Hi folks I just got an email last night from Oklahoma from some crypto miners who wanna print microchips for building miners... Maybe that might be THE solution for our issue with dealy poisonous chemicals when running a classic silicon based process in a home lab? What do you all think?
Cheers -lev
On Tue, Mar 8, 2022 at 4:52 PM David Lanzendörfer leviathan@libresilicon.com wrote:
Hi folks I just got an email last night from Oklahoma from some crypto miners who wanna print microchips for building miners...
if they have several tens of billions of dollars available once converted out of {insert-currency-of-choice}, then they can buy a Foundry, and release the PDK just like Skywater did.
caveat: that's only if it is *their* PDK - a number of Foundries actually buy in their PDK and process from *other Foundries*, such as the India Lab, which bought the Israeli Tower Semi process (Tower Semi is currently being bought by Intel).
Global Foundries was available lock stock and barrel for 4.5 billion a few years ago.
l.
Hi No. They've asked about printing chips, which would include some highly specialized submicron nozzle and special polymers: https://pubs.rsc.org/en/content/articlelanding/2020/tc/d0tc01341b
It might really be a feasible option for manufacturing LibreSilicon layouts...
Any chemists around?
Cheers -lev
On Tuesday, March 8, 2022 5:48:20 PM WET Luke Kenneth Casson Leighton wrote:
On Tue, Mar 8, 2022 at 4:52 PM David Lanzendörfer
leviathan@libresilicon.com wrote:
Hi folks I just got an email last night from Oklahoma from some crypto miners who wanna print microchips for building miners...
if they have several tens of billions of dollars available once converted out of {insert-currency-of-choice}, then they can buy a Foundry, and release the PDK just like Skywater did.
caveat: that's only if it is *their* PDK - a number of Foundries actually buy in their PDK and process from *other Foundries*, such as the India Lab, which bought the Israeli Tower Semi process (Tower Semi is currently being bought by Intel).
Global Foundries was available lock stock and barrel for 4.5 billion a few years ago.
l. _______________________________________________ Libresilicon-developers mailing list Libresilicon-developers@list.libresilicon.com https://list.libresilicon.com/mailman/listinfo/libresilicon-developers
On Tue, Mar 8, 2022, 10:00 David Lanzendörfer leviathan@libresilicon.com wrote:
Hi No. They've asked about printing chips, which would include some highly specialized submicron nozzle and special polymers: https://pubs.rsc.org/en/content/articlelanding/2020/tc/d0tc01341b
It might really be a feasible option for manufacturing LibreSilicon layouts...
Any chemists around?
well...i have nearly no practical experience, but i do know a lot of chemistry...
From what I can tell, they're comparing organic semiconductors that they
made with an electro-spinning-style process with organic semiconductors that were spin-coated onto a SiO2/Si substrate (the substrate is implied to be just a handy surface to put their organic semiconductors on, it doesn't have silicon transistors afaict)...their process looks like it could be quite useful, if you want organic semiconductors. organic semiconductors generally have several problems when compared to silicon transistors: 1. they are much larger, the resolution you can achieve with silicon is usually several orders of magnitude smaller (tens of nm vs 10-100um). 2. organic semiconductors generally have worse electrical performance than silicon, often by several orders of magnitude -- this is beyond just what the size difference would give. 3. organic semiconductors often degrade quickly (years) vs. silicon which can last tens - hundreds of years. Think OLED screen burn-in.
if you want to build cryptocurrency miners and you care about energy efficiency (hashes per joule), organic semiconductors are a pretty terrible choice, because 1 and 2 combine to give terrible performance and low density.
if you're working with a cryptocurrency where the algorithm changes every week, organic semiconductors might have an advantage, since you could print new ones each week, and hopefully the circuits would be better than a general purpose processor.
Jacob
Hi Jacob Thanks for your contribution! So much appreciated! <3 Hmm... while the initial circuits may not be suitable for hardcore number crunching for hashing blocks, we might be able to use it for prototyping our layouts at home, which already would be awesome progress, especially if it could be done with chemicals which can be bought without ending up on a terror watch list. It would be interesting to see, whether we could build simple circuits like a 555 or maybe even a simple RISC-V MCU with it. Hagen shared another document here BTW: https://sci-hub.ee/https://doi.org/10.1002/adma.201302278
What's in your opinion the best way for patterning of structures using such polymer? I was thinking, we could mix some photoreactive polymer like the photoresist commonly used for photomasks in semiconductor manufacturing and then modify the chemical formulas in order to give them those dopant accepting properties. Then we could even use acryllic glas or so as a substrate (given the surface is enough smooth)
Cheers -lev
On Tuesday, March 8, 2022 7:28:47 PM WET Jacob Lifshay wrote:
On Tue, Mar 8, 2022, 10:00 David Lanzendörfer leviathan@libresilicon.com
wrote:
Hi No. They've asked about printing chips, which would include some highly specialized submicron nozzle and special polymers: https://pubs.rsc.org/en/content/articlelanding/2020/tc/d0tc01341b
It might really be a feasible option for manufacturing LibreSilicon layouts...
Any chemists around?
well...i have nearly no practical experience, but i do know a lot of chemistry...
From what I can tell, they're comparing organic semiconductors that they made with an electro-spinning-style process with organic semiconductors that were spin-coated onto a SiO2/Si substrate (the substrate is implied to be just a handy surface to put their organic semiconductors on, it doesn't have silicon transistors afaict)...their process looks like it could be quite useful, if you want organic semiconductors. organic semiconductors generally have several problems when compared to silicon transistors:
- they are much larger, the resolution you can achieve with silicon is
usually several orders of magnitude smaller (tens of nm vs 10-100um). 2. organic semiconductors generally have worse electrical performance than silicon, often by several orders of magnitude -- this is beyond just what the size difference would give. 3. organic semiconductors often degrade quickly (years) vs. silicon which can last tens - hundreds of years. Think OLED screen burn-in.
if you want to build cryptocurrency miners and you care about energy efficiency (hashes per joule), organic semiconductors are a pretty terrible choice, because 1 and 2 combine to give terrible performance and low density.
if you're working with a cryptocurrency where the algorithm changes every week, organic semiconductors might have an advantage, since you could print new ones each week, and hopefully the circuits would be better than a general purpose processor.
Jacob
On Tue, Mar 8, 2022, 14:25 David Lanzendörfer leviathan@libresilicon.com wrote:
Hi Jacob Thanks for your contribution! So much appreciated! <3 Hmm... while the initial circuits may not be suitable for hardcore number crunching for hashing blocks, we might be able to use it for prototyping our layouts at home, which already would be awesome progress, especially if it could be done with chemicals which can be bought without ending up on a terror watch list. It would be interesting to see, whether we could build simple circuits like a 555 or maybe even a simple RISC-V MCU with it. Hagen shared another document here BTW: https://sci-hub.ee/https://doi.org/10.1002/adma.201302278
What's in your opinion the best way for patterning of structures using such polymer?
unfortunately, that's exactly the kind of question where I won't have a particularly useful answer, since that would require experience, and/or lots of research into the specifics of how such polymers behave, the different methods by which polymers can be patterned, etc. That's more the domain of materials engineering rather than chemistry... That said, I'd make a wild guess that electro-spinning may work well similar to how FDM 3D Printing works, except that the flow of fluid through the nozzle could be quickly changed by changing the electric field, rather than just adjusting a filament-feeder servo.
I was thinking, we could mix some photoreactive polymer like the
photoresist commonly used for photomasks in semiconductor manufacturing and then modify the chemical formulas in order to give them those dopant accepting properties.
that seems likely to break the photoreactive properties and/or the dopability. that said, it never hurt to try, though it may be time-consuming and expensive (specialty chemicals aren't generally cheap).
Then we could even use acryllic glas or so as a substrate (given the surface
is enough smooth)
hmm, i'd use something like glass, or some other ceramic material, since the solvents used can likely easily dissolve acrylic (iirc dcm, which is used in a lot of commercial acrylic solvents/bonders, is very chemically similar to chloroform, which i think is what they were using in that rsc.org paper). glass is easily cleaned, and is quite chemically similar to the SiO2 that they were using in the rsc.org paper, though you may have to watch out for contaminants contained in the glass, which depends on the exact composition of glass you have. I remember hearing about how sodium ions in common glasses can contaminate semiconductors, though that is Si-based semiconductors rather than organic semiconductors. fused silica may be a better option, since it is SiO2, though is more expensive.
Jacob
Hi So yesterday I was having some brainstorming with two guys from here in Portugal, who're going to join our maskless stepper aligner project. We were talking a bit about the substrate too. I think I'll use small glass squares, because for an initial stepper mechanism it's easier to just target single die exposure of like 1cm x 1cm or so, using the laser exposure technique used for photo epoxy 3D printers. I found some modules which might be suitable, so that we don't have to design it from scratch. Photoreactive dopant accepting polymer would really be a game changer tho. Otherwise we could try to "just" scale down a normal additive printer design to a micrometer scale... It will require multiple nozzles tho, because we need to feed more than just one material... p-doped, n-doped and isolator...
Cheers -lev
On Wednesday, March 9, 2022 5:07:33 AM WET Jacob Lifshay wrote:
On Tue, Mar 8, 2022, 14:25 David Lanzendörfer leviathan@libresilicon.com
wrote:
Hi Jacob Thanks for your contribution! So much appreciated! <3 Hmm... while the initial circuits may not be suitable for hardcore number crunching for hashing blocks, we might be able to use it for prototyping our layouts at home, which already would be awesome progress, especially if it could be done with chemicals which can be bought without ending up on a terror watch list. It would be interesting to see, whether we could build simple circuits like a 555 or maybe even a simple RISC-V MCU with it. Hagen shared another document here BTW: https://sci-hub.ee/https://doi.org/10.1002/adma.201302278
What's in your opinion the best way for patterning of structures using such polymer?
unfortunately, that's exactly the kind of question where I won't have a particularly useful answer, since that would require experience, and/or lots of research into the specifics of how such polymers behave, the different methods by which polymers can be patterned, etc. That's more the domain of materials engineering rather than chemistry... That said, I'd make a wild guess that electro-spinning may work well similar to how FDM 3D Printing works, except that the flow of fluid through the nozzle could be quickly changed by changing the electric field, rather than just adjusting a filament-feeder servo.
I was thinking, we could mix some photoreactive polymer like the
photoresist commonly used for photomasks in semiconductor manufacturing and then modify the chemical formulas in order to give them those dopant accepting properties.
that seems likely to break the photoreactive properties and/or the dopability. that said, it never hurt to try, though it may be time-consuming and expensive (specialty chemicals aren't generally cheap).
Then we could even use acryllic glas or so as a substrate (given the surface
is enough smooth)
hmm, i'd use something like glass, or some other ceramic material, since the solvents used can likely easily dissolve acrylic (iirc dcm, which is used in a lot of commercial acrylic solvents/bonders, is very chemically similar to chloroform, which i think is what they were using in that rsc.org paper). glass is easily cleaned, and is quite chemically similar to the SiO2 that they were using in the rsc.org paper, though you may have to watch out for contaminants contained in the glass, which depends on the exact composition of glass you have. I remember hearing about how sodium ions in common glasses can contaminate semiconductors, though that is Si-based semiconductors rather than organic semiconductors. fused silica may be a better option, since it is SiO2, though is more expensive.
Jacob
On Wed, Mar 9, 2022 at 3:56 PM David Lanzendörfer leviathan@libresilicon.com wrote:
Photoreactive dopant accepting polymer would really be a game changer tho. Otherwise we could try to "just" scale down a normal additive printer design to a micrometer scale... It will require multiple nozzles tho, because we need to feed more than just one material... p-doped, n-doped and isolator...
the creator of this - which i thoroughly recommend (it has a 40:1 wormdrive built-in and an automotive-grade flexible driveshaft)
https://flex3drive.com/product/g5-flex-extruder/
also created a dual-extruder head (2-in to 1-nozzle).
there do exist triple extruders https://www.matterhackers.com/store/printer-accessories/polystroooder-tri-ho...
i am assuming of course that the material being "fed" comes in the standard 1.75mm extrusion roll, which, duh, may not actually be the case, duh.
l.
Now we only need help from Ray Palmer and his Atom suit in order to shrink it down to a size where it's usable :-) I think the route of trying to produce a photoreactive polymer is more promising...
-lev
On Wednesday, March 9, 2022 4:04:29 PM WET Luke Kenneth Casson Leighton wrote:
On Wed, Mar 9, 2022 at 3:56 PM David Lanzendörfer
leviathan@libresilicon.com wrote:
Photoreactive dopant accepting polymer would really be a game changer tho. Otherwise we could try to "just" scale down a normal additive printer design to a micrometer scale... It will require multiple nozzles tho, because we need to feed more than just one material... p-doped, n-doped and isolator...
the creator of this - which i thoroughly recommend (it has a 40:1 wormdrive built-in and an automotive-grade flexible driveshaft)
https://flex3drive.com/product/g5-flex-extruder/
also created a dual-extruder head (2-in to 1-nozzle).
there do exist triple extruders https://www.matterhackers.com/store/printer-accessories/polystroooder-tri-ho tend-1.75mm
i am assuming of course that the material being "fed" comes in the standard 1.75mm extrusion roll, which, duh, may not actually be the case, duh.
l. _______________________________________________ Libresilicon-developers mailing list Libresilicon-developers@list.libresilicon.com https://list.libresilicon.com/mailman/listinfo/libresilicon-developers
I just had an idea! You folks were talking about using a laser from a DVD reader/writer, right? Lets use this idea to build a small exposure unit for hardening out dopant accepting polymers! That solves ALL the problems we had so far with poisonous chemicals!
Cheers -lev
On Tuesday, March 8, 2022 5:48:20 PM WET Luke Kenneth Casson Leighton wrote:
On Tue, Mar 8, 2022 at 4:52 PM David Lanzendörfer
leviathan@libresilicon.com wrote:
Hi folks I just got an email last night from Oklahoma from some crypto miners who wanna print microchips for building miners...
if they have several tens of billions of dollars available once converted out of {insert-currency-of-choice}, then they can buy a Foundry, and release the PDK just like Skywater did.
caveat: that's only if it is *their* PDK - a number of Foundries actually buy in their PDK and process from *other Foundries*, such as the India Lab, which bought the Israeli Tower Semi process (Tower Semi is currently being bought by Intel).
Global Foundries was available lock stock and barrel for 4.5 billion a few years ago.
l. _______________________________________________ Libresilicon-developers mailing list Libresilicon-developers@list.libresilicon.com https://list.libresilicon.com/mailman/listinfo/libresilicon-developers
libresilicon-developers@list.libresilicon.com
-
David Lanzendörfer -
Jacob Lifshay -
Luke Kenneth Casson Leighton