Re: Silicon Machining

> > From: "Kirk W. Fraser" <kirk@...>
> > Has anyone considered making their own desktop chip foundry? ...
>
> From: Jon Elson <jmelson@...>
> There are a couple of problems with his idea. First, you are not likely
> to be able to make a CPU, due to the cleanliness problems...
>
> 2nd, modern chips are made with sub-micron feature sizes, so any
> source of red light is too long a wavelength...

I like Kirk's question. There's lots of opportunity for DIY chips.
Not that we could ever compete with Intel using our Chinese Mill-Drills
adapted into patterning tools, but we really don't compete with
Hardinge or Bridgeport when we make our own CNC, either.

My job is IC engineer. My employer has two wafer fabs on site. We
have the equipment to make chips, and make millions per day. Yet we
don't compete with Intel, either. The leading edge companies are
working with 300 mm wafers and 0.16 micron feature sizes. We aren't.
We have a number of different production processes today. Our older
processes are still running and generating great products, yet the
finest line widths on our oldest processes are 8 microns (0.3 mils).
That kind of feature size is within the grasp of the HSM. It's not
easy, but it's within reach.

A good desktop printer is capable of 1200 DPI, which is better than
1 mil spot size. So that kind of tool should be good enough to
draw 1 mil lines on wafers. 2400 DPI printers are still exotic,
but they're available, too. So the line width from lasers is within
reach.

When we bring up a new process or a new fab, our first steps are to
process batches of silicon wafers through individual operations. We
won't attempt to make transistors until we have a process that lets
us grow pure oxide. Then we develop the process to sputter down a
layer of metal. Then we develop the process to selectively implant
impurities into the wafer. Finally, we make wafers sprinkled with
every different size transistor and other component imaginable.
Then we evaluate the components and refine the subtle nuances of the
process before we ever get to making flip-flops, op amps, or any
larger functional blocks.

If you're going to experiment making your own semiconductors, you'd
do well to follow this same strategy. Don't attempt to machine a
Quorn before you've learned to knurl and thread on the lathe. Alas,
semiconductor manufacturing has a wide range of operations, so you
need to either set your sites low or expect to spend a long time
developing the skills and the tool set.

But it can be done in a basement shop. I've seen fabs that look
dirtier than many common metal shops. And they make good semiconductors
in those fabs. Not 256 meg RAMs and not Pentium III processors, but
diodes, transistors, and SCRs.

It won't be cheap and it will involve significant safety questions.
Raw silicon wafers are only sold in large quantities. But you could
get out-of-spec wafers for a song and they should be adequate for
experiments. The safety issues need to be addressed one at a time.
You can't do it without chemicals. Even if you use lasers and plasmas,
you're going to need some chemicals.
--
Bob Neidorff