DIY Laser Lithography: Micron resolution
Vložit
- čas přidán 7. 06. 2024
- DIY photolithography using a direct write laser machine.
This is a maskless technique (similar to projection lithography) to create patterns in a photosensitive resin (photoresist). This is a common step in many microfabrication operations like transistor/semiconductor fabrication, MEMS devices, microfluidics, etc.
The advantage of maskless litho over traditional masked lithography is that you don't need to generate a mask first (which often itself requires photolithography). Maskless is faster for one-off and prototype work, but mask is superior once you need to create millions of devices.
🥰 CONSIDER SUBSCRIBING 🥰
🔬Patreon if that's your jam: / breakingtaps
☕Or buy me a coffee? www.buymeacoffee.com/Breaking...
📢Twitter: / breakingtaps
📷Instagram: / breakingtaps
💻Discord: / discord
==== Links ====
- "Maker Foundry" Hackaday project: hackaday.io/project/25260-mak...
==== Analysis Equipment ====
nGauge AFM from ICSPI (www.icspicorp.com/)
Phenom XL SEM from Thermo
Gwyddion for AFM post-processing
Blender for 3D AFM rendering
==== Timeline ====
0:00 Direct write laser lithography
1:01 Some exposed photoresist
2:06 Spatial filtering
3:59 Focus detection (unused)
5:08 Walking the beam
6:05 Galvo scanner
7:17 Scan/Tube lens
7:56 Microscope Objective and substrate
8:35 Optical Rlay Full Explanation
10:35 2x Beam Expander
11:33 Results (Datak Er-71)
12:40 Results (Microposit s1818)
16:37 Problems to fix - Věda a technologie
🚨 *Addendum* 🚨
- "Why not fix the laser and just move the stage under the laser?"
It's a good question! Speed mostly, it's generally a lot slower to draw the pattern with a mechanical stage, particularly if you need a lot of "flood fill" regions. And in theory a galvo can give very small resolution even if the laser spot is relatively large. I.e. you could position a 2 micron line (due to 2um spot size) 200nm away from another line. Depends on the quality of a stage, but precise movements of a large stage are a lot harder than precise rotations of a lightweight mirror, especially considering backlash and such on the stage. But it's a totally viable technique, and I might try it if I can't make the galvo behave. 🙂
- "Why direct-write laser instead of DLP chip?"
No particular reason (my first attempt years ago was with a pico projector actually), I just wanted to try the direct-laser approach. There are some pros/cons to each. With DLP you run into limitations with pixel size quickly, which means you need correspondingly better optics to continue reducing feature size. Direct write laser can "scale" in that aspect a lot cheaper: a good clean beam passed through a 60x objective could theoretically hit sub-micron features. OTOH, you have to deal with issues like the galvo linearity/stability. So different trade offs. You can find commercial devices of both varieties competing in the same space for example :)
Hah, thanks for answering that. What about a stage that is based on one of those flexure designs? That would allow tiny movements.
Thank you for the considered answer! Have you also thought about using a Wicked Lasers LaserCube?
Its blue diode is 1300mW at 445nm, which is within the range (350nm-450nm) specified by Microposit's datasheet.
Is there any way you can see how far off the laser is coming out of the galvo and just use the stage to correct it? You could try to make the lines straighter and maybe even see if for some reason it's repeatably off mark the same amounts in when doing similar things? Kind of like how the Shaper Origin handheld cnc router uses a small but accurate cnc to improve the positioning you do by hand.
Like fgbhrl suggested flexure xyz stages can be very precise, zero backlash, readily characterized and driven by very standard gcode. Any speed lost due to greater inertial mass may be offset by running many parallel beams at once. You could even make mirror images cut at the same time that would register perfectly with their counterparts - which may be useful somehow in the construction of MEMS devices.
Another argument for using a mechanical stage is that your beams (and thus the walls of your resist) could be made to be exclusively perpendicular to the substrate rather than all converging at the microscope lens.
I mean, you already have a sub-micron stepping stage, you just had to make a spatial filter for your laser and half a decent focusing optics to get better results with less effort... Also, is your laser polarized, and was the dichroic designed to work at 45°? Most of the off-the-shelf dichroic are designed to work at normal incidence and suffer big degradation in performance at high angle, especially if the laser is polarized in the wrong direction
I was going to click the like button, then realized I had already clicked! Really nice project.
Thanks Ben! Glad you enjoyed it ♥
@@BreakingTapsI’m a big fan of both of you. You guys do amazing work!
Between the 2 of you I would expect to learn how to like a video twice
Beginner mistake, no worries, it´ll go away... By "Breaking Taps" u click "like" always before starting the video, everyone knows... ;-)
I see a colab in the future
Nice project Zach! I've got a remark: given the resolution of your patterns, your resist layer thickness is at least several microns. Generally, you don't need that kind of thickness if you want to do subsequent etching. Using a resist layer thickness of less than 300nm will improve your image definition dramatically, because fully exposing a thick layer of photoresist in a very short time all the way to the bottom is very difficult. This is because of the optical density at the laser wavelength and the fact that photobleaching of the photosensitive component isn't instantaneous.
That is amazing to see so many wizards in one comment section!
Thanks for the tips, that's great to know! I wasn't really sure what thickness to be aiming for, so just chose an arbitrary speed near from the datasheet (3500RPM) and left it constant so there was one less variable to think about. Good news is that it seems pretty consistent, around 1.7-2um each time. I'll try increasing the speed and get a thinner layer for my next tests, thanks!
@@BreakingTaps if 2um is your thickness at 3500rpm, then increasing RPMs is probably not going to give you a significantly thinner layer. Adding a little solvent is a better strategy. Or alternatively you could use a resist like az1505.
Hm yeah, the datasheet maxes out with 500nm @ 7000 RPM. My coater should be able to hit that, I'll give it a shot and see. Good idea about adding some solvent. Will look into AZ1505! I've been wanting to get some SU8 as well, just because it looks very useful for "structural" applications where the resist is left in place.
@@BreakingTaps Be careful with exposed SU8! This stuff is ridiculously hard to remove, nearly had to scrap several fully structured 8" MEMS wafer because someone was curious about my stuff and opened the box outside the yellow light area... (partially my fault because I stored the box outside the litho area, but the stepper was down and all shelfs were full).
The thickness of 150µm wasn't helping this either. Warm acetone or NMP over several hours followed by plasma ashing in a O2/Ar plasma somewhat removed it (I also read that warm Piranha solution makes quick process of the SU8 but would have definitely destroyed my wafer) . At least some devices survived.
Neato! I'm impressed with the quality you were able to get out of an old 3d printer mirror box. What kind of setup are you hoping to replace it with? some "better" COTS part, or a homebrew with a mirror bolted to a few piezos or something? Hmmm... are piezos too small for this? it looked like your angular range was actually pretty big.
Actually I guess that scaling is a separate question - when you had the "demo raster" running while you walked through the beam path, the lateral displacement of the beam was on the order of mm. How many microns across the surface was that once sent through the rest of the optics and microscope objective and whatnot?
Thanks! I'm not sure what v2 will look like to be honest. The deflections are pretty big (+/- 20degrees? something like that), but that could be changed... I was trying to maximize the full range of the galvos because I figured they would not have much accuracy if trying to make very small deviations. If the a piezo galvo or whatever had very small range, we can tweak the focal distances in the relay system to make it work, probably just with much longer focal distances.
I have a few options that I'm thinking about:
- Getting a better galvo, the "digital" ones designed for etching should be tuned better.
- Some kind of tip/tilt flexure mechanism, actuated by voice coils or high TPI threads or something. I don't _technically_ need all the speed of the galvo, so we can use a slightly slower mechanism too
- a few different more "mechanical" designs (coupling a ballscrew to a rotary mechanism), so we can take advantage of high resolution in the ballscrew mechanism
- A sorta silly idea someone mentioned on Discord which is growing on me: a CoreXY style 3d printer mechanism that "rasters" above the tube lens in the system, replacing the galvo entirely. Similar to how CO2 laser cutters have a traveling mirror, but coupled into a microscope objective. Still mulling that one over.
- Or just admitting defeat and using the wafer stepper for all motions. Would be very slow though.
The "demo raster" was an "X" shape and just about filled the entire field of view (I used it to help align various components). It depends where you look, but on the first lens it travels a few mm, on the second lens it's traveling a centimeter or two. And ultimately the FOV is either 2x2mm (20x objective) or 500x500um (40x objective).
I suggest looking either for proper galvos (like for actually your purpose) or to keep in mind that the mirror surface of your galvos has optical quality. I assume the galvos you used don't actually have the needed surface quality and that this might create the "wobbly" structure of your trial structures.
Just a suggestion from my side - I hope I'm not too late 😅
@@1xBublex1 theoretically if error is repeatable u could calibrate wobble into "control system". But yeah if there is huge variation and bad repeatability is another case.
I find this channel to be a great compliment to Applied Science. Both channels are doing really complicated things in a home shop, things that you would think would stay in the lab environment.
This guy is doing some crazy work I would never have dreamed could be approached in a DIY setup. The other channel I have seen that blew my mind in this regard is Thought Emporium.
Wow once again I am blown away. I am fresh out of EE, and watching your videos has got me inspired to work on my own laser scanning microscope.
Do you know if commercial LSMs have enough power to bleach the photoresist?
Criminally cool. The combo of the work your doing now with the etching is incredible.
The stuff you manage to make in a home shop always blows my mind. You’re like a whole other level of Maker.
You continually make some of the most interesting stuff I’ve ever seen on the internet. Thanks for all of it.
Way fun and interesting! It is super nice to experience new ideas and principals from a source that elucidates these concepts so understandingly in both general idea and detail. Seriously, well done.
This is some serious DIY. The fact that you are creating the beginnings of an Integrated Circuit with off the shelf hardware and 3d printed parts is seriously impressive.
Another awesome video and project! I have no doubt that you’ll get it all working flawlessly. Imagine making MEMS devices in a home shop…. Good grief. You and Applied Science are always pushing the limits, and I love it!
You beat me to my dream project. Absolutely stellar job, so great to see that hobbyists can do this!
Just listened to your within tolerance episode and I’m stoked to watch your backlog of videos again! Such good content.
Another great video! I'm always excited when I've seen you've posted a new one. Keep 'um coming!
I love your honesty and comments on plastic parrs..good work!
You have done so many cool projects! I am truly impressed. Look forward to the next iteration! How do you not have a million subscribers I do not know.
You've got to look into green lithography technology. I covered some in my quantum hardware lecture series on QuantumGrad. Basically, egg whites diluted with water can function as a polymer photoresist and even e-beam resist. This means that egg whites can serve as resist material for both micron and nano scale features. Plus, it's non-toxic : D
Looking forward to watching this evolve. Great work!
I am everyday fascinated how far we have come. We can do that stuff now AT HOME. Great job!
You're fast becoming one of my favourite youtubers
Another incredible video, as usual. Blows my mind what "home makers" are capable of, these days. Keep up the great work, looking forward to seeing more in the future!
Astounding work! What you have achieved is very amazing, almost incredible. I'm looking forward to seeing how you will refine it further. Thanks for sharing!
This project wins a Golden Opossum award for showing us a really cool project made with less than stellar components. Can't wait to see V2. (:
This is a great video! I'm new to the channel but I'm SO impressed with the quality and clarity of your videos!!!! Long term subscriber here hopefully!
Excellent explanation of the optics between the galvos and microscope objective!
Fantastic ..I worked on optics back in the day .I get surprise about all kinds of new solutions ,that are already available this days .20 year ago something like you have put together will be very ,very expensive ..And only corporations or universities will be able to afford .To see you doing this type of research is a great accomplishment !!
Omg... You are unbelievable.... What is your work ethic that you can finish such complex projects....
Powered by project-ADHD 😂 I usually have 3-4 projects going at once, and when I get bored or burnt out I switch to a different one :) This project for example was 90% done in January but I just couldn't make myself finish, so went off to do other things until I got interested again :)
Pretty cool. I’ve been in semiconductors for over 20 years, servicing mask aligners. 😆
We live in interesting times. It's such a pleasure to see highly intelligent people do what they excel at.
homemade slicon cpu when? you have tried silicon doping, you have the laser machine....
also your channel is awesome
i guess i now have a project to follow on this channel for the forseable years to come
You guys are just awesome. One of you will eventually come up with something that can spit out usable chips within the next 5 years.
Bloody fantastic work.
Fascinating project. Like that it is 3Dp parts ebay/amazon parts. Very cool. Also, happy to see you are wearing the 'favorite' ball cap.
Absolutely beautiful work, very, very nice! :-)
Yeah, diode laser beam quality is beyond poor. If you want maximum available power and a nice clean beam, have you considered fiber coupling the diode? The beam focused into the end of a nice length of Silica SMA905 fiber, will yield a really nice Gaussian beam profile, plus you can run the fiber wherever you want it...
does the beam need to enter the fiber parallel or can it enter it focused to a point? what angular distribution does is have when it exits?
Thanks Les, means a lot coming from an actual optical and laser expert! :) I haven't really looked into fiber coupling to be honest, and no real reason either, just didn't ocurr to me 🙂 Can you couple in multiple wavelengths to a fiber (so I can also include the red focusing laser)? I assume I'd need a single mode fiber instead of the bigger multi-mode stuff? Thanks for the help! My fiber-fu is pretty weak 😅
@@BreakingTaps I've never done it but I believe so. You'll have to make sure that the fiber is rated for the range of wavelengths used. You also need a lens that optimally couples the free space laser to the single mode fiber. Newport has a good article on this if you search for "Fiber Optic Coupling".
Nice project! I built the DLP version of this a few years ago, using a UV light engine and a wafer inspection microscope, rigged to step and repeat. Pretty good results at comparable resolution, although with some imperfections on the tiling.
Took me weeks to align the mirrors on my K40 crapbox, i am in awe of a setup like this, yes different worlds, i know
Dude, you are changing my life over here....
Your channel blows my mind
Very impressive as always.. loved the diagram-animations
this man! like the quality is just bullshit good- always outstanding
Amazing as always.
It looks like the waves are mostly due to the galvo controller. My light Show laser (unity elite pro3) was wavy similar to yours until I aligned it using the potentiometers on controller board. Basically projecting a grid test pattern with pangolin quickshow software and adjusted the controllers until the corners are as close to square as possible. I'm fairly new to pro lasers but these adjustments solved my wavy problems
Great project, thank you! I'm thankful there are folks like you around. If society ever collapses, you're the kind of guy who will help build everything back. (Sorry if I have a depressing opinion of our world and where it's going, but you at least give me hope.)
Excellent video!! Thank you so much for sharing.
I was attempting DIY PCB lithography at the time this video dropped. What a nice coincidence
This video is perfect! thank you so much for sharing your knowledge!
Thanks I had selected the same project for pbysics and now with your contribution I can do it
Brilliant video! I hope to see further iteration on this project!
i struggle with a ikea shoe shelf and this man just diys himslef some lithography im truly stoopid
really well explained for something as magic as optics to my smooth brain
great job. I can see you have spent a lot of time with this.. awesome..
Amazingly, my Recommended feed actually proved useful for once by suggesting this video which led me to discover (and subscribe to) your fantastic channel. Cheers!
Utterly impressive that this can be made as a DIY system.
the pinnacle of scien for general public explained in a very clear way. yes this videos will start a revolution!
As always very informative and entertaining video
Amazing work, subscribed. Reminds me of taking image stacks through the microscope, Driven by steppers it was move, wait for vibration to stop, open shutter, wait for vibration to stop,flash led light, close shutter. Had to move it from an attic to a cellar due to vibrations. The frame wasn't sturdy but with the right setup it worked very well, these vibrations are always there however big the machine. three phase FOC motors help enourmously with smooth control. Galvo, electrostatic, Henry at UMIST in 1990 built ascanning optical microscope with two 3" loudspeakers and two sweep generators, im feeling inspired.
Very interesting and thanks for taking the time to compile this video. Thanks!!!
I see your uploads and drop everything to watch at 1x!
How about a Risley prism instead of a galvo? It might make for a much more robust and stable assembly.
Oh! Interesting! I wasn't aware of these devices, looks super cool. Will do some more reading, cheers for the tip!
Well done mate. It just goes to show how complicated this process is. How Intel and AMD do the things they do blows my mind.
Super clever stuff! Very cool.
I've been meaning to take the next step in a similar project. Using the LCD for an sla 3-D printer to make pcbs. I've always been curious if I can get the features small and precise enough using a similar method with Optics.
Thanks for getting me motivated to go back and experiment with that project/concept some more.
amazing video, thank you! i learned so much
Man, this is CZcams gold, amazing!
Fascinating work!...cheers.
Awesome project man! 🤙
Really neat stuff! While I certainly understand how this works on a macro level, I'm not quite sure how some of the different components are interacting together to give the final output. I think I need to re-watch your explanation a few more times. As always with your videos, the topic and application of your ideas to achieve some final outcome is infinitely interesting. Thanks for sharing all of your hard work with us!
Incredable work, I’m so impressed buy your dedication and know how!
Really nice project!!! 👍👍👍
Amazing video Dude ! Very Impressive 🤩
Fantastic work and amazing video 👍
Hey, confocal microscopy guy here! I'm super impressed with what you managed to do diy with a super low budget.
Before blaming galvos, try to see if the unreliability is due to vibrations: they can be absolutely awful, especially in a setup suspended from the table, and with those long 3d printed parts holding heavy optics and metal parts. If you can, try to launch an etching experiment remotely from a pc in another room, consider doing it at night if you are close to a busy road/railway, and if you are not at the ground floor. If the result is much better, it's probably you whole setup wobbling about as you walk, talk, even breathe around it. Better yet, use your extra focusing optical path behind the dichroic to get a microscopy image of a sample on a camera with some decent frame rate (30 fps is enough), if you have vibrations, it will be extremely noticeable.
Also, you can gain a lot of power (if you need it) by focusing the laser on the pinhole, instead of just placing the pinhole in the collimated beam path. As long as the resolution of the lens you use is worse than the size of the pinhole, you should still get a very clean beam out of it, it will just be more finnicky to align.
Thanks for the tips, appreciate getting help from actual experts! Yeah I definitely think vibration is playing a role, although I might have underestimated how much. v2 will probably sit on a granite plate, and I might put some kind of air damper underneath to help with vibrations just to rule it out. That's a clever tip with the camera, will keep that in mind!
And re: laser, I probably explained it poorly. The diode has an integral lens which is focused onto the pinhole, so the efficiency isn't as bad as it seems. But I have the pinhole intentionally misaligned because the aberration from a misaligned pinhole ended up giving better results than full intensity through the too-large-pinhole. That's why the ultimate spot ended up looking so dim. :)
This is the type and level of diy we need to future proof agents SHTF! The more that know this stuff the better off and safer we are.
Bless you . Salute .
Great explanation with animations at 9:30. I'd be interested to see a version with Huygen's spheres overlayed onto the video
yep, that was amazing
huygens spheres are problematic with singular rays that can't actually exist. as in, they only exist because the beam is much thicker than the wavelength but its wave structure is inconvenient to visualize and it also causes inevitable diversion over distance.
The feature size of the defects caused by the aberrations is pretty impressive. I wonder if at some point real photolithography machines will be based on the aberrations we made along the way.
Oh you'll love this: sub-resolution assist features. Not quite the same, but these are features that are added to masks which help compensate for diffraction effects. What's neat is that they are smaller than the minimum feature size that can be made with the mask, but due to diffraction they still "add" to the feature and help fix problems or adjust the final shape. Super neat stuff!
Years ago I had a set of piezoelectric galvos for very fine deflection.
Cant wait to see updates on this project. If you get it run properly someday, you could think about trying to mode-lock a diode laser for fs-pulses and go for two-photon-polymerization, basically enabling full 3d printing.
Amazing you got such good results with 3d printed parts!
I was pretty shocked to be honest, much better than I expected!
Have you seen Robin Renzetti on CZcams? He mostly does very precise machining, stuff like machine spindles. He also machines his own surface plates and measures them for flatness with an autocollimator. To do this, he made custom mirror mounts, and I think you'd like his kinematic mounts. He made a custom differential screw to get better resolution when adjusting the mirrors. Maybe something there would help you out?
Very nice videos, keep them up!
I love Robin! That man has so much knowledge packed into his head it's amazing. Not sure I saw the custom mirror mounts though, will go dig through his videos to find it. Thanks for the tip!
Great start
cool prototype. looks like it could benefit from a solid steel structure and an enclosure so you can temperature control it and eliminate air currents and so forth
Incredible!
What could help you to get steeper walls in your etched layer would be to use a microscope optic with a longer focal length so the cone of light that hits your substrate is allot thinner/sharper and doesn't spread out as much towards the top.
Also planning on building something like this but instead of galvos I'll use glass cube prisms that displace the beam when rotated.
Will adapt your solution of using a microscope lens for the final focusing though, very smart.
This is amazing
Super cool!
Wonderful! What made you choose Galvos instead of a DLP chip like Sam Zeloof uses?
Amazing stuff.. if long plastic parts are creeping i can suggest mounting them with a long screw going through the long axis and pressing the part to the extrusion.. maybe it will provide a better stiffness.
kind of wicked. im bussy with this kind of stuff, just now :) its faaar more then i need. but hell very interesting !!
Damn! That is a really fantastic prototype! I cant wait to see how this thing does once you've gotten it strapped to a ground cast iron base and weighs 100lb. Might even outdo the resolution of the xy table.
Good call. I bought a 250 kg cast iron surface table to use on vibration-isolated mounts for this sort of experiment. It needs a rustproof optical bench plate bolted and glued to the top. Even just using air bag supports, the thing is ridiculously stable. A quarter of a ton lump in the lab is a little inconvenient as it needs an engine crane to move it around. I have some granite surface tables as well, but drilling and tapping those needs epoxy anchors. I can drill and tap directly into the cast iron table.
Another very interesting project! I am amazed by your project diversity.
When I saw the issues you have with the galvos I wondered why not use a DLP (as others apparently did). After you mentioned you got the galvos from a 3D printer I first thought about the LCD printers. So this might be an option to increase reproducability. You could use a cheap chinese resin 3D printer (or the optical spare parts) and use 2 lenses to collect as much UV from the LCD as possible (you'd loose a lot but then again expose everything at the same time and I assume exposure time is not an issue compared to all the other prepping time). Probably those UV LEDs output way more UV power than the laser. First lens relays the picture of the LCD to a second lens (e.g. used 50 mm prime camera lens: cheap on ebay and the have smooth focusing mechanics) and second lens collimates for the microscope lens.
Oh and another nice advantage of the LCD approach: you could use rather simple image manipulation algorithms for corrrect for light intensity falloff at the edges and even improve contrast on thin features/over etching in areas where two lines converge.
Side note: Do you know about this software for doing (simple) optical design simulations clles WinLens by Qioptiq?
Awesome project. I don't know much about lasers, but could you use blu-ray laser with its lense from the blue-ray burner/reader to do the same as you did? I think the holes that they burn in to discs are about 130nm wide?
Could try sitting it on a paver and dense foam for isolation to rule out at least some of the vibration. Surprising such a hodgepodge setup even works, let alone with results that are kind of promising.
Amazing 👍👍
Awesome work! Always love seeing your projects. Do you know what % thruput you're getting after the pinhole? Id expect somthing much higher than what I think you're getting (pending effort of alignment) unless the diode is multi transverse mode... then high loss is the name of the game
A very small amount, maybe 5-10%? There was some extra explanation but it rambled too much and I cut it from the edit: I have the pinhole intentionally misaligned by a small amount which is cutting down intensity a lot. The extra modes I was getting when correctly aligned (3-4 extra "rings" around the center spot) were enough to cause issues under the 40x objective... lots of "roughness" on the edge of the exposed line. So I ended up misaligning to cut a lot of the junk out. The circularity isn't quite as good, but there's a lot less "leakage" (apologies for the made up terms here).
It also worked out well because the resist is _super_ sensitive and needed reduced intensity. V2 will have a correctly sized pinhole and an ND filter to do it correctly :)
Impressive!!!
It looks very cool!
Maybe if you use thin wire to suspend the mirrors and a magnetic or electro-actuator, you'll get a much better result.
ASML in shambles EUV in danger haha I love it. As janky is looks very good work sir.
Really nicely done. I use SU-8 spinned on a Si wafer in combination with a 365nm Led and a pattern printed on a transparancy (outsourced to a specialized printing company cost around €20 for an A4). Placed the transparancy on the wafer with a thick piece of UV tranparant glass. Expose and develop. Very high aspect ratio's possible with a spatial resolution down to 3-5 micrometers.
Dr Nick, can you share the company that you used to print the masks?
@@arguetav ProArt BV in Groningen, The Netherlands. Ask for film with high density in black/white. Transparant for UV and black silver blocks UV. Resolution in micrometers. Just give them a call and discuss your needs.