Semiconductors from stone knives and bear skins

[hombre]

I'm wondering if it's possible to make LEDs, displays and semiconductors using a big ion deposition machine. The machine would be like a very slow and doubtless huge 'replicator' that can manufacture a wide variety of things by spray-painting atoms onto a substrate.

Ideally the device would also dissociate and purify the silicon, metals and other substances as they are used, like a mass spectrometer; it would not have to be frequently dismantled for cleaning or maintenance (unlike a mass spectrometer); and it would not contain anything extraordinarily rare or costly. If anyone takes an interest, how big would such a machine have to be for a yield of, say, a gram of product per day? What would the basic design look like?

Right... I'll get it started myself by calculating the obvious. Lets assume that what we're making is mostly silicon. One gram of Si is 2.7 x10^22 atoms. That's 4.4kC (4400 'coulombs') of Si per day (assuming the atom stream is of silicon atoms with one extra or one missing electron); which comes to just 50mA. That's a lot for an ion gun. It's a lot even for 100 ion guns.

Using many guns in parallel confuses things a bit. Maybe one biggun?

My purpose is to make electronics and reagents without pollution or mining.

 

[Aqueous Id]

Welcome to SciForums. Y Bienvenidos tambien.

I'm not sure about the gun question.

As for power consumption you can take this to Iceland (another post) where they have an abundance of virtually free electricity, and there is grassroots political will to use it responsibly.

You can purify the silicon there and similarly refine the dopants. Presumably you will still need a reactor to do that.

As for feasibility, again, political will, since you will have to take a working process and reverse engineer it into a nontoxic-intensive one. So you would be dispensing with a lot of tried and true steps to the fabrication process, all which have been subjected to continuous quality vs cost reviews.

Gallium Arsenide has a foul enough name - there will be materials that will no longer be available, so that's a limitation of a different scope. Etching still seems problematic as far as mild chemicals are concerned.

I like your idea, rethinking this from ground zero. Even if you were to try to do this in stages, choosing, say, photocell production as a proof of concept, and especially by drawing your electricity from renewable sources, you could even possibly garner the needed political will.

The marriage of environmental engineering with semiconductor fab seems like a great niche.

 

[hombre]

Thanks for answering!

Lets assume power isn't an issue, even if it's a lot.

It would be nice not to have to purify the silicon and other materials first -- at least not beyond a minimum. I was hoping to feed the machine glass and use a mass spectrometer-like means of getting the silicon. Basically take a time-of-flight mass spec, stick a targeting mechanism on the rear end and greatly increase the throughput. It's true that it will consume a ton of power for a little product, and some of the potentials required may be a little extreme; but I just need a convincing theory in place.

The initial costs should also be considered a non-issue. This thing is for my vision of a zero-growth commune, so only the labor required is for maintenance. It doesn't matter how big it is, or ugly or even fragile, it's ok if it needs stabilizing against vibration or if it has to be put in a faraday cage in a pit. It doesn't have to be practical in a marketing sense.

Toxic substances abound of course, but in nanograms (how much does a chip need?), I ignore it. If it's not above the natural background level it's not a pollutant to me. It may not even be possible to scrounge up enough gallium to make use of.

So... I'm talking about a machine that takes (I imagine) fine wires of glass, metal, and quite dilute doping agents and makes devices by vaporizing them and shooting the atom stream onto a substrate with great precision. Basically a bucket of dirt and sand -> ICs. I'm hoping someone out there can give me an idea of what we're looking at. It will take a pretty big ion shooter.