This is probably my most audacious and stupid proposal, but allow me to explain.
This proposal is about building a CMOS processing all-in-one tool at home, so not Grant Thompson's aluminum melter.
Modern CMOS processes are based on a collaboration of multiple devices, there are special equipment for lithography/spinning/baking/developing, special equipment for etching, special equipment for implanting, and special equipment for deposition.
In between, there are cleaning, planarization and handling equipment, and all of them housed in a clean room.
My proposal is to build a single ion beam (E field lens and linear accelerator) based equipment that can scan patterns to etch, deposit and implant, using both additive and subtractive methods, driven by a configurable ion source (mass spectrometer and electric field ionization source) that spits noble gas ions, dopant ions and protons.
A second beam of scanning electrons gives real time image feedback of the process to control focus of the ion beam as well as to profile depth and to do close loop compensation of deflection caused by mechanical vibration and H field. With controlled e beam irradiation, antenna effect can also be negated, thus allowing designs that were not possible previously.
When massive quantity of material is needed for deposition, the wafer can be heated and gas being introduced for CVD, the excess can be etched by focused (with adjustable focal length) ion beam in a DRIE atmosphere.
The wafer stays in place from cleaned and mounted virgin wafer to finished wafer waiting for dicing, so alignment requirement is minimum, plus the SEM can provide image feedback for fine adjustment to compensate contraction/expansion, vibration and other sources of misalignment.
The only other equipment needed is a sputter machine, which deposits Ti/Ni/Au triple layer to make die bump pads, with a relatively low resolution stencil. Sputtering is moved away due to its messiness, I think everyone worked in a clean room knows how messy this process is.
The finished product can then be diced either in house or sent to a packaging house to do dicing, testing, bumping and C4 with encapsulation.
---------------------------
Do you guys think this is a viable solution that will allow the more advanced hobbyists to make 0.35um chips at home for less than, say, $50k investment?
Since everything is scanning based, rather than lithography based, this is no where close to production throughput, but just for prototyping niche ideas, I think this is a viable solution.