Desktop Semiconductor Foundry

A desktop semiconductor foundry is a hypothetical custom manufacturing device small enough to fit on a desk, and would make semiconductor parts (power transistors and/or integrated circuits). Such a device might also include the ability to fabricate electrical components (resistors, capacitors, and inductors with values different than what's possible with semiconductors).

People need power transistors and integrated circuits to build other custom manufacturing devices. As of 2007, such semiconductor parts are fabricated in semiconductor fabrication plants, which are large buildings filled with large, expensive machines.

From a website about a new semiconductor fabrication plant:


 * "With equipment, the plant will cost a total of $3 billion." ]

So what are the prospects for home based integrated circuit design and manufacture?
 * Already semi-custom designs can be fabricated at home, using FPGAs and OTP devices. FPGAs are 'Field Programmable Gate Arrays'. Like downloading data to USB flash ROM, data can be used to program the circuits, setting up the logic connections between different elements. Such devices cost a few dollars. 'One Time Programmable' devices are even cheaper.
 * Some calculation also shows that the density of data spots on a typical CD ROM is approaching the density of earlier generations of IC [Philips]. A laser based scanning device based on CD ROM technology would be slower and lower resolution than modern silicon fabrication plants, however it would also be able to use mature optical and resist technology.
 * While the RepRap project is still mostly a thermoplastic extruding machine, work is underway to enable it to print its own and arbitrary electronics, either through extruding conductive and semiconductive poymers or by inkjetting appropriate nanoparticles [RepRap Wiki].
 * Another approach to a desktop semiconductor foundry is 'Spray on Silicon' which could be printed using technology similar to that of an inkjet printer. However this is considerably lower resolution than would be required for the density of circuitry we are accustomed to.
 * Home made scanning tunneling microscopes already exist; modyfying one so it can do dip pen nanolithography should be possible, although not high enough speed to create modern chips in bulk. It could be used to create stencils for soft lithography or masks for photolithography.

Analysis
The expensive clean rooms that keep dust out are not needed for burning a CD-ROM, because the data surface is sealed by a transparent plastic layer. The same principle could be used for desktop IC manufacture. Other methods of manufacture that could be achieved at home include inkjetting nanoparticles and extruding thermoplastics.

While these technologies would enable macro-scale fabrication of application-specific integrated circuits, miniaturization will prove to be a challenge. Initially, home fabrication will likely be used for simple components and prototyping. This would drastically reduce the expense generally associated with developing application-specific integrated circuits, accelerating the existing trend towards system on one chip even further.