Hello List! Just a small reminder for our next Mumble Sessions on this Sunday, 12/01/2020, at 21.00 Hong Kong Time. Please join us as usual at our Mumble Server murmur.libresilicon.com at Port 64738, the Channel is IC.
Thanks for the attention!
Hello List.
Today, while talking about different possibilities to get back into a clean room and to continue with our work there, we had once again an other technical topic in our weekly Mumble session regarding the feature size on the steppers [0],[1].
All other machines are quite feasible in every clean room, but the resolution of the stepper determinate the size feature size of our technology node. Or, with smaller stepper resolutions we can manufacture smaller structures with Photolithography [2],[3] on silicon. Short, this machines limiting our possibilities, and, of course, they need masks [4] which are quite expensive.
Already a couple of weeks ago came up the idea of using quite common display techniques instead of classical masks, which would allow to manufacture / photolithography mask-less but faster than usual Laser plotters are. Today Ference proposed to use the also quite common DPL technique [6],[7] which is used in Off-the-Shelf Projectors.
So we are now thinking about designing / building a Stepper from scratch, or to replace the mask-holding parts / optics by something like Laser + Diffusor + DLP.
Of course, this machine building stuff should be also open source, public and community-driven as well as our LibreSilicon process. In the beginning there is no need for keeping this machine in a clean room under controlled conditions. Just later, when it comes to production a machines has to get clean.
So are there physicist out there which like to deal with this topic of constructing / hacking hacking such equipment?
Please, if somebody of you on the list, or someone in your social surroundings, is interesting in this topic, let us know it. We are happy to seeing this kind of hacktivism also rising :-)
Best Regards, Hagen.
[0] https://en.wikipedia.org/wiki/Stepper [1] https://de.wikipedia.org/wiki/Stepper_(Halbleitertechnik) [2] https://en.wikipedia.org/wiki/Photolithography [3] https://de.wikipedia.org/wiki/Fotolithografie_(Halbleitertechnik) [4] https://en.wikipedia.org/wiki/Photomask [5] https://de.wikipedia.org/wiki/Fotomaske [6[ https://en.wikipedia.org/wiki/Digital_Light_Processing [7] https://de.wikipedia.org/wiki/Digital_Light_Processing
Hi I've now sent an email to a mass-tape-out capable factory in Portugal, and asked whether they'd be interested into participating into LibreSilicon. I'm now waiting for an answer.
About the stepper aligner with mask less lithography feature: You don't need a clean room for refurbishing an old stepper aligner and replacing the mask exposure part with a projector.
I'm right now in Hong Kong and will most likely move over to Portugal, as soon as I finally get a feedback from the European Union on the status of my offer to continue our research there.
I don't have the time to work full time on reverse engineering a stepper aligner besides developing a manufacturing process, but I can certainly give any volunteer for this task a helping hand with my long years of experience in doing so.
Cheers -lev
On Monday, 13 January 2020 12:44:27 AM HKT Hagen SANKOWSKI wrote:
Hello List.
Today, while talking about different possibilities to get back into a clean room and to continue with our work there, we had once again an other technical topic in our weekly Mumble session regarding the feature size on the steppers [0],[1].
All other machines are quite feasible in every clean room, but the resolution of the stepper determinate the size feature size of our technology node. Or, with smaller stepper resolutions we can manufacture smaller structures with Photolithography [2],[3] on silicon. Short, this machines limiting our possibilities, and, of course, they need masks [4] which are quite expensive.
Already a couple of weeks ago came up the idea of using quite common display techniques instead of classical masks, which would allow to manufacture / photolithography mask-less but faster than usual Laser plotters are. Today Ference proposed to use the also quite common DPL technique [6],[7] which is used in Off-the-Shelf Projectors.
So we are now thinking about designing / building a Stepper from scratch, or to replace the mask-holding parts / optics by something like Laser + Diffusor + DLP.
Of course, this machine building stuff should be also open source, public and community-driven as well as our LibreSilicon process. In the beginning there is no need for keeping this machine in a clean room under controlled conditions. Just later, when it comes to production a machines has to get clean.
So are there physicist out there which like to deal with this topic of constructing / hacking hacking such equipment?
Please, if somebody of you on the list, or someone in your social surroundings, is interesting in this topic, let us know it. We are happy to seeing this kind of hacktivism also rising :-)
Best Regards, Hagen.
[0] https://en.wikipedia.org/wiki/Stepper [1] https://de.wikipedia.org/wiki/Stepper_(Halbleitertechnik) [2] https://en.wikipedia.org/wiki/Photolithography [3] https://de.wikipedia.org/wiki/Fotolithografie_(Halbleitertechnik) [4] https://en.wikipedia.org/wiki/Photomask [5] https://de.wikipedia.org/wiki/Fotomaske [6[ https://en.wikipedia.org/wiki/Digital_Light_Processing [7] https://de.wikipedia.org/wiki/Digital_Light_Processing
Hello everyone, i am Noah, M.Sc. in EE / Nanoelectronics from RWTH Aachen University. would be very interested participating in building a stepper scanner. What is your idea for tackling the problem? It would be helpful to understand what kind of expertise is needed in the team (as I do not know the current members of this group and couldnt join the mumble chat yesterday) to see if i could be of much help. Small disclaimer, I would be able to participate come the second week of February. Right now my schedule is way too packed already.
Best, Noah
________________________________ Von: Libresilicon-developers libresilicon-developers-bounces@list.libresilicon.com im Auftrag von David Lanzendörfer david.lanzendoerfer@lanceville.cn Gesendet: Sonntag, 12. Januar 2020 18:32:45 An: libresilicon-developers@list.libresilicon.com Betreff: Re: [Libre-silicon-devel] Reminder - Next Mumble session on Sunday, 12/01/2020, 21.00 HKT
Hi I've now sent an email to a mass-tape-out capable factory in Portugal, and asked whether they'd be interested into participating into LibreSilicon. I'm now waiting for an answer.
About the stepper aligner with mask less lithography feature: You don't need a clean room for refurbishing an old stepper aligner and replacing the mask exposure part with a projector.
I'm right now in Hong Kong and will most likely move over to Portugal, as soon as I finally get a feedback from the European Union on the status of my offer to continue our research there.
I don't have the time to work full time on reverse engineering a stepper aligner besides developing a manufacturing process, but I can certainly give any volunteer for this task a helping hand with my long years of experience in doing so.
Cheers -lev
On Monday, 13 January 2020 12:44:27 AM HKT Hagen SANKOWSKI wrote:
Hello List.
Today, while talking about different possibilities to get back into a clean room and to continue with our work there, we had once again an other technical topic in our weekly Mumble session regarding the feature size on the steppers [0],[1].
All other machines are quite feasible in every clean room, but the resolution of the stepper determinate the size feature size of our technology node. Or, with smaller stepper resolutions we can manufacture smaller structures with Photolithography [2],[3] on silicon. Short, this machines limiting our possibilities, and, of course, they need masks [4] which are quite expensive.
Already a couple of weeks ago came up the idea of using quite common display techniques instead of classical masks, which would allow to manufacture / photolithography mask-less but faster than usual Laser plotters are. Today Ference proposed to use the also quite common DPL technique [6],[7] which is used in Off-the-Shelf Projectors.
So we are now thinking about designing / building a Stepper from scratch, or to replace the mask-holding parts / optics by something like Laser + Diffusor + DLP.
Of course, this machine building stuff should be also open source, public and community-driven as well as our LibreSilicon process. In the beginning there is no need for keeping this machine in a clean room under controlled conditions. Just later, when it comes to production a machines has to get clean.
So are there physicist out there which like to deal with this topic of constructing / hacking hacking such equipment?
Please, if somebody of you on the list, or someone in your social surroundings, is interesting in this topic, let us know it. We are happy to seeing this kind of hacktivism also rising :-)
Best Regards, Hagen.
[0] https://en.wikipedia.org/wiki/Stepper [1] https://de.wikipedia.org/wiki/Stepper_(Halbleitertechnik) [2] https://en.wikipedia.org/wiki/Photolithography [3] https://de.wikipedia.org/wiki/Fotolithografie_(Halbleitertechnik) [4] https://en.wikipedia.org/wiki/Photomask [5] https://de.wikipedia.org/wiki/Fotomaske [6[ https://en.wikipedia.org/wiki/Digital_Light_Processing [7] https://de.wikipedia.org/wiki/Digital_Light_Processing
Hello Noah!
On 1/12/20 9:08 PM, Wickel, Noah wrote:
Hello everyone, i am Noah, M.Sc. in EE / Nanoelectronics from RWTH Aachen University. would be very interested participating in building a stepper scanner. What is your idea for tackling the problem? It would be helpful to understand what kind of expertise is needed in the team
Well, how familiar you are with the silicon chip manufacturing process?
For our 1 um process we should be able to bring well-shaped structures at least in this size and with good accuracy on the photo resist.
The wavelength of light is between 750 nm (red) and 380 nm (violett).
Older machines on the market are in the 650 down to 500 nm resolution range (as we used at HKUST).
But for smaller structures, we need better resolutions. Mostly newer machines using diffraction (and multi-pattering) therefor.
So I guess, the knowledge someone needs here is a mixture of - photolithography (how this is done on the wafers) - physics of Light - optics for scaling down bigger structures in the mask down to the small structures we need on the resist - mechanical (and systems) engineering - electronics (for controlling the machine)
Our issue here is the following: in HKUST we had a stepper which was feasible. But now looking around to get into other clean rooms, we often see that there are *no* steppers. This also means, that a good photolithography for small clean rooms on Universities is not so quite common as expected.
We could buy and install one in the clean room we like to go into (assuming we had a budget for that). But other folks which like to use our LibreSilicon process in their clean room probably facing the same problem. Hence, finding a solution which closes this gap for others too would be nice.
Okay, this sounds a little bit scaring, I know. But probably you can set your footprint into that field and get your job offer from a equipment supplier if you like.
Regards, Hagen
Good day,
didn't want to spam the mailling list with too much personal introduction right away ;) But ofc you should know a little about me if we might be working on a project together!
how familiar you are with the silicon chip manufacturing process?
Studying electrical engineering i had the chance to go through the whole manufacturing process for 4µm gate length SOI nmos and pmos devices in a guided weekly lab course. It has been about two years since i was active in the lab, but i have some understanding of the general processes: standard clean, dry and wet etching, lithography with positive and negative processes, growing oxide... the bottom line is, i understand most of it but never worked in industry, or using cutting edge tools. And I am 99% sure RWTH's lab does not have a stepper. We used visible/near UV contact lithography in that course.
If I understand correctly, you plan to use double patterning to make 1µm work on a regular projector's pixel grid and a condenser optics?
I guess that is both a very ambitious but at the same time very tempting project. Looking forward to seeing self-made litho steppers pop up in hackspaces ;D
I guess I could best help with electronics and control programming, since that is an area i have some practical experience (microcontrollers as well as desktop). But not nearly as much as most people you'd meet at CCC -- i just think this is a very interesting project and i have some shallow understanding of almost all the puzzle pieces involved.
Best, Noah
________________________________ Von: Libresilicon-developers libresilicon-developers-bounces@list.libresilicon.com im Auftrag von Hagen SANKOWSKI hsank@posteo.de Gesendet: Montag, 13. Januar 2020 02:40 An: libresilicon-developers@list.libresilicon.com Betreff: Re: [Libre-silicon-devel] Reminder - Next Mumble session on Sunday, 12/01/2020, 21.00 HKT
Hello Noah!
On 1/12/20 9:08 PM, Wickel, Noah wrote:
Hello everyone, i am Noah, M.Sc. in EE / Nanoelectronics from RWTH Aachen University. would be very interested participating in building a stepper scanner. What is your idea for tackling the problem? It would be helpful to understand what kind of expertise is needed in the team
Well, how familiar you are with the silicon chip manufacturing process?
For our 1 um process we should be able to bring well-shaped structures at least in this size and with good accuracy on the photo resist.
The wavelength of light is between 750 nm (red) and 380 nm (violett).
Older machines on the market are in the 650 down to 500 nm resolution range (as we used at HKUST).
But for smaller structures, we need better resolutions. Mostly newer machines using diffraction (and multi-pattering) therefor.
So I guess, the knowledge someone needs here is a mixture of - photolithography (how this is done on the wafers) - physics of Light - optics for scaling down bigger structures in the mask down to the small structures we need on the resist - mechanical (and systems) engineering - electronics (for controlling the machine)
Our issue here is the following: in HKUST we had a stepper which was feasible. But now looking around to get into other clean rooms, we often see that there are *no* steppers. This also means, that a good photolithography for small clean rooms on Universities is not so quite common as expected.
We could buy and install one in the clean room we like to go into (assuming we had a budget for that). But other folks which like to use our LibreSilicon process in their clean room probably facing the same problem. Hence, finding a solution which closes this gap for others too would be nice.
Okay, this sounds a little bit scaring, I know. But probably you can set your footprint into that field and get your job offer from a equipment supplier if you like.
Regards, Hagen
-- support LibreSilicon to get back into the Clean Room https://www.gofundme.com/f/libresilicon-cleanroom-rent _______________________________________________ Libresilicon-developers mailing list Libresilicon-developers@list.libresilicon.com https://list.libresilicon.com/mailman/listinfo/libresilicon-developers
Hi
Studying electrical engineering i had the chance to go through the whole manufacturing process for 4µm gate length SOI nmos and pmos devices in a guided weekly lab course. It has been about two years since i was active in the lab, but i have some understanding of the general processes: standard clean, dry and wet etching, lithography with positive and negative processes, growing oxide... the bottom line is, i understand most of it but never worked in industry, or using cutting edge tools. And I am 99% sure RWTH's lab does not have a stepper. We used visible/near UV contact lithography in that course.
Sounds already good alright.
If I understand correctly, you plan to use double patterning to make 1µm work on a regular projector's pixel grid and a condenser optics?
For 1µm we might actually still fly ok with normal exposure using a projector but for higher resolutions we will have to use double patterning because you can only reduce so far with a lens system.
I guess that is both a very ambitious but at the same time very tempting project. Looking forward to seeing self-made litho steppers pop up in hackspaces ;D
Exactly. Yes. We can use a DMD[1] which already can go down to 500nm, as I've seen recently[2], assuming, we get the lens system right.
I guess I could best help with electronics and control programming, since that is an area i have some practical experience (microcontrollers as well as desktop). But not nearly as much as most people you'd meet at CCC -- i just think this is a very interesting project and i have some shallow understanding of almost all the puzzle pieces involved.
I've checked on AliExpress and there are plenty of inexpensive DMD chips available for sale. One of those in combination with a good lens system and positioning system, might actually do the trick.
-lev
[1] https://en.wikipedia.org/wiki/Digital_micromirror_device [2] https://www.nanosystem-solutions.com/en/product/maskless
Hello.
On 1/14/20 11:08 AM, David Lanzendörfer wrote:
I guess that is both a very ambitious but at the same time very tempting project. Looking forward to seeing self-made litho steppers pop up in hackspaces ;D
Exactly. Yes.
:-)
We can use a DMD[1] which already can go down to 500nm, as I've seen recently[2], assuming, we get the lens system right.
A issue I see here is the pattern structure of DMDs. Between all the small mirrors are always even smaller gaps.
If we like to draw a bar, eg. poly or whatever (ASCII-Art)
------------------------- ----- ^ | v ------------------------- -----
the mirrors are with a gap instead
+-------+ +-------+ ----- | mirror| |mirror | ^ | | gap | | | | | | | v +-------+ +-------+ -----
|<----->| ???
This means that it becomes difficult to "stitch" two mirrors gap-less. For human eyes it does not matter, the eyes does not see the pixel / pattern structure. But I guess, the photo resist will see 'em - so we have to tricked them out somehow.
Can we use here diffraction? Or we already need double-pattering here?
Regards, Hagen.
DLP/DMD is promising. I'd also say check out the short throw laser projectors - in particular the new Xiaomi 4k model has native 0.5um pitch, 5000 lumens and only costs ¥10999rmb/€1433 for CNY specials - it uses ALPD 3.0 though as I understand it Appotronics came out with an even newer 4.0 spec in 2018: http://www.appotronicsusa.com/alpd_technology.html -pmg
On Tue, Jan 14, 2020 at 7:00 PM Hagen SANKOWSKI hsank@posteo.de wrote:
Hello.
On 1/14/20 11:08 AM, David Lanzendörfer wrote:
I guess that is both a very ambitious but at the same time very tempting project. Looking forward to seeing self-made litho steppers pop up in hackspaces ;D
Exactly. Yes.
:-)
We can use a DMD[1] which already can go down to 500nm, as I've seen recently[2], assuming, we get the lens system right.
A issue I see here is the pattern structure of DMDs. Between all the small mirrors are always even smaller gaps.
If we like to draw a bar, eg. poly or whatever (ASCII-Art)
^ | v
the mirrors are with a gap instead
+-------+ +-------+ ----- | mirror| |mirror | ^ | | gap | | | | | | | v +-------+ +-------+ -----
|<----->| ???This means that it becomes difficult to "stitch" two mirrors gap-less. For human eyes it does not matter, the eyes does not see the pixel / pattern structure. But I guess, the photo resist will see 'em - so we have to tricked them out somehow.
Can we use here diffraction? Or we already need double-pattering here?
Regards, Hagen.
-- support LibreSilicon to get back into the Clean Room https://www.gofundme.com/f/libresilicon-cleanroom-rent
Libresilicon-developers mailing list Libresilicon-developers@list.libresilicon.com
https://secure-web.cisco.com/1X48t-eiDZPXuS3di6IC1xHDUuEQ5PNtCMljSC7HHWWK2fL...
Thanks for the Link!
As I understood, APLD versions are based on different techniques.
ALPD 3.0 has red and blue laser APLD 4.0 has blue laser plus phosphor
Former versions are with the typical colour wheel which we do not need.
We just need the laser source, and if this is blue, we have shorter wave length which is fine. Photoresists are sensitive to short wave length light, hence the yellow light inside clean rooms. If there is red light (close to the insensitive yellow spectrum) also we could use this for the mask position marker.
Sounds nice! So have a deeper look at that!
Hagen.
On 1/14/20 3:59 PM, Paul Gerhardt wrote:
DLP/DMD is promising. I'd also say check out the short throw laser projectors - in particular the new Xiaomi 4k model has native 0.5um pitch, 5000 lumens and only costs ¥10999rmb/€1433 for CNY specials - it uses ALPD 3.0 though as I understand it Appotronics came out with an even newer 4.0 spec in 2018: http://www.appotronicsusa.com/alpd_technology.html -pmg
On Tue, Jan 14, 2020 at 7:00 PM Hagen SANKOWSKI <hsank@posteo.de mailto:hsank@posteo.de> wrote:
Hello. On 1/14/20 11:08 AM, David Lanzendörfer wrote: >> I guess that is both a very ambitious but at the same time very tempting >> project. Looking forward to seeing self-made litho steppers pop up in >> hackspaces ;D > Exactly. Yes. :-) > We can use a DMD[1] which already can go down to 500nm, as I've seen > recently[2], assuming, we get the lens system right. A issue I see here is the pattern structure of DMDs. Between all the small mirrors are always even smaller gaps. If we like to draw a bar, eg. poly or whatever (ASCII-Art) ------------------------- ----- ^ | v ------------------------- ----- the mirrors are with a gap instead +-------+ +-------+ ----- | mirror| |mirror | ^ | | gap | | | | | | | v +-------+ +-------+ ----- |<----->| ??? This means that it becomes difficult to "stitch" two mirrors gap-less. For human eyes it does not matter, the eyes does not see the pixel / pattern structure. But I guess, the photo resist will see 'em - so we have to tricked them out somehow. Can we use here diffraction? Or we already need double-pattering here? Regards, Hagen. -- support LibreSilicon to get back into the Clean Room https://www.gofundme.com/f/libresilicon-cleanroom-rent _______________________________________________ Libresilicon-developers mailing list Libresilicon-developers@list.libresilicon.com <mailto:Libresilicon-developers@list.libresilicon.com> https://secure-web.cisco.com/1X48t-eiDZPXuS3di6IC1xHDUuEQ5PNtCMljSC7HHWWK2fL8eXVUZId5FIJWetqtwlyRCH5TEcdYnDp14vrL98QEsWiGOFel2DA1ukp9mAyCrJLQmOvAsxeKV-I1Q8EICCVOhZ5RD3US1X4ydX1EU7Wr19BVDpfJLzjtYqYZ1gjgnWmHymClgWu4aBNU69M5q03lMmBUkIyWAN6B3tim8ZdhwUQtl-zw_aUBJnrN404AkvSpHHx3UimhVNuCtwEQmQ1Sb78Qqpil9mwQqtde853NA5J93_BVscLDMkrgRgbVl7QMQ-bHwP7WnCV76gUxRoxtFrjOWhC9544WXosy8W9-VodD2fJPBSwwchPmz6SIa8Ph4ivjZq7fF4c_nmX3VXQd9Cyz0WHZl7lR3_QrxcK-RcqUciHG5JAA5lauAiWBeHUMh0ITEFZ12cSuvcq_jwN7f65GUg8ZMh5atoqeMrA/https%3A%2F%2Flist.libresilicon.com%2Fmailman%2Flistinfo%2Flibresilicon-developers
Libresilicon-developers mailing list Libresilicon-developers@list.libresilicon.com https://list.libresilicon.com/mailman/listinfo/libresilicon-developers
All,
One of the difficult engineering problems for maskless lithography is throughput. So maybe the first task for a guy looking further into this is compute what the expected throughput will be of the maskless lithography system one wants to develop.
The first thing to do is determine how many pixels you want to use for printing a minimum width feature. If you stick to use one pixel for a minimum feature width and the layer is for example polysilicon you will be limited in the variation you can make of width, length and pitch of the poly lines. If this is done one knows how much area will be covered by the resolution of the selected projection system and thus how many images are needed to cover a full waver. Finally step is to determine how many images per second one can print, including alignment after each position change and this will determine the time it will take to print a full waver.
An alternative is to use such a litho system to make your own masks and use these then for printing wafers.
greets, Staf. Paul Gerhardt schreef op di 14-01-2020 om 22:59 [+0800]:
DLP/DMD is promising. I'd also say check out the short throw laser projectors - in particular the new Xiaomi 4k model has native 0.5um pitch, 5000 lumens and only costs ¥10999rmb/€1433 for CNY specials - it uses ALPD 3.0 though as I understand it Appotronics came out with an even newer 4.0 spec in 2018: http://www.appotronicsusa.com/alpd_technology.html-pmg
On Tue, Jan 14, 2020 at 7:00 PM Hagen SANKOWSKI hsank@posteo.de wrote:
Hello.
On 1/14/20 11:08 AM, David Lanzendörfer wrote:
I guess that is both a very ambitious but at the same time very tempting
project. Looking forward to seeing self-made litho steppers pop up in
hackspaces ;D
Exactly. Yes.
:-)
We can use a DMD[1] which already can go down to 500nm, as I've seen
recently[2], assuming, we get the lens system right.
A issue I see here is the pattern structure of DMDs. Between all the
small mirrors are always even smaller gaps.
If we like to draw a bar, eg. poly or whatever
(ASCII-Art)
^ | v
the mirrors are with a gap instead
+-------+ +-------+ -----
| mirror| |mirror | ^
| | gap | | |
| | | | v
+-------+ +-------+ -----
|<----->| ???This means that it becomes difficult to "stitch" two mirrors gap-less.
For human eyes it does not matter, the eyes does not see the pixel /
pattern structure. But I guess, the photo resist will see 'em - so we
have to tricked them out somehow.
Can we use here diffraction?
Or we already need double-pattering here?
Regards,
Hagen.
_______________________________________________Libresilicon-developers mailing listLibresilicon-developers@list.libresilicon.com https://list.libresilicon.com/mailman/listinfo/libresilicon-developers
libresilicon-developers@list.libresilicon.com
-
David Lanzendörfer -
Hagen SANKOWSKI -
Paul Gerhardt -
reminder＠libresilicon.com -
Staf Verhaegen -
Wickel, Noah