OpenASIC : technology for DIY microchip manufacturing

by Mikhail Svarichevskiy

Mikhail Svarichevskiy wants some feedback on this project. What do you like? What could be better? Anything missing?

No feedback yet! Be the first to leave feedback

This is not a live project. This is a draft shared by Mikhail Svarichevskiy for feedback.

This project does not meet the Kickstarter Project Guidelines. Mikhail Svarichevskiy may disable this link at any time.

OpenASIC : technology for DIY microchip manufacturing

by Mikhail Svarichevskiy

  • Home
  • Updates 0
  • Backers 0
  • Comments 0
Photo-main
Play
00:00
00:00
  • Embed

This technology would allow anyone to manufacture open source microchips in their garage lab.

OpenASIC: What is this and why it's important?

Today all open source hardware is built around proprietary integrated circuits - so even if schematics, PCB layout and firmware is open, you still depend on manufacturer of this central proprietary part. OpenASIC project aims to develop open source microchip manufacturing technology which would allow anyone with reasonably equipped garage-lab to make their own microchips.

Yes, initially they will be very limited from the technology side, and will take a lot of labor - but as anything open source - it will evolve over time.

OpenASIC technology, it's limitations and trade-offs

You might have seen schematic MOSFETs - this is what we are trying to make:

In the simplest case we aim to get NMOS/PMOS logic working, which means we only need 1 type of transistors. NMOS transistors are faster, PMOS - are more tolerant to impurities. Doing only 1 type of transistors is easier because we only need 1 doping - and it's the most realistic for home lab.

Most old integrated circuits are made with self-aligned silicon gates - but silicon deposition generally requires toxic and explosive gas - silane, so we won't use it. Instead, we going back to ancient Aluminum gates - this significantly lowers transistor switching speed, but makes whole process much safer and simpler.

How slow is slow? Probably around 1Mhz, give or take. No threat for Intel (yet).

Another simplification is in transistor insulation (like LOCOS or STI) - as we don't need maximum transistor density, we can omit insulation at all and just increase distance between transistors until they don't interfere with each other. 

And finally - we would not have more than 2 levels of metalization, even 2 might be hard for home lab as we won't have planarization, rough surface might cause problems with the second layer. So, in the worst case, we will have to use only 1 level of metalization. Anyway, even 1 layer is enough to be able to make useful microchips.

Based on this technology other semiconductor devices are also possible, like power diodes or small solar cells.

Main microchip manufacturing steps

  • We start with semiconductor grade silicon wafers, which are already doped p or n type. They are not expensive and easy to get. As we are building small lab, we only work with 1x1 cm wafer pieces (would call it a 'wafer' anyway).
  • RCA clean
  • Grow thick oxide layer in a furnace in steam+oxygen
  • Deposit photoresist, expose it with 'first doping' mask using a microscope, develop.
  • Etch off thick oxide (~500nm) through photoresist in BOE
  • Do a predeposition and drive in of our doping chemical in furnace (might be BBr3, POCl3 or spin-on dopant, at about 1000C)
  • Deposit photoresist, expose it with 'gate' mask, develop.
  • Etch off thick oxide through photoresist, and the grow thin (50-100nm) high-quality gate oxide in furnace in pure oxygen.
  • Deposit photoresist, expose it with 'source/drain contact' mask, develop.
  • Etch off SiO2 under source/drain contacts
  • Deposit Aluminum in vacuum everywhere
  • Deposit photoresist, expose it with 'M1 metal' mask, develop.
  • Etch off Aluminum. Alternatively, lift-off process might be used here.
  • Anneal to form ohmic contacts of Aluminum to silicon.

Obviously, each step have a lot of details and specific parameters we need to figure out exactly, and this will take a lot of time.

Safety notes

This is not going to be magic box, which you can place on your desk near your bedroom (like RepRap) due to safety limitations: some acids and toxic substances are unavoidable in microchip manufacturing, as well as high-temperature operations (up to 1100°C / 2000°F).

So developing safety guidelines & providing safety datasheets for all chemicals required for the process is essential, and only reasonably equipped home- and university-labs with all required protection would be able to make microchips using this technology.

What was already done on the project

I decided to find out how to make my own microchips about 2.5 years ago - and during that time I bought all possible books on the topic, read through tons of papers on various steps of manufacturing technology, was able to get detailed process description (with all steps, temperatures and concentrations) from actual semiconductor foundry which I would be able to use as a reference, and bought almost all chemicals needed to make microchip at home (although, not sure yet if they are pure enough for microchips), and some required equipment (vacuum pump & furniture, metallographic microscope, parts for 1100C furnace).

Why do I need the money

First of all I need to improve safety - at the moment everything is stored and all tests are being done in my bedroom which is way beyond any safety rules, so moving all chemicals to a separate place would be much safer. There are some other things like full-body protection suite, laminar boxes, proper air filtering.

Then there are things which make work easier and faster - nitrogen generator, deionized water generator, and test equipment. When working in separate place, I would be able to use pressurized oxygen (at the moment I am going to make it chemically or through electrolysis).

The only materials I still need are spin-on dopants/glasses - they would replace toxic BBr3 and POCl3, but they are quire expensive and hard to buy for an individual (would be happy if someone would help me to buy them from Honeywell for example).

If the project would exceed it's funding goal, it would be possible to extend the scope and put more effort on CMOS and self-aligned gates (any of these features would make the technology much more useful).

Timeline

After project is funded I expect that about 3-5 months will take to receive latest pieces of materials and equipment that I will be buying. Then there will be 6-9 months of work on technology, trying out different variants until it all works fine.

When it all works for me - I will be writing detailed documentation on how to reproduce it, including all steps, chemicals and equipment, and then I'll start working on manufacturing microchips I will need to ship, which should be done about 18 months after project is funded.

Risks and challenges Learn about accountability on Kickstarter

Doing something which is generally believed to be impossible to do at home - is inherently risky. But given the amount of research already done I believe the only question is how many transistors we would be able to fit reliable in 1 integrated circuit, and would it only have simple NMOS or PMOS transistors or much more flexible CMOS circuits.

Another risk is that it will take more time to 'debug' the technology - this might lead to 6-12 months delay on delivering technology description and working chips, but at the moment I do not anticipate any show stoppers.

0
Backers
$0.00
pledged of $35,000 goal
Me_chem.medium
Project by
Mikhail Svarichevskiy

Moscow, Russia Contact me

  • First created · 0 backed
  • Has not connected Facebook

See full bio

  • Pledge $10 or more
    You selected

    0 backers

    You get your personal copy of OpenASIC manufacturing technology emailed to you, which would allow you to make your own microchips (or fully understand how exactly these microchips are made).

    Estimated delivery:
  • Pledge $25 or more
    You selected

    0 backers

    All of the above plus access to preliminary documentation revisions and your name will be forever listed as OpenASIC founding supporter.

    Estimated delivery:
  • Pledge $100 or more
    You selected

    0 backers

    All of the above plus A4-size high quality photo of the world-first working microchip made at home with autograph.

    Estimated delivery:
  • Pledge $500 or more
    You selected

    0 backers

    All of the above plus actual working device powered by homemade microchip. Depending on technology level we would be able to achieve it might be microcomputer with small LCD display, LED clock or in the very worst case blinking LED (that would be sad). It will have acrylic case so you would be able to amaze your visitors with it's homemade heart. Getting 1 microchip working might take up to 100 hours (10 hours per chip, 10% chance that chips would work) - hence the reason why I cannot offer working device for more backers.

    Estimated delivery:
  • Pledge $2,500 or more
    You selected

    0 backers

    All of the above plus we will write your name on each homemade microchip we make and surely it will appear on all microchip photos. Your name will be part of history!

    Estimated delivery: