Hi,

- spin coating

I finally tried to build a makeshift spin coater the way Sam Zeloof did, by mounting a plastic can on top of a CPU fan (actually a case fan from another device I had lying around). I can't say this is the way to go... the fan is way too weak to spin the can at a meaningful speed. It is also very hard to center the can exactly, so it wobbles a lot and that probably reduces the speed even more. Mounting a die without a container would allow the fan to spin faster, but would likely clog the fan with chemicals in no time. I'll next try my original idea, to mount the can to an axle that I'll stick into a power drill.

One thing I did find out is that the coating liquid should be evenly applied before spinning. It will NOT spread to the whole surface by itself, but rather run to a single direction in a concentrated stream, probably due to surface tension.

- chemicals

I did find this resist which is meant for PCB etching, so I don't know yet whether it has some kind of "grain size" that limits the resolution. The datasheet explicitly states that it can take up to 40% hydrofluoric acid, though, and be used for glass etching (so I hope it will actually stick to glass). It is also liquid until soft-baking, so spin-coating should be possible.

WRT hydrofluoric acid, there is something called "glass etching cream", the best-known brand being "Armour Etch". This contains fluorides, i.e. HF salts, is *somewhat* safer than HF itself (less fumes I think, but don't get it on your skin), and can be ordered by private individuals. The downside is that I could not find any information on whether it contains an abrasive.

- photolithography

Sam Zeloof used a modified DLP projector for his litho setup. He either used magic to do that or chose exactly the right projector. I took apart a used DLP and the UV filter is the only thing that could be taken out without destroying the projector. The color wheel, and even more so the optics, are buried under tons of other stuff, most of which is glued together. Also, the optics path itself is mostly glued. So I think David's plan of taking the DMD chip and building his own optics setup is a far better idea than modding a projector.

To me as a layman in these things, stiching a larger image together from multiple images with a stepper sounds more logical than using multiple DMD chips. This XY table is cheap and has stepper motors that can position the table in increments of 5 microns. I have actually wondered if a first prototype should use a DMD at all, or just focus a single UV LED and "draw" by moving the XY table.

- strategy

I also thought a bit about who would be interested in making your own chips. High-volume production has far better, established processes, and low-volume production would use an FPGA or other off-the-shelf parts. Silicon testing before production isn't useful when the process is different. This leaves education and hobbyists. In other words, a free silicon process must be sustainable based solely on education and hobbyists, otherwise it won't stand a chance.

There is another problem that goes in the same direction: a "democratized" chip making process should not rely on actors like INL, because there are too few of them -- less than 10 in the whole of Europe. Even if INL agrees to host the LibreSilicon process, a single decision by INL that chip making isn't interesting anymore can shut the whole thing down.

I'm not particularly enthusiatic about the "glove box" idea either because I don't think it is practical, nor doable in a useful price range.

The usual answer to the "expensive tools" problem is shared labs, usually as "fab labs", "makerspaces" or similar. I think this is a way in which chip making could actually work, especially if it works on dual-use tools that benefit a makerspace in other ways, like an XY table, microscope etc. I wanted to contact a local makerspace in my town, but unfortunately Germany just went into full Covid lockdown, so it's going to take a while until that.

Greetings,

Martin