royalties so large that the products made by any other company could not be cheaper than their own.
There are many examples of patents which had never been implemented in any product before the day when they expired.
In this case, Shin-Etsu would have probably been willing to supply Ga-doped silicon wafers to PV cell manufacturers, but only at prices too high in comparison with the current prices of the PV panels, so such wafers have never seen any significant use.
E.g. satellite manufacturers? Military? Remote installations?
If they sold a consumer version, then they're at risk of losing that business to shuckers (like 'external' HDs that are cheaper than bare units).
End result - good chunk of a generation of aspiring digital artists lost due to unaffordable tools.
Now when their patents finally expired and they have actual competition (xp pen, huion and others) it turns out huge surprise they actually can produce good quality and achievable drawing tablets like the Wacom One.
Progress of arts and sciences indeed!