Fuggedaboudit.
I once had the pleasure of working on an IC design in as
close to "home fab" as you could get - a university fab
line, ca 1982. The equipment occupied more than a garage
and we had sweet 2" wafers to work with. The equipment
was laughable by today's production standards, we were
perhaps the last class to actually cut rubylith. The process
was NMOS-only (E/D) because there was only one dep
source. The closest we got was a partially functional
chip (they all failed in different ways). And this was under
a hundred transistors all told.
You have many issues ahead of you such as cleanliness
and contamination (go directly to defect density yield
and short-term reliability), the noxious to toxic nature of
the specialty gasses (thus, the low likelihood that you will
be able to get them delivered to you or for that matter,
survive your attempt at plumbing) and solvents, a need
for much ancillary equipment (cleaning baths, photoresist
expose, develop, strip, aligners, and have you looked at
mask costs (and can you get masks made still, for any wafer
diameter that has afffordable equipment on the aftermarket
(with useful life remaining - good luck on the spares, for
gear whose manufacturer went belly-up a couple of decades
back))?
But on the plus side, I have seen companies make deals
to take over trailing-edge fabs for zero dollars so long as
they promise to take on the toxic waste liabilities from the
previous owner, and then sell it off later at no net loss and
no net liability. Of course the lawyers got paid.
If you are really wanting to stir the cauldron, go to work
in a fab. If you are really wanting to design and hold in
your hand something you invented, I suggest finding folks
who jump on multiproject wafer runs, and convince them
to let you put your little piece in a corner of it (for free,
of course?). The fab wants its money, but maybe someone
at a university has a square millimeter of white space the
week before tapeout and it costs them nothing, then.
Now, there are semiconductors and there are semiconductors.
Maybe you could have fun with things less demanding of
purity and quality than a silicon IC. For example there looks
to be plenty left to do in II-VI semiconductors for PV and
so on - Iron sulfide, copper oxide, etc - and what you need
there is more like "garage chemistry" and an idea of crystal
growth and doping, contact and metallization, all of it maybe
doable with wet etches and evaporative deposition (we had
one ancient flash gold bell jar setup at one place I worked,
invaluable in its day (before FIB-writing of tungsten was
de rigeur) for laying down a post-fab interconnection).