Home Chip Fab
sam.zeloof.xyz
sam.zeloof.xyz
[1] https://www.minimalfab.com/en/ [2] https://www.youtube.com/watch?v=WsOVbmfYxoM
[1] https://www.yokogawa.com/yjp/solutions/solutions/minimal-fab...
It seems like this fab is a demonstrator, and is planned for sale. This makes me wonder if/when this will be available for maker style setups, at which process nodes & price per piece. Would be nice to be able to design some fully custom ip, have it tested, bonded and packaged, and optionally soldered, like you can have it with pcb services today.
One can dream, yes?
[1] https://asia.nikkei.com/Business/Biotechnology/Minimal-fab-t...
According to wikipedia the following CPUs were built with that:
[2] https://en.wikipedia.org/wiki/350_nanometer
Which i think of as more than sufficient to finally being able to implement something like SCED, WAM, CHERI, Applecore, whatever in whichever way one is able to wrap his brains around it. Asynchronous, fully static?
https://en.wikipedia.org/wiki/Mead_%26_Conway_revolution (with MEMS(for sensing)) from scratch?
I want it all! I want it NOW!
[1] https://en.wikipedia.org/wiki/R4200
under R4300i which mentions 45mm² for the die.
Also mentioned here [2] https://bits-chips.nl/artikel/small-series-of-chips-profitab...
are 0.25-micron to be released this year, with 190nm and smaller on the roadmap.
Which leads us to [3] https://en.wikipedia.org/wiki/250_nanometer at least.
Maybe not comparable in die size for all the chips mentioned there, i don't care so much, because i don't want to clone or emulate them. I want to go simpler. Rebranch from the 70ies so to speak, to take all the roads not taken since then. Just to see what's there :-)
0: The main reason to use a rectangular die is that they tesselate better to fit many dies per wafer.
What is missing of course is what their feature sizes are and the part at the end where you cut the die and package it. It is those things that would define the set of things you could put on a single chip.
[1] https://bits-chips.nl/artikel/small-series-of-chips-profitab...
under "E-Beam" for 0.25-micron node and roadmap.
So your 12.5mm wafer can have a 8.8 x 8.8 mm square inside of it, or 78.125 mm^2. If I did the math right that is on the order of 156M transistors given a 4t ram cell that is about 39 million bits of RAM. So basically a pretty useful amount of space for "jelly bean" type applications. A synchronized fab line with a median processing time of 1 minute can produce 60 dice per hour. Assuming a physical plant cost of $8M US (that is "several million Euros + the office space to hold it) and a depreciation cycle of 12 years that is about $500 / day for the machinery we can add another $500 / day for staff + electricity, figuring 8 hour days, that's $125/hour to operate for 60 chips is a bit more than $2/dice.
Well the pencil math works (with all of those assumptions) but even assuming its off by an order of magnitude, $20/dice isn't a deal breaker for your own custom chip that does your special thing. You'll also notice that the 50 weeks a year 40 hours a week assumption. I'm guessing you can get better utilization than that which would offset your depreciation costs.
[1] www.lip6.fr/public/2017-03-17_Shimizu.pdf
Mentioned are workflows, used tools, intendend audience and goals, estimated prices, nda-freeness of spice models and design rule check, open source, open-cores, github, and so on.
If they really make this widely available, then my mind is blown...
Thanks for this info.
I see a lot of people say this, but I rarely see any actionable advice.
How does your average residential person find and dispose of chemicals like this? Seems like most people end up pouring them down the drain simply because they don't know how to actually find a better means of disposing them.
Ultimately if you want to use chemicals that have the potential to poison our shared environment you need to be willing to do the legwork to be confident that you are doing things responsibly.
If you're planning out your own home fab, you're researching a bunch of other things and giving serious thought to a lot of nontrivial problems. Choosing to ignore waste disposal is irresponsible, and you deserve any legal hassle you run in to.
https://www.sccgov.org/sites/rwr/hhw/Pages/What-is-Household...
Other states probably have similar things?
Very much worth researching before you get into a hobby where you start accumulating hazmats!
How's this? If you don't know how to safely dispose of the toxic chemicals used in chip fabrication don't fabricate chips at home.
If you have industrial quantities of chemicals (more than a gallon or so), you need to call the relevant entities.
If you are lucky enough to have a hazardous waste disposal locally, obviously use that.
If you have stuff that doesn't break down well in water (cooking grease, for example), pouring that down the drain is always a recipe for trouble. You're simply going to clog your pipes. You need to dispose of that properly. There is a reason why restaurants have grease traps. Normally, residential quantities of this stuff can be placed in normal trash.
You can try to dispose of motor oil at gas stations, service stations, etc., but a lot of places won't take oil from end consumers anymore as it may be contaminated and their recycler will charge them. I actually had a very difficult time disposing of motor oil about 15 years ago. (I don't do my own oil changes anymore for this reason). I got told by the local enforcement "At the end of the day, dishwashing liquid and pour it down the toilet and the sewage treatment plant will chew it up the rest of the way." Obviously if everybody does this, it's a problem, but if it's really a one-off, it's okay-ish.
If, however, your stuff is soluble in water and you have a relatively small amount of it and your waste goes to a sewage treatment plant, pouring it down the drain while diluting it with a lot of water is often your only choice (be careful--solvents and acids can produce fairly noxious vapors even when diluted heavily) Quite often industrial disposal sites simply will not take small quantities of waste from individuals as there are liability issues involved.
Now, you may not like what even a dilute solution does to your plumbing, but that's a different issue. If your sewage doesn't go to a sewage treatment plant, but instead goes to something like a septic system, then you probably don't want to do this.
DO NOT POUR STUFF DOWN STORM DRAINS. Those normally do NOT go to sewage treatment plants (there are exceptions--but they are rare) and, as such, are a really quick way to contaminate the environment.
How do you dispose electronics? Batteries? Oil? Tires? Cars? Furniture? construction materials? ...
You go to the webpage of your local garbage disposal authority, and read their FAQ, which typically contains where all their disposal centers are, their addresses, opening hours, etc. and what can you dispose on each one.
If what you want to dispose is not listed anywhere, you call them and ask them.
In my country if you want to buy these types of chemicals, you need to ask a company for a price, and the company will ask: who are you? what do you want them for? what's your process for the chemicals? Etc. If you fail to answer any of the questions, they are obligated to report that a "sketchy" party tried to buy some chemicals from them. That might get you a visit from the police, asking even more questions.
That's balancing your freedom to do whatever you want with chemicals with my freedom to enjoy a world that hasn't been polluted by idiots that didn't know what they were doing.
+1
Please develop a workable waste disposal strategy BEFORE thinking about cool disruptive home brew nanofab MEMS/circuit hacks.
The first step is easy: neutralize them with any base, such as baking soda. You don't need precision, just add a lot of it, and make sure the pH is close to 7 (or beyond, baking soda is not corrosive) at the end. After this step, the solution is no longer acidic or corrosive and much safer to handle.
Unfortunately, the next step is tricky. The solution is safe, but it still contains a lot of Cu+ ions, which is a heavy metal pollutant and poisonous.
On the other hand, if your sulfuric or hydrochloric acid is unused for etching anything (clean without those nasty ions), you are good to go, just dump them in the sink after you've neutralized them (test with pH paper) it's perfectly safe.
The solvents can be broken down to things that are compatible with the sewer system.
The hard part to deal with is the material you dissolved with the solvent.
Or is it something you can just ignore, because at the end of the day everything goes to a wastewater treatment plant, and your volume/concentration is too low to be considered hazardous, and actually not more harmful than the wastewater of commercial chemical cleaners, and well within the wastewater processing capabilities for small volumes?
Can anyone give an authoritative answer to this question?
I think the biggest problems for chip fabrication waste (not circuit board etching) are HF and nonpolar organic solvents. I'd think neutralizing HF with chalk would yield fluorspar, which is resistant to weathering even over geological timescales. But again I don't know what the official answer is. Maybe dumping fluorspar in your yard will get you arrested.
From what I understand the last thing you want to experience is an explosion of hydrofluoric acid.
The internet seems to suggest lime (->fluorspar) or soda lye (->NaF, used for fluoridating water) to neutralize HF, but that's two other substances you wouldn't want raining on your head... be careful out there.
I suggested fluorspar rather than NaF because a backyard full of fluorspar is a pretty rock garden, while a backyard full of NaF is a toxic waste dump.
Don't mix this stuff in dilution in an enclosed space either as the H2 gas can create another explosion hazard.
As for hydrogen, neutralizing acids with bases doesn't generally produce it, but ventilation is still a good idea.
https://www.mgchemicals.com/products/prototyping-and-circuit...
> The solution must not be put down the drain because of residual copper ions left in it. To make it safe for disposal, you can add sodium carbonate (washing soda) or sodium hydroxide to it to neutralize it, until the pH value goes up to between 7.0 and 8.0, testing it with indicator paper. Copper will be deposited as a sludge. Allow the sludge to settle, pour off the liquid, further dilute it with water and then it can be poured down the drain. Collect the sludge in plastic bags and dispose of it as required by your local waste authority.
Cu though is an aquatic herbicide (used to rid decorative ponds of all plant life including algae) and can mess up things.
> at the end of the day everything goes to a wastewater treatment plant
Some of these things, especially the photosensitive chemicals in particular ammonium dichromate will annihilate the wastewater treatment plant. Experienced this problem in an industrial context once, resulting in tens of millions of dollars in damages to taxpayers and a giant fiasco. As a result our privileges to use the municipal sewage system were withdrawn and we had to spend tens of millions in evaporation towers since no water could leave our facility ever again.
I read through the OP - he's making good choices and doing it much as we did fifty years ago. I would have killed to get that maskless photolith setup, though.
For a small fab (three furnace tubes, two fume hoods), the main issue is acids from photoresist ashing, wafer cleaning, tube etch cleaning, and metal etching. These are easily neutralized with a vat of marble chips built into the plumbing under the fume hoods.
If you are doing this, avoid epitaxy, which involves arsine and phosphine. I would say never attempt those! You can buy epi wafers. Either use spin-on dopants like he did, or use oxidized boron nitride wafers and small amounts of phosphorous oxychloride like we did.
The effluent from a plating facility or a dry cleaning establishment would be far more concerning.
Our furnace and photolith operation fit in a 20 x 30 ft room. It was one of the leading operations of its kind in the world at that time.
Seriously impressive though.
Even if you're the next Albert Einstein you will never be able to even attempt a project like that unless your parents are rich or you can find a generous benefactor at some institution who allows you to use their equipment.
https://www.youtube.com/watch?v=HW5Fvk8FNOQ
BBC Horizon: 1977-1978 Now The Chips Are Down
https://news.ycombinator.com/item?id=16955549
Edit: I should mention that one can make a rectifier using household chemicals. Caution is still warranted of course.
http://www.hpfriedrichs.com/radioroom/cu-diode/rr-cu-diode.h...
EDIT: This is the pic I was asking about: http://sam.zeloof.xyz/wp-content/uploads/2018/04/na_1.png
If so, that's not an SEM issue, that's just what it looks like. Photo exposure by it's very nature doesn't edge things 100% properly, diffraction or tiny misalignments causes stuff to end up being 'soft'.
Also, keep in mind that was from 1997. Fab tech has gotten way better; although we don't have a scale reference, I imagine if you were to make the same device on modern tech everything would look a lot 'sharper'. You still can't get away from some softness though.
Yes to both of these questions. "Soft" is probably a better way to articulate this, thanks. I'm not following the explanation of this being caused by photography however. Is this specific to photography at this scale?
>"I imagine if you were to make the same device on modern tech everything would look a lot 'sharper."
I don't have an example handy however I have seen more recent micrographs and they display the same "softness" and lack of true 90 degree angles etc.
Yes. A smaller scale makes it worse, a larger scale makes it better.
If you reduce things enough, everything will become so blurred that you won't be able to see anything (that's why we can't photograph atoms using visible light), and if you make things large enough your camera will become the bottleneck so a very small amount of blur is added to any object, whatever the size.
As for 'modern tech' vs. old tech, I'm referring to if you made the same size silicon structure on newer tech. When you get down to 7nm or whatever, the same problems show up (or even new ones) because the sizes get smaller. I wouldn't be surprised if the image from 1997 is 180nm or larger nodes.
Also, unlike the other replier, I don't think there's any meaningful diffraction (causing softness) by the actual SEM images... they don't use glass lenses like traditional cameras.
The resist has a finite thickness (often comparable to the size of the feature you are trying to make).
The developer solution works downwards through the exposed resist, but also outwards at the edges (where you will have some proximity effect in the exposure).
Deposited metal or insulator sometimes sticks to both the bottom of the pit and the top of the resist without a clean break, or contains chunks or grains that make the coating uneven.
Wet chemical etchants don't etch straight down, and often create rounded corners in cross-section.
Every aspect of this has decades of optimization behind it, and its success in any one fabrication run is vulnerable to a frustrating number of variables. If there's an SEM cross-section of a device in a paper, it's likely because fabricating it was an accomplishment, and/or the design is in some way new (not ruling out other reasons of course: there's also "we have an SEM and not a lot else to put in the paper").
TL;DR: I agree. These things are very small and if we could make the corners sharper, that would remove one major barrier to making them even smaller. And then they wouldn't look so tidy anymore.
I didn't quite understand what you meant here. Could you elaborate on this? What are some of those limits?
>"These things are very small and if we could make the corners sharper, that would remove one major barrier to making them even smaller. And then they wouldn't look so tidy anymore."
Aren't these contradictory? Did you mean to say "And then they would look so tidy"?
A few factors:
There's always some finite transition region between the fully-exposed area and the fully-unexposed area of the polymer resist area. With optical lithography, to get the smallest features possible, we're basically trying to make a shadow edge that's perfectly sharp and vertical through the thickness of the resist, by pressing a transparent mask with a pattern of thin metal on it right against the resist.
The thicker the resist, the blurrier the shadow down at the substrate. This blur takes the form of partially-exposed resist, which will develop faster than unexposed resist but more slowly than exposed resist.
With electron-beam lithography, there's the focus and control of the e-beam writer, but the sharpness and precision of the exposed pattern also depend on electron scattering and charging effects in the sample being exposed. Again, thicker resist makes the pattern less sharp as it needs a higher electron dose and will suffer more scattering.
We use a chemical to dissolve the exposed resist (there are processes where it's the unexposed resist that dissolves instead, but that's a tangent here). It takes time to dissolve the resist right down to the substrate, and as it's doing so, it's attacking the edge of the pattern too, albeit more slowly, making corners rounder and changing the vertical edge profile of the resist.
Say we get a decent pattern in our developed resist, and the next step is a liquid chemical etch. The etchant will dissolve the substrate at a certain rate. But it won't only etch straight down. If it etches equally in all directions, it will round the corners and etch under the resist, which makes the etched-away areas bigger than the pattern in the resist, with the rounded vertical profiles we often see in device cross-sectional images. This is assuming perfect resist adhesion.
A really tiny wet-etched pit has to start with a really, really tiny resist pattern and may well consist mostly of rounded undercut.
Fluid dynamics and chemistry in confined spaces often mean some areas etch faster than others and lines can get wobbly (this can be a factor in development too).
If, instead of etching, we want to deposit material like insulator or gate material, the developed edge of the resist has to have a profile that the deposited film cannot just run continuously up, or when we try to "lift it off" in the unwanted areas by dissolving the remaining resist, the patterned film will tear unevenly (at best). An overhanging resist profile is best for this. It's very difficult to lift off a deposited film that's thicker than the resist, so in general if you're depositing a film, you're starting with the challenge of a thicker resist layer.
Now imagine you've done a couple of steps on the devices already. You have perhaps some etched features, ohmic contacts, and an insulator, and now it's time to pattern some gates on top. The devices are now 3D. The resist is thicker in some places than others. The insulator tends to charge up in the electron beam. It's all the same principles again but a little more complex.
There is a plethora of techniques and chemistries to mitigate the issues encountered at every step, and you mix and match the things you are allowed to do and can afford to do and the skill level of available personnel to get the best result you can. The more automated you can get it, the better, because it's crazy how tightly every environmental variable has to be controlled to make a complex process reproducible, but in academic labs you still have students with gloves and tweezers and beakers and stopwatches, tweaking their process and device design to make something new happen.
BUT, the image is also made worse by how the image is captured. The picture you are looking at is a FIB (Focused Ion Beam) cut cross-section of a transistor which is then imaged by a SEM (scanning electron microscope). Probably in one of their fancy dual source beam devices. Depending on how you cut with the FIB, you can also push material downwards into lower layers and create a waterfall effect that will blur edges. It's can be very noticeable in rookie cuts but doesn't look like a major impact here.
Imaging with the SEM itself can also cause distortion and blurring as trapped charge accumulates on the surface and deflects the electron beam slightly. You can see this where the edges get increasingly bright because charges gather and become trapped at the boundary of the conductive and insulating layers. The image isn't sharp enough for me to tell if that is poly-silicon or aluminum contact traces.
Thanks for the detailed response. Might you have a link to one of those other posts or some other resources that talks about how these processes tend to produce rounded structures?
Given the volume and mass payload capabilities of the upcoming Starship/Superheavy, cheaply lifting bulk quantities of lead into space may actually be a reasonable proposal in the next five years rather than ultra-expensive pie-in-the-sky fantasy as it was during the age of disposable rockets.
I was lucky to have this at my university. We had a professor who build a complete mini fab, mostly from industry donations of discarded equipment. Our semester built a complete and functioning surface acoustic wave (SAW) filter from the wafer to the bonded and housed chip. One of the best practical courses I've ever had.
Thanks.
For one, this person seems to have some very advanced knowledge of some very advanced processes - I am not even sure where one would gain such knowledge from easily, though they do seem to have access to MIT in some manner. Regardless of how it was gained, this knowledge of some of the very basic processes and advanced tools involved would be foundational to this entire enterprise (and it seems this person went to the trouble to build up to this - but what foundation of knowledge they had prior to that point seem unknown?).
Secondly, they appear to have something more than just a simple "2-car garage" - unless the camera perspective is really skewed. The building they are in looks to be like a large steel fabricated unit, 40'x 20'x 12' - my numbers might be off but it doesn't look like a simple garage or workshop, but rather something you'd either pay a lot of money for to have erected on your (likely sizeable) property, or its space at a leased business/industrial complex.
Third, from what I gather, much of their equipment is resold industrial surplus off ebay (and probably other places); I know of one company, local to me (Phoenix), which sells online via ebay such equipment (Equipment Exchange) - assuming they are still in business. While what they sold was inexpensive - even at "pennies on the dollar" you are not talking small amounts of money, because some of that equipment is anything but inexpensive brand-new. I'm sure there are other vendors as well for this kind of equipment, and if you know what you want and need, you can probably strike some good bargains, but I still can't imagine this being budget friendly for most people.
This all adds up to what seems to be a "hobby" that would require both extreme dedication (time, energy, and knowledge acquisition) and considerable monetary resources to implement. Not something that just anyone can do - it's not a "maker friendly" endeavour. Maybe it can be miniaturized and simplified now that it's been "proven" to be possible, which is a great thing in itself (after all, it was more or less done with 3D printing and home CNC). Though I am not sure how far that can be taken - then again, people have fabricated some quite complex stuff before...
So all in all - this is amazing work, and does show the possibility. But temper that with the reality of the situation: This person has some very specialized knowledge gathered over who knows what kind of timeframe, and the resources to purchase and bring together all of these tools and parts to do this kind of thing. Even with all that, they still had and have a bunch of trial and error. I'm amazed it could be done at all in a nominally "DIY" manner. Congratulations to that!
If anybody would like to get into this kind of thing I highly recommend that book, it's very approachable for somebody coming from computing without a particular semiconductor/chemistry/material science background.
Kirt R. Williams, Kishan Gupta, Matthew Wasilik - Etch Rates for Micromachining Processing—Part II
Mark R. Jackson - Effects of Radio Frequency Power and Sulfur Hexafluoride Flowrate on Etch Rate of Silicon Dioxide
S. A. Moshkalyov, C. Reyes-Betanzo, R.C. Teixeira, I. Doi, M.B. Zakia, J.A. Diniz, J. Swart - Etching of Polycrystalline Silicon in SF6 Containing Plasmas
I.J. Kima, H.K. Moona, J.H. Leea, N.E. Leea, J.W. Jungc, S.H. Cho - Silicon nitride etch characteristics in SF6/O2 and C3F6O/O2 plasmas and evaluation of their global warming effects
http://microchem.com/pdf/PMMA_Data_Sheet.pdf
These for further general process design/example values:
http://ww2.che.ufl.edu/unit-ops-lab/experiments/semiconducto...
These are by no means the be all and end all, they were just useful for the parts of the process I've looked at in detail so far and there are often several options at every step for chemicals and approaches, these were suited to what I was aiming for.
Well, he hadn't graduated high school yet. Your points about the expense are probably sound, unless he was doing some serious wheeling and dealing (which is certainly possible, some high schoolers flip sports cars and make money that way, etc) but a large part of that would be an artifact of being one of the first to attempt this sort of thing at this scale. Modern manufacturing often doesn't scale down very well. Such steps are necessary, however, to improve the level of capability of the hobby community in general. The next project will probably be done cheaper, learning where there is flex in the process, etc.
OTOH, considering a hypothetical group of dedicated enthusiasts who are willing to pool resources / funds / knowledge /etc. and work on something like this collectively, I'd just about be willing to wager than any moderately urban area in a "1st world" country has the raw potential to have a "garage chip fab" show up. Especially areas near one or more research universities and with a heavy tech presence (and therefore lots of technically adept people in the local population).
Of course not everybody will bother, because arguably there isn't much point other than education and a sense of self-satisfaction, given this bit: "Update 7/8/19: FET gate length (feature size) reduced to <5µm, bringing this project to be state-of-the-art in about 1975".
Anyway, this seems like a pretty cool demo at worst, and "hats off" to Sam for pulling it off!
and their bank account...
which is not to diminish his accomplishments. It's impressive as hell what he has managed to do, but it's kinda hard to overlook that detail.
A project like this would be impossible for the vast majority of the population of the richest country in the world, even for adults with a college degree. Besides the cost, the time investment required for a hobby like this is huge, and what you produce would be effectively worthless in 2019 (financially, at least).