<qu1j0t3>
this might be an entirely redundant observation, but can you use the high resolution imagesetters that expose film for making printing plates? dot res is up to 5000 dpi, more commonly 2540 dpi, and they should be pretty cheap now as obsolete gear. they often have pin registration
<diginet>
qu1j0t3: you're here too now?
<qu1j0t3>
mais oui!
<qu1j0t3>
i am like a cockr04ch
<diginet>
heh yeah
<john_cephalopoda>
One inch is 25.4 mm.
<john_cephalopoda>
2540 dpi would be 100 dots per mm.
<john_cephalopoda>
That would be a resolution of 10 µm.
<john_cephalopoda>
Not sure if it would work well. Would have to be tested, I guess.
<qu1j0t3>
yeah i know it's much coarser than commercial cmos, but i'm not sure where you're aiming
<qu1j0t3>
it certainly gets you very far from the masking tape ballpark :)
<john_cephalopoda>
As dpi says, it's dots per inch. Not sure if you can make proper lines with that.
balrog has quit [Quit: Bye]
<john_cephalopoda>
The Intel 4004 (1971) used a 10µm process.
<pie__>
if you use that to create a mask you might be able to increse the resolution with some optic
<pie__>
*optics
<pie__>
idk what quality of lens it would take to say apply 2x magnification to that
<john_cephalopoda>
Some guy used a projector and some lens system to do maskless lithography.
<pie__>
and i guess surface imperfections would be a huge pain in the ass at that scale (idk)
<pie__>
*i mean on the lens
<john_cephalopoda>
He was inspired by Jeri's videos.
<diginet>
john_cephalopoda: he was going off of a paper published back in I wanna say 2005
<diginet>
there's a large body of literature on DLP lithography
<john_cephalopoda>
I like the idea of using direct-write laser lithography, possibly in conjunction with soft lithography.
<diginet>
yes, specifically in microfluidics is where I've seen this
<diginet>
at 10μm I feel like it just makes more sense to have a mask made
<diginet>
they aren't even that expensive at that range
<john_cephalopoda>
Masks take time to be made and shipped though.
<john_cephalopoda>
Doing rapid prototyping isn't easy, also you rely on an external supplier.
<john_cephalopoda>
And worldwide shipping can be very costly, when you don't happen to live close to a supplier.
<john_cephalopoda>
Laser diodes and electric motors, on the other hand, can be bought for little money anywhere, and the gears and mechanical parts could possibly be 3D-printed.
<john_cephalopoda>
The question is, how exact those mechanical parts can work.
<pie__>
diginet, i see you are also everywhere lol
<pie__>
WE ARE EVERYWHERE
<diginet>
yes, I am everywhere
<diginet>
I have yet to find a good solution to the patterning problem
<pie__>
i really need to write an irc plugin that shows people you have cmmon channels with
<diginet>
10μm isn't terrible, but 5 or even 2 would be IMO significantly more useful
<pie__>
im inclined to say lets even get there :P unless it should obviously be possible and we should just jump straight to the next level
<john_cephalopoda>
pie__: I haven't seen a working homemade IC yet, except for the stuff Jeri made.
<pie__>
pretty sure sam made somehting relatively quite impressive
<pie__>
but i think he might have had some $hardware, never got around to looking into it
<john_cephalopoda>
Oh, yeah, Sam has made one with gate size 175μm, smallest features (test pattern) at 2µm.
<qu1j0t3>
john_cephalopoda: it's for graphic repro -- not dots per se, but solid areas
<qu1j0t3>
john_cephalopoda: it's drum scanned, so a raster, but the exposure and photo material and processing are such that you will get fully exposed regions
<qu1j0t3>
john_cephalopoda: the issue you will first hit will be resolution, you won't get 10µm features, but something a bit larger
<qu1j0t3>
perfect if you're only aiming at 50µm or 100µm, though
<qu1j0t3>
& may also be very suitable for further reduction ... these things are quite large
<qu1j0t3>
e.g. 2540dpi @ 550x750mm exposure area, with pin reg
<john_cephalopoda>
100µm is basically what you can get with a cutter knife and Rubylith tape.
<qu1j0t3>
with reduction tho?
<qu1j0t3>
this thing could do it directly
<qu1j0t3>
at a 550x750mm bed size
<qu1j0t3>
(these machines replaced all rubylith processes in graphic arts :)
<john_cephalopoda>
Heh.
<john_cephalopoda>
If one is available for cheap, that could be an option.
<qu1j0t3>
i'll keep an eye out. those specs are from the last type that i personally used
<diginet>
registration/alignment are always going to be what kills you: I've been interested in some variant of scanning probe litho, but the alignment problem seems almost intractable
<pie__>
so what we gotta do is everyone buys a piece of hardware and we just mail samples around lol
<qu1j0t3>
diginet: pin registered, not sure what tolerance they were aiming for for print
<diginet>
I've read a few papers on using a BD drive head to get sub-micron
<qu1j0t3>
diginet: but should be close to the limits of film handling
<diginet>
my main concern is a vacuum-free process
<diginet>
high-vac equipment is expensive and often unreliable, most people aren't cut out to service their own turbos
<pie__>
what about not so high vac
<diginet>
and you really need one with a molecular stage so you can use a diaphragm pump as backing
<pie__>
meh, i need to get around to reading my litho book.
<pie__>
:c
<diginet>
acquiring resist is another rather hard problem
<diginet>
almost no one will sell to individuals
<diginet>
I've been working on a water developable resist that can be made from relatively easy to acquire precursors
<diginet>
with a little bit of chemistry finesse
<qu1j0t3>
diginet: nice work
<pie__>
i keep wanting to write something like openscad but better and integrable with other tooling
<pie__>
eventually... :P
kristianpaul has quit [Ping timeout: 252 seconds]
<john_cephalopoda>
pie__: International shipping is cheap, so having a decentralized manufacturing process would be possible.
<pie__>
i was joking, could that really work
<john_cephalopoda>
I'm located in Germany. Shipping 500 g of stuff to the USA for 3.70€.
<pie__>
i guess if you batch things and have working individual process steps but...
<pie__>
also need diagnostic tools to make sure your incoming batch is usable?
<pie__>
may or may not have to redo cleaning
<pie__>
packaging for shipping...?
<pie__>
i mean it could work, idk
<john_cephalopoda>
Depends on which products you want to ship.
kristianpaul has joined #homecmos
<pie__>
there seems to be a few of us in EU but all the really well set up guys i know are in us
<john_cephalopoda>
For most things you will have to do diagnostics before every step anyway.
<pie__>
makes sense
<pie__>
what i really meant to say is there would be a lot of diagnostic hardware duplication?
<pie__>
not that theres anyone that doesnt want more diagnostic hardware ;P
kristianpaul has quit [Ping timeout: 240 seconds]
<pie__>
that would be pretty awesome though, to have a distributed project like that actually work
<pie__>
sounds super inefficient though :X
<john_cephalopoda>
Hmm, let's take a look at the steps.
[X-Scale] has joined #homecmos
<john_cephalopoda>
I see three steps that can be done separately: Mask creation, CMOS fabrication and the packaging/BEOL.
X-Scale has quit [Ping timeout: 272 seconds]
[X-Scale] is now known as X-Scale
<pie__>
dont you need to loop mask and litho and cleaning?
<john_cephalopoda>
What do you mean by "loop"?
<pie__>
well, im being extremely imprecise
<pie__>
oh i see what you're saying
<pie__>
well yeah on that level i guess it could work
<john_cephalopoda>
To create masks, you need special printers or a process that creates soft litho stamps.
<john_cephalopoda>
Masks can be checked with a microscope and without too much work.
<john_cephalopoda>
For the actual fabrication checks, you will need a microscope anyway.
<john_cephalopoda>
Welp.
<john_cephalopoda>
Right now it doesn't really matter since nobody has anything set up yet :D
<john_cephalopoda>
An interesting thing to find out and document soon is where to order photo masks, wafers, etching supplies, dopants, etc in the EU.
<john_cephalopoda>
Everything I've seen by now is pretty US-centered and probably costs quite a lot to ship to the EU.
<john_cephalopoda>
Night
john_cephalopoda has quit [Quit: Trees can see into your soul. They lurk everywhere.]
<qu1j0t3>
not wrong
Bike has quit [Quit: Lost terminal]
azonenberg_work has quit [Ping timeout: 245 seconds]
azonenberg_work has joined #homecmos
rqou has quit [Remote host closed the connection]
rqou has joined #homecmos
pie___ has joined #homecmos
pie__ has quit [Ping timeout: 246 seconds]
sn00n has quit [Ping timeout: 246 seconds]
john_cephalopoda has joined #homecmos
<john_cephalopoda>
Morning
<pie___>
o/
azonenberg_work has quit [Ping timeout: 272 seconds]