Microprocessors manufactured locally - is it possible?

Can anyone comment on the technical feasibility of decentralized microchip fabrication plants?

Having been in an older generation fab in college, I doubt it would be possible. Current fabs cost billions. I think the best anyone will do in the near (20-300 Years?) future on a desktop at home will be small scale integrated circuits with at most a few tens of transistors. I would love to be proven wrong. Not sure what the point would be though. You would likely replace a $5 part with something that costs hundreds each to make. Microcontrollers benefit from economies of scale to a massive degree. If you want to design your own hardware try FPGA's, or some of the hybrid MCU's with programmable logic attached. (Which might be a great way to improve reprap performance without running the bogomips race)Otherwise I think basking in the cheapness of the commercial microcontrollers is a great idea. Just remember availability (for which cost is a great proxy) is also part of the self replication goal. If availability suffers from advances in self replicatability then I think the big picture is being lost.

Realizing I didn't actually answer the question, here are the requirements.

Si wafers -- You'll just have to buy those
Photo resist/developer-- ditto
Light source -- perhaps a scanned blue laser would work well enough for small scall integrated circuits
Optics -- grind your own ??
Masks -- this is where it starts to get hard
Ion implantation for doping the transistor regions -- good luck with that
Metal vapor deposition -- probably not too hard with a vacuum chamber and a heating coil
Oven for growing SiO2 -- likely expensive but doable
wire bonding machine or some other way to connect to the chip -- reprap 2099?
Way to register layers on top of each other with sub micron accuracy -- if you get the optics and masks worked out

Probably some stuff I missed

Not with the current technological paradigm.
With upcoming technologies, almost certainly - even to the point of total decentralization.

For example ice lithography, which can provide higher integration and reduced equipment cost.



This could even be a sidekick project to metallicarep at which point we would have reached total self replication capabilities.
Awesome stuff but there is a long way ahead, if we even make it.

ElectricMucus, that is an exciting possibility. I've found an article that discusses ice lithography. One concern I have is that the silicon substrate is cooled down to 110K (-163 C). How much energy would that require?

Nanotechnology on ice

I wouldn't be concerned about the amount of energy too much since everything would have to be done in vacuum anyway. The major power drain would be the pump. For cooling we could either try sandwiching peltier elements or cryogenics. The latter would be more complicated, especially if we have to do this in a vacuum.
But as far I know the above configuration is a fairly standard sem configuration. So at best we would start here. And we would need a sem for any high integration work anyway.

The first major step here, as with metallicarap would be how we can construct a vacuum chamber. Everything else could be either scavenged (electron tubes, etc...) or built with regular 3 axis cnc.

I wonder what has been done in the field of self-built sems... They are around long enough that someone should have done that outside academia and industry....

As for the process itself, we wouldn't necessarily need something as far-out as carbon nanotubes. The process, Chemical Vapor deposition can easily be used to grow doped and undoped silicon in of any kind. Combine that with the ability to grow metal contacts we would have a complete system.

The problem is, we would most likely need multiple chambers, to accomplish this.

Well here it is, this fella built himself a sem pretty much from scratch:

[benkrasnow.blogspot.com]

Edited 1 time(s). Last edit at 07/13/2011 01:25PM by ElectricMucus.

Is it safe to assume this could be done on a plastic substrate?

Plastic microchip