@@wormball it was actually a test, I pulled the English text through Google translate. But apparently the translation sucks, as other russian viewers have also remarked before you.
I was so nervous after watching Sam Zeloof that there would be only one person on the planet working on such subject matter. Wow! I’m so happy that I was wrong! I can’t wait to find more revolutionizing the accessibility to this enigma of technology. Great work and presentation!
yeah, it would be very cool if Huygens Optics would put project BOM, CADs and description at hackaday - it definitely have chance of gaining community attention
I will look into this Hackaday functionality, although I'm not sure where to find it on the website. Hackaday has featured a number of my video's in the past and I think the blog really features highly interesting subjects. By the way, I'm not really focussed on getting as many subscribers or views as possible. I'd rather just reach people that are genuinly interested in optics. So if that is just a limited number, that is fine with me.
@@HuygensOptics I honestly believe it would get you both kinds(optics people and subscribers) of attention. Hackaday is all about electronics, embed programming and DIY manufacturing processes(including re-purposing old used hardware) and your project ticks all checkboxes for the platform and more importantly people on it.
That Censor lens brought back memories. I used to work for Perkin-Elmer in the mid-80's as a young optical engineer on the SRA wafer steppers both before and after they bought the Censor company in Liechtenstein. Stitching together multiple fields is a real challenge. Here are some issues you may have already thought about. If not I hope this helps. The reticle (or DMD in this case) has to be rotated parallel to the stage movement within a small fraction of a pixel width to avoid jags at the field boundaries. Hopefully the X and Y axes of your exposure stage are perpendicular to this same level. The asymmetric distortions like keystone also have to be corrected to a similar degree for stitching. You may need to tilt the DMD to dial this in. Illumination uniformity also has to be incredibly good to avoid noticeable line width changes at the boundaries. Simply Kohler illumination might not be quite good enough, you may need a light pipe uniformer or fly's eye integrator. At this NA, you will probably need to focus at each field. There will always be an offset between the focus detector and the best focus determined by an exposure/focus matrix that will have to be determined every few hours. Since it must be calibrated anyway, this means that you potentially could use a dichroic mirror in place of the beamsplitter, and use a separate wavelength for focusing and not waste the 405 nm light.
Hi Adam, that is some very good advice! Some things I had taken into consideration (like perpendicularity and field rotation), others not in that much detail. Illumination uniformity is actually one of those aspects that is quite tricky, I consider using only part of the total HD field for exposures, because the 16x9 aspect ratio of the chip is far from ideal. And it would not be a big problem to correct the step size for this in the software when in stitching mode. As for focussing, I actually designed a height map feature in the software, but only in first order (so just linear tilts in X and Y). So I hope this is sufficient for holding focus. Anyway, I'm not planning to expose 300mm wafers, but rather limit the field to maybe 10x10mm. But in general, I try not to consider every problem I might encounter in the future, but rather tackle it when I get there. Your description of what lies ahead is greatly appreciated though.
Wow, i'm researching how to build a high resolution LDI machine for PCB development on a budget. Applied science linked your channel, i'm glad they did..
Generally you can focus dlp-projectors very close by, making it possible to project high-resolution images without a single modification to the projector, in an area of for example 50x100mm. The light generally also contains quite a lot of uv, which allows for dry-film photoresist exposures.
@@HuygensOptics Indeed, i'm looking at the ts pico projector development board, (small sized, 100$) www.ti.com/lit/ug/dlpu049c/dlpu049c.pdf?HQS=TI-null-null-mousermode-df-pf-null-wwe&DCM=yes&ref_url=https%253A%252F%252Fwww.mouser.com%252F&distId=26 Just remove the lightsource and replace it with an uv laser. I am also thinking of using belt instead of ballscrew in order to circumvent ballscrew non linearity issues... But in any case,.. goed gedaan man !! Love your channel, very inspiring..
In the case of a development board, you could use a 50mm lens or so for high resolution projection. By the way, if you use a laser instead of an LED source, you can encounter diffraction effects on the DMD-chip, because of the coherent nature of the light. This may cause non-uniform projection intensity.
@@HuygensOptics Thank you for the tip of the lens, I will certainly read up on the topic of diffraction effects. I actually think I have one lying around. Thanks....
Great work! I have watched several vedios from your channel. In university, my major was optics, but we did not learn how to make lens, just studied theory. Very appreciate for your video， now i want to learn optics again. I think dvd optical pickup head is deserved to be studied, it is cheap and easy to get
Those DMD chips are very useful. I used them when I built my own two-photon lithography tool about 4 years ago. It was capable of printing 100 nm lines with an 800 nm laser. However, I don't know how to do movement very well so I kept the microscope stage stationary.
Hi, I bough s-planar lens from ebay that supposedly is 1:10 but text on it clearly states 1:1,6 f=50mm. Is there any way to verify if datasheet matches description? Also what does f number mean in case of constant focal length lens Is it effective focal length?
It's an F/1,6 and yes it is designed for 1:10 magnification and 1500 lpm. See more info on this page: www.marcocavina.com/articoli_fotografici/Zeiss_cute_DFR_DDR_lenses/00_pag_English.htm
I admire the effects of your work. Please forgive my ignorance in the field of optics but I have a question regarding the optical scheme of your system presented in 8:35 of your video. Why is there no tube lens in the path between the DMD and the microscope objective? I saw tube lenses in all the diagrams using the infinity corrected microscope objectives, but these were microscopic diagrams, not projectors. Also in video microscopes. What is the difference? My plan is to build a stepper similar to yours. I have collected a Nikkon Plan Apo 20x/0.75 DIC N2 INF/0.17 WD1.0 objective bought on ebay at a good price and also identical to your piezoceramic stage with the driver. I also have a small DC motors driven 2-axis 25x25mm Ficontec/National Aperture stage with Renishaw RGH25/RGB25H linear encoders with 50nm resolution so the stage can operate in closed loop feedback. I do not need a resolution as high as you, I think 2.5 to 5 micrometers would be sufficient as I am interested in preparing substrates for microwave thin film systems on Al2O3 substrates. However, in my application it is important to combine individual images into one larger one. Ultimately, I wanted a 50x50 mm working area but I found a 25x25 mm stage and I think I'll start with that. I actually have 2 of these stages and the first idea was to stack them on top of each other to get a 50x50mm working area, but I will rather start testing with a single stage. When it comes to DMD, I was thinking about a ready module with 405 nm backlight. I was looking at EKB modules (one of the TI DLP design houses). Ready modules with around 95% illumination uniformity, 2.5 W LED 405 nm or 1.5 W LED 385 nm. Not the cheapest ($ 1,600 +), but maybe without the projection lens it would have been a bit cheaper. By entrusting this part of the project to specialists, I would increase the chances of success at the expense of higher costs. As always...
Sounds like you are all set to make your own wafer stepper! As for the tube lens: others have asked me the same question. In principle you are right about the objective requiring a tube lens, and under some circumstances you could encounter field deformation or spherical aberrations. However, in the configuration used, the distance between objective and DMD is quite large (180mm). And in practice you do not notice any field deformation at all when leaving it out. (You might when you go to into the submicron scale and start stitching, but evidently that is not your goal). Also, since placing and aligning the lens caused additional issues given the current geometry, and since the exact position had a large influence on magnification, I decided not to use it after all. I have one suggestion for you to improve your build: add the Z-translation to the wafer stage and not to the objective. It will be way easier to check your focus. Good luck with your project, it would be great to hear about your progress and results.
@@HuygensOptics Thank you. I appreciate your advice very much, such practical knowledge is hard to find in books and scientific articles. In fact, in addition to two pcs. X/Y stages, I also bought a Z-axis stage with the same encoder with 50 nm resolution and 12.5 mm range. And also a rotating stage with a diameter of 65 mm with a hole in the center, which allows me to easily bring the vacuum to the wafer holder, but with the encoder on the axis of the motor and not on the axis of the stage. Maybe I will actually stack all these stages to be able to operate on (almost) all the axes I need. There was also a small theta axis stage included, but I rather assume that a single wafer projection area will be flat and will fit within FOV and no additional tilt will be required. The driving part is easier for me to develop because it is related to what I do professionally. I have a long but very fascinating road ahead of me with this project.
Unfortunately it is probably impossible to make metasurface from a DMD projected through an objective without DUV. What sets metasurface apart from conventional diffractive optics is each scatterer are subwavelength and they provide a phaseshift whit thickness less than 1 wavelength, they behave more like waveguide and resonators. Comparing to conventional diffractive optics often has macroscopic features that are several wavelength in size and thickness of several wavelength as well. The way the optical wavefront is shaped with metasurface is that spherical wave from each individual scatterer with different phase create the desire wavefront at the farfield. So each element (they are dielectric waveguide actually) typically range from 70-150nm in diameter. Maybe you can make something with a DUV stepper, but optical system in that wavelength will be very expensive. So optical stepper is probably more suites for conventional diffractive optics than metasurface.
The microscope objective lens you are using seems to be corrected for cover glass thickness, as this is typically the case for most CFI Nikon Plan Apos, especially those with higher magnifications. I wonder if you see any noticable spherical aberration if used without cover glass. Have you measured the point spread function of this system so far?
Thanks for the suggestion Jaka, I actually did not know that, so I checked the documentation. The Plan Apo VC 20x that I use does not have a cover glass correction. I have not measured the optical quality of the objective myself but I probably should. The main issue so far seems to be a rather limited contrast, probably caused by the "APO" properties, which I do not need using only a limited wavelength range. I improved the contrast quite a bit by introducing a few limiting aperture stops.
Sure. You can download it from the nikon website. www.nikon.com/products/microscope-solutions/support/download/brochures/pdf/2ce-muzh-4.pdf. The objective I'm using is the MRD70200
Thanks to both of you for discussing this issue. A photomacrography forum has been discussing this issue in relation to your work. I don't see any specs on the Nikon website for the particular objective you are using. The Nikon 20x 0.75 VC is clearly labelled for 0.17mm coverslip, even in the diagram that you used in your video. The OEM versions are generally optimized for different (usually thicker) "coverslips" (often glass-walled receptacles in gene sequencers). If you have information that your particular model does not need a coverslip, that would be very interesting and exciting news for many of us. Can you explain in more detail why you think your objective does not need a coverslip? Thanks.
@@loujost Thanks for the info. So, indeed a cover glass is indicated in the figure, but I am not sure that means it is corrected for this. However, check out page15 of the Nikon document linked above. With the 60x, the correction is explicitly mentioned. The 100x oil emersion does not need it, so it is logical it's not mentioned. With the 20x there is no mention of cover glass correction. I'm no expert at all, so I might very well be wrong about this. PS: if the cover slip correction is mainly chromatic, it is not important for my application since I use monochromatic light.
If you use monochromatic UV source, maybe you don't need APO lens? APO lens I think are useful only if you want to reduce chromatic aberration. But if you use monochromatic light, that is I think unnecessary, and can reduce other aspects of the image quality, like field curvature, coma and contrast, due to the design tradeoffs. Achromatic / Apochromatic lenses surely are useful in the inspection of the wafer tho using external secondary light. High quality apochromatic objectives can be pretty expensive and have 15 lens elements, which does reduce microcontrast and light transmission in general, so it is probably better to avoid them if possible. You will notice on your huge Zeiss S-Planar, that it is marked as 405nm. It is optimized for this specific wavelength, and it is not achromatic / apochromatic, maybe just slightly, but really in very narrow range 390nm-420nm maybe. The complex construction inside is mostly to improve the field flatness, reduce geometric deformations, reduce / eliminate spherical aberrations, vignetting, and reduce effects of colimation errors and thermal expansion.
Actually, I bought the objective for the "plan" indication, not so much for the "apo" properties. But as you also suggested, it could have advantages when using a secondary wavelength, which is outside the exposure range of the photoresist. For example when checking the focus through the same objective, before doing exposures. So far, loss of contrast and field curvature/coma do not seem to play a large rol at this point, but I guess we have to find out. Anyway, thank you for the valuable suggestions. By the way, did you see the video's specifically made on the S-planar? In the first one I discuss the 405nm design wavelength and the implications for chromatic aberration (around 5.30min).
@@HuygensOptics I just noticed your other videos on photolitography, and you clearly know about the design tradeoffs and more than me on the topic. :D Thanks and keep us posted on the progress of your project! Yes, I just watched the S-Planaer videos, they were very cool and indeed, it is sharp and optimized for 405nm. The chromatic aberration is enormous in it, but it is by design :D
Is it possible to make a beam splitter that splits it non-uniformly, sending, say, 90% of the light one way and 10% the other? I guess it would be quite a custom part, but it could get you better exposure times, since I guess you don't need much light to focus by.
You can do this projector using a UV led array, a cheap screen masking lcd like the ones used in 3d printers and lenses to project the image into the microscope objective instead of a wall. Like making a cheap diy projector. That could give you even 4K in resolution.
The work presented is commendable and described with a remarkable level of quality! I do have a question. What are the specifications of the beamsplitter you used? Thank you.
@@HuygensOptics I've watched it since, must say I was so enthused that I had replied on the spot. I've done a bit of micro fabrication myself, including microoptics (Fresnel lenses mostly) , but using photoreduction or metal masks, not optical direct writing : your setup is amazing (caveat : resolution and alignement tends to always become a limitation at some point or other, so it is a major design parameter in my experience). Though I'm mostly into mems and biosensors these days, this kind of tool would be very valuable, at least for prototyping... One minor suggestion if I may : you could probably improve your defect density by building yourself a little air box to prepare the plates and box them before taking them to the patterning machine.
Thank you for a very nice video. One question I would like to ask: If I replace the light source in Acer DLP projector with a UV light source, do I need to put a dummy signal in the light source driver to make it look like the original projector light source is working? I would be grateful for your guidance on these procedures.
Good question, I forgot to discuss this in my video. Unlike other brands, where one can just modify an opto coupler, It's difficult to trick the Acer p1500 into thinking a lamp is present, because it uses serial communication between the lamp driver and the main PCB. The easiest way to fool the Acer is to connect a +/-1000W ceramic cooking plate or a heating element to the 2 lamp wires. These can easily dissipate the 100W of lamp power and will trick the Acer into thinking a lamp is present.
3:54 What kind of software or source code do you use to generate the input data (such as on/off mode and gray level of each mirrors) for maskless optical proximity correction?
I do not have software for that (yet). In the video I just mentioned that it is possible. But if you just use simple images for exposures, one can easily use for example Photoshop to generate the specific mask patterns to test the feasibility and the effects.
can we use this technology to reduce the exit pupil size in a telescope with a large aperture and focus all the light into the eye pupil, to bring out the color of galaxies and nebulae, so that no light is wasted outside the eye pupil?
It was hard to get the outer edges at the same intensity as the center so I generally use the central area of 1080x1080 instead of the total 1920x1080px.
First thanks for the videos, But I'm a bit confused about the UV light source. Was it a UV laser or an array of UV LEDs? and if you don't mind how powerful should the light source be?
It is just one 3W diode of 405nm with a surface area of a few mm2. The beam is collimated to approx. A few cm2 at the dmd-chip. Only a few tens of milliwatts makes it into the microscope, however the image is projected into a very small area so the wattage per area is still high.
Wow really cool project! You hear now also in the cinemas e.d. about laser projectors provdiing better contrast. Would using a 405nm laser have any advantages for your system?
The high resolution (4K for some systems) could be advantageaous. Also it would probably be quite easy to replace one of the lasers with a powerfull 405nm laser. I have mixed experiences with using laser light in photolithography. One of the problems I encountered is the fact that you can get "grainy" images when reflecting monochromatic light on a DMD-chip.
"producing mirrored Poynting vectors", is that a euphemism for "watching naked women on motor cycles"? Just kidding, I don't know what the end result will be. For me this project is about the fun of builing and making. And although the general plan is to make micro-optics, I think the detailed design of meta surfaces and writing the software for that is way over my head. So if you have any suggestions in this field, they are very welcome. The first things I will try to make will be miniaturized versions of conventional lens designs.
@@HuygensOptics Following a normal design process could result in all the fun being drained out of this very enjoyable series! But I believe that you could pick up quickly on design by following some of the examples available on github of resonator analysis, some of which utilize the python 'meep' module, others use Matlab. I'd even bet that you could duplicate a published long wave IR resonator and observe results if you have access to a bolometer imager. This is just my contribution to the universal internet world of lousy advice from strangers, as in 'caveat emptor'. I really enjoy your channel!
@idical idical, thanks for the references and advice. So IR is probably the only application that I can make actual devices for considering the rather limited resolution of setup. To make devices for visible light you need to go far beyond the 400nm resolution limit and I'm not sure if I can extend the resolution to let's say 50nm by just using multiple exposure steps. Using e-beam lithography would be a more logical choice for making devices in the visible wavelength range.
Can you please share some details about the UV source and the condensing optics as I understand you need something close to a point source for DMD chip.
What you need is a highly colimated beam of light incident on the DMD. you can use a point source and a collimating lens or mirror to achieve this. What I used was an LED, then a short focal distance lens and then an additional collimating mirror that reflectes the light onto the DMD at the right angle and focusses the light into the microscope objective. I need to get as much light into the narrow opening of the microscope objective. It is a bit like I explained in this video: ruclips.net/video/E9uU2h1Ldcg/видео.html (starting at around 8 minutes)
@@HuygensOptics Thanks for the explanation, its really helpful. I ran into this problem a few years ago when I tried to convert a DLP projector for mask-less PCBs . I might revisit it.
@@HuygensOptics Thanks for the explanation, its really helpful. I ran into this problem a few years ago when I tried to convert a DLP projector for mask-less PCBs . I might revisit it.
Wow. I'm clearly no optical engineer at all, but I heard that certain 3D optical structures can redirect light around objects. Could you make structures like that with this machine, or would that require smaller nanometer size objects? Another difficulty might be that the structures may require a 3D shape to gradually shift the light around. I'm not sure how you'd get that smooth stepping. Varying intensity levels or exposure times? Haha. I'm probably missing something, but either way- it would be super cool to see.
I'm curious about the projector teardown. Seems like a cheaper and less stressful way that having to assemble all the various hardware needed for a TI dev kit. Do you have any images of the projector that you can share? Was it easy to get access to the DLP mirror?
Unfortunately, I do not have any photos of the disassembly. However, getting access to the dmd chip was pretty easy because generally these chips can be replaced if they break. In the case of this projector, I had to modify the part holding the optics to get free access to the dmd chip.
@@HuygensOptics thank you. I’m also trying to get a dev kit from DLi but my application is fairly straightforward so this seemed to be a quick solution.
how can you possibly see such a fine details (half a micrometer) using optical microscope and visible light? i thought such details, being comparable to the size of the light wavelength should be invisible
Hey, love your video, just a quick question. With the Microscope objective how did you get it close enough to the DMD While equally allowing the light source to hit the DMD at the same time? Or what kind of "Condensing Optics" did you use? When ever I use my DMD it detracts orders all over the place, even with collimated light, so how did you get the image to focus so far away from the DMD itself? Any info will be greatly appreciated I've been racking my brain with this problem!
The objective is actually infinity corrected, so you need a "tube lens" as a corrector between dmd and microscope objective. Unfortunately, that part was introduced later in the machine so it is missing in the video. Generally this is a lens with 150-200mm focal distance, with that distance to the dmd .
@@HuygensOptics Do you happen to have the make of thew tube lens you used? Was it from Thorlabs or something similar? sorry for questions but thank you for answering me first question your video is great!
@@danielblackmore4718 yes, the Thorlabs or Edmund tube lenses work fine. But in fact, more "regular" lenses also work well. Just as long as they have limited spherical aberration, generally meaning a fairly high f-number (say F/6 or higher), I guess you are safe.
@@HuygensOptics I'm really sorry but I'm still a bit confused, so did you manage to make a tube lens system with individual lens? if so what kind did you use? I've currently got a 100mm to a 50mm bi-convex in a kinda 4f system to demagnify the image down to the microscope. But the DMD Diffracts all the orders around too much so I never get a clear image, so did you have a similar set up of lens after the DMD before the Microscope? Do you remember what lenses you used? Sorry again for all the questions thats my last one I promise!
@@danielblackmore4718best look on this page and look for infinite conjugate. It shows how the tube lens works in a microscope. In the dmd projections it works in reverse where the dmd is in the focal plane of the ocular. A tube lens eliminates the requirement for a fixed tube length, because of the infinity projection. www.edmundoptics.eu/knowledge-center/application-notes/microscopy/understanding-microscopes-and-objectives/
Hello! @Huygens Optics I am doing a similar project on maskless lithography like yours. But I really lack the knowledge of programming. I am wonder if you had uploaded your control program on GitHub or could you please teach me how to project the picture with the projector. (is it just using an hdmi wire or by program)
Awesome. What is the minimum feature width you think you will be able to achieve? Also I would be interested in your thoughts on thermal stability, particle contamination and vibration considerations.
Hi Nick, regarding the resolution: With 193nm it is now possible to produce dimensions below 50nm using multiple exposures and very advanced mask design. So if you would use a similar factor to 405nm, a 100nm should be achievable. But of course, my "mask" is very crude and it is not possible to deliberatly introduce interference effects to enhance resolution, so it would be quite hard to achieve. Bottom line is: I don't know at this point. I guess we just have to find out. Thermal stability and vibrations will be discussed in the next video on mechanics. Please be patient, it will take a few weeks before that video is ready.
@@HuygensOpticsyou don’t have masks. You only would need to change the pattern. So easy. Mount 4 polarised light sources around your DLP (lasers are polarised). Turn on top, bottom for features in one direction, change pattern, turn on left right lights. Or rotate around and animate DLP ?? Oh I see. All needs to be off axis.
I did this in my lab a few years ago. I will be interested in how well you can do field stitching. I actually made it so I could swap optics and align though the lens with red light before switching to nUV. I had much more field distortion than a real stepper though.
Sometimes you find gold for free....like I found your channel on RUclips. If only people knew how exciting your work is... great work, sir.
Let's hope you can keep a secret, I don't want you to cause a gold rush ;-).
@@HuygensOptics 😄😄
@@HuygensOptics whoever translated your video into russian, he/she/it does not know russian or english, or perhaps both.
@@wormball it was actually a test, I pulled the English text through Google translate. But apparently the translation sucks, as other russian viewers have also remarked before you.
This is seriously one of the most fascinating channels I've found on RUclips
I was so nervous after watching Sam Zeloof that there would be only one person on the planet working on such subject matter. Wow! I’m so happy that I was wrong! I can’t wait to find more revolutionizing the accessibility to this enigma of technology. Great work and presentation!
sudden jump of ebay DLP projector prices soon 😅
And Nikon Plan Apo's.
Impressive! You present with such clarity, it's very engaging, despite optics not being my field. Your channel is severely underrated.
yeah, it would be very cool if Huygens Optics
would put project BOM, CADs and description at hackaday - it definitely have chance of gaining community attention
I will look into this Hackaday functionality, although I'm not sure where to find it on the website. Hackaday has featured a number of my video's in the past and I think the blog really features highly interesting subjects. By the way, I'm not really focussed on getting as many subscribers or views as possible. I'd rather just reach people that are genuinly interested in optics. So if that is just a limited number, that is fine with me.
@@HuygensOptics I honestly believe it would get you both kinds(optics people and subscribers) of attention. Hackaday is all about electronics, embed programming and DIY manufacturing processes(including re-purposing old used hardware) and your project ticks all checkboxes for the platform and more importantly people on it.
That Censor lens brought back memories. I used to work for Perkin-Elmer in the mid-80's as a young optical engineer on the SRA wafer steppers both before and after they bought the Censor company in Liechtenstein.
Stitching together multiple fields is a real challenge. Here are some issues you may have already thought about. If not I hope this helps. The reticle (or DMD in this case) has to be rotated parallel to the stage movement within a small fraction of a pixel width to avoid jags at the field boundaries. Hopefully the X and Y axes of your exposure stage are perpendicular to this same level. The asymmetric distortions like keystone also have to be corrected to a similar degree for stitching. You may need to tilt the DMD to dial this in. Illumination uniformity also has to be incredibly good to avoid noticeable line width changes at the boundaries. Simply Kohler illumination might not be quite good enough, you may need a light pipe uniformer or fly's eye integrator. At this NA, you will probably need to focus at each field. There will always be an offset between the focus detector and the best focus determined by an exposure/focus matrix that will have to be determined every few hours. Since it must be calibrated anyway, this means that you potentially could use a dichroic mirror in place of the beamsplitter, and use a separate wavelength for focusing and not waste the 405 nm light.
Hi Adam, that is some very good advice! Some things I had taken into consideration (like perpendicularity and field rotation), others not in that much detail. Illumination uniformity is actually one of those aspects that is quite tricky, I consider using only part of the total HD field for exposures, because the 16x9 aspect ratio of the chip is far from ideal. And it would not be a big problem to correct the step size for this in the software when in stitching mode. As for focussing, I actually designed a height map feature in the software, but only in first order (so just linear tilts in X and Y). So I hope this is sufficient for holding focus. Anyway, I'm not planning to expose 300mm wafers, but rather limit the field to maybe 10x10mm. But in general, I try not to consider every problem I might encounter in the future, but rather tackle it when I get there. Your description of what lies ahead is greatly appreciated though.
@@HuygensOptics at 16x9 you can try doing anamorphic lenses to get bigger NA
I'm not very young, still you sir, you are my Santa. And every month when your video comes up is my Christmas.
We said one man ARMY, and you are one man ASML ! This is totally in same engineering skills!
Wow, i'm researching how to build a high resolution LDI machine for PCB development on a budget. Applied science linked your channel, i'm glad they did..
Generally you can focus dlp-projectors very close by, making it possible to project high-resolution images without a single modification to the projector, in an area of for example 50x100mm. The light generally also contains quite a lot of uv, which allows for dry-film photoresist exposures.
@@HuygensOptics Indeed, i'm looking at the ts pico projector development board, (small sized, 100$) www.ti.com/lit/ug/dlpu049c/dlpu049c.pdf?HQS=TI-null-null-mousermode-df-pf-null-wwe&DCM=yes&ref_url=https%253A%252F%252Fwww.mouser.com%252F&distId=26
Just remove the lightsource and replace it with an uv laser.
I am also thinking of using belt instead of ballscrew in order to circumvent ballscrew non linearity issues...
But in any case,.. goed gedaan man !! Love your channel, very inspiring..
In the case of a development board, you could use a 50mm lens or so for high resolution projection. By the way, if you use a laser instead of an LED source, you can encounter diffraction effects on the DMD-chip, because of the coherent nature of the light. This may cause non-uniform projection intensity.
@@HuygensOptics Thank you for the tip of the lens, I will certainly read up on the topic of diffraction effects. I actually think I have one lying around. Thanks....
this is an amazing project. Genuinely impressed
What your doing is fascinating and its helping me understand things with real world examples. Thanks for posting.
Great work! I have watched several vedios from your channel. In university, my major was optics, but we did not learn how to make lens, just studied theory. Very appreciate for your video， now i want to learn optics again. I think dvd optical pickup head is deserved to be studied, it is cheap and easy to get
Those DMD chips are very useful. I used them when I built my own two-photon lithography tool about 4 years ago. It was capable of printing 100 nm lines with an 800 nm laser. However, I don't know how to do movement very well so I kept the microscope stage stationary.
Wow, you seriously perked my interest in optics. Incredible project.
Great channel! Thank you for mentioning the surplus website in the other video!
Hi, I bough s-planar lens from ebay that supposedly is 1:10 but text on it clearly states 1:1,6 f=50mm. Is there any way to verify if datasheet matches description? Also what does f number mean in case of constant focal length lens Is it effective focal length?
It's an F/1,6 and yes it is designed for 1:10 magnification and 1500 lpm. See more info on this page: www.marcocavina.com/articoli_fotografici/Zeiss_cute_DFR_DDR_lenses/00_pag_English.htm
Fascinating stuff. Thanks for explaining and sharing your knowledge.
I admire the effects of your work. Please forgive my ignorance in the field of optics but I have a question regarding the optical scheme of your system presented in 8:35 of your video. Why is there no tube lens in the path between the DMD and the microscope objective? I saw tube lenses in all the diagrams using the infinity corrected microscope objectives, but these were microscopic diagrams, not projectors. Also in video microscopes. What is the difference? My plan is to build a stepper similar to yours. I have collected a Nikkon Plan Apo 20x/0.75 DIC N2 INF/0.17 WD1.0 objective bought on ebay at a good price and also identical to your piezoceramic stage with the driver. I also have a small DC motors driven 2-axis 25x25mm Ficontec/National Aperture stage with Renishaw RGH25/RGB25H linear encoders with 50nm resolution so the stage can operate in closed loop feedback. I do not need a resolution as high as you, I think 2.5 to 5 micrometers would be sufficient as I am interested in preparing substrates for microwave thin film systems on Al2O3 substrates. However, in my application it is important to combine individual images into one larger one. Ultimately, I wanted a 50x50 mm working area but I found a 25x25 mm stage and I think I'll start with that. I actually have 2 of these stages and the first idea was to stack them on top of each other to get a 50x50mm working area, but I will rather start testing with a single stage.
When it comes to DMD, I was thinking about a ready module with 405 nm backlight. I was looking at EKB modules (one of the TI DLP design houses). Ready modules with around 95% illumination uniformity, 2.5 W LED 405 nm or 1.5 W LED 385 nm. Not the cheapest ($ 1,600 +), but maybe without the projection lens it would have been a bit cheaper. By entrusting this part of the project to specialists, I would increase the chances of success at the expense of higher costs. As always...
Sounds like you are all set to make your own wafer stepper! As for the tube lens: others have asked me the same question. In principle you are right about the objective requiring a tube lens, and under some circumstances you could encounter field deformation or spherical aberrations. However, in the configuration used, the distance between objective and DMD is quite large (180mm). And in practice you do not notice any field deformation at all when leaving it out. (You might when you go to into the submicron scale and start stitching, but evidently that is not your goal). Also, since placing and aligning the lens caused additional issues given the current geometry, and since the exact position had a large influence on magnification, I decided not to use it after all.
I have one suggestion for you to improve your build: add the Z-translation to the wafer stage and not to the objective. It will be way easier to check your focus. Good luck with your project, it would be great to hear about your progress and results.
@@HuygensOptics Thank you. I appreciate your advice very much, such practical knowledge is hard to find in books and scientific articles. In fact, in addition to two pcs. X/Y stages, I also bought a Z-axis stage with the same encoder with 50 nm resolution and 12.5 mm range. And also a rotating stage with a diameter of 65 mm with a hole in the center, which allows me to easily bring the vacuum to the wafer holder, but with the encoder on the axis of the motor and not on the axis of the stage. Maybe I will actually stack all these stages to be able to operate on (almost) all the axes I need. There was also a small theta axis stage included, but I rather assume that a single wafer projection area will be flat and will fit within FOV and no additional tilt will be required. The driving part is easier for me to develop because it is related to what I do professionally. I have a long but very fascinating road ahead of me with this project.
Awesome work look forward to seeing more!
Nice work , congratulations.
Unfortunately it is probably impossible to make metasurface from a DMD projected through an objective without DUV. What sets metasurface apart from conventional diffractive optics is each scatterer are subwavelength and they provide a phaseshift whit thickness less than 1 wavelength, they behave more like waveguide and resonators. Comparing to conventional diffractive optics often has macroscopic features that are several wavelength in size and thickness of several wavelength as well. The way the optical wavefront is shaped with metasurface is that spherical wave from each individual scatterer with different phase create the desire wavefront at the farfield. So each element (they are dielectric waveguide actually) typically range from 70-150nm in diameter. Maybe you can make something with a DUV stepper, but optical system in that wavelength will be very expensive. So optical stepper is probably more suites for conventional diffractive optics than metasurface.
Amazing. Thanks for sharing. I have been eorking on image stitching for high resolution photo printing so got some good ideas from this.
Fascinating work you do !...cheers.
The microscope objective lens you are using seems to be corrected for cover glass thickness, as this is typically the case for most CFI Nikon Plan Apos, especially those with higher magnifications. I wonder if you see any noticable spherical aberration if used without cover glass. Have you measured the point spread function of this system so far?
Thanks for the suggestion Jaka, I actually did not know that, so I checked the documentation. The Plan Apo VC 20x that I use does not have a cover glass correction. I have not measured the optical quality of the objective myself but I probably should. The main issue so far seems to be a rather limited contrast, probably caused by the "APO" properties, which I do not need using only a limited wavelength range. I improved the contrast quite a bit by introducing a few limiting aperture stops.
@@HuygensOptics Thank you for your answer. Interesting - is it possible for you to share the documentation?
Sure. You can download it from the nikon website. www.nikon.com/products/microscope-solutions/support/download/brochures/pdf/2ce-muzh-4.pdf. The objective I'm using is the MRD70200
Thanks to both of you for discussing this issue. A photomacrography forum has been discussing this issue in relation to your work. I don't see any specs on the Nikon website for the particular objective you are using. The Nikon 20x 0.75 VC is clearly labelled for 0.17mm coverslip, even in the diagram that you used in your video. The OEM versions are generally optimized for different (usually thicker) "coverslips" (often glass-walled receptacles in gene sequencers). If you have information that your particular model does not need a coverslip, that would be very interesting and exciting news for many of us. Can you explain in more detail why you think your objective does not need a coverslip? Thanks.
@@loujost Thanks for the info. So, indeed a cover glass is indicated in the figure, but I am not sure that means it is corrected for this. However, check out page15 of the Nikon document linked above. With the 60x, the correction is explicitly mentioned. The 100x oil emersion does not need it, so it is logical it's not mentioned. With the 20x there is no mention of cover glass correction. I'm no expert at all, so I might very well be wrong about this.
PS: if the cover slip correction is mainly chromatic, it is not important for my application since I use monochromatic light.
Fantastic work!
Sir, you are a god. Keep up the good work.
If you use monochromatic UV source, maybe you don't need APO lens? APO lens I think are useful only if you want to reduce chromatic aberration. But if you use monochromatic light, that is I think unnecessary, and can reduce other aspects of the image quality, like field curvature, coma and contrast, due to the design tradeoffs. Achromatic / Apochromatic lenses surely are useful in the inspection of the wafer tho using external secondary light. High quality apochromatic objectives can be pretty expensive and have 15 lens elements, which does reduce microcontrast and light transmission in general, so it is probably better to avoid them if possible. You will notice on your huge Zeiss S-Planar, that it is marked as 405nm. It is optimized for this specific wavelength, and it is not achromatic / apochromatic, maybe just slightly, but really in very narrow range 390nm-420nm maybe. The complex construction inside is mostly to improve the field flatness, reduce geometric deformations, reduce / eliminate spherical aberrations, vignetting, and reduce effects of colimation errors and thermal expansion.
Actually, I bought the objective for the "plan" indication, not so much for the "apo" properties. But as you also suggested, it could have advantages when using a secondary wavelength, which is outside the exposure range of the photoresist. For example when checking the focus through the same objective, before doing exposures. So far, loss of contrast and field curvature/coma do not seem to play a large rol at this point, but I guess we have to find out. Anyway, thank you for the valuable suggestions. By the way, did you see the video's specifically made on the S-planar? In the first one I discuss the 405nm design wavelength and the implications for chromatic aberration (around 5.30min).
@@HuygensOptics I just noticed your other videos on photolitography, and you clearly know about the design tradeoffs and more than me on the topic. :D Thanks and keep us posted on the progress of your project! Yes, I just watched the S-Planaer videos, they were very cool and indeed, it is sharp and optimized for 405nm. The chromatic aberration is enormous in it, but it is by design :D
@@HuygensOpticsscanners only illuminate a circle sector to make curvature non-critical
Waiting to see more!
Is it possible to make a beam splitter that splits it non-uniformly, sending, say, 90% of the light one way and 10% the other? I guess it would be quite a custom part, but it could get you better exposure times, since I guess you don't need much light to focus by.
Actually not that custom: www.edmundoptics.com/f/plate-beamsplitters/12424/
CD players use QWP and a polarisating beam splitter
Just found your channel so cool
You can do this projector using a UV led array, a cheap screen masking lcd like the ones used in 3d printers and lenses to project the image into the microscope objective instead of a wall. Like making a cheap diy projector. That could give you even 4K in resolution.
The work presented is commendable and described with a remarkable level of quality!
I do have a question. What are the specifications of the beamsplitter you used?
Thank you.
would you not also require rotation adjustement for being able to stitch ?
This issue is actually discussed in Part 2 (rotation is mentioned around 11min 22 sec)
@@HuygensOptics I've watched it since, must say I was so enthused that I had replied on the spot. I've done a bit of micro fabrication myself, including microoptics (Fresnel lenses mostly) , but using photoreduction or metal masks, not optical direct writing : your setup is amazing (caveat : resolution and alignement tends to always become a limitation at some point or other, so it is a major design parameter in my experience). Though I'm mostly into mems and biosensors these days, this kind of tool would be very valuable, at least for prototyping... One minor suggestion if I may : you could probably improve your defect density by building yourself a little air box to prepare the plates and box them before taking them to the patterning machine.
Thank you for a very nice video. One question I would like to ask: If I replace the light source in Acer DLP projector with a UV light source, do I need to put a dummy signal in the light source driver to make it look like the original projector light source is working? I would be grateful for your guidance on these procedures.
Good question, I forgot to discuss this in my video. Unlike other brands, where one can just modify an opto coupler, It's difficult to trick the Acer p1500 into thinking a lamp is present, because it uses serial communication between the lamp driver and the main PCB. The easiest way to fool the Acer is to connect a +/-1000W ceramic cooking plate or a heating element to the 2 lamp wires. These can easily dissipate the 100W of lamp power and will trick the Acer into thinking a lamp is present.
3:54 What kind of software or source code do you use to generate the input data (such as on/off mode and gray level of each mirrors) for maskless optical proximity correction?
I do not have software for that (yet). In the video I just mentioned that it is possible. But if you just use simple images for exposures, one can easily use for example Photoshop to generate the specific mask patterns to test the feasibility and the effects.
The idea of metasurface optics reminds me a bit of the Fresnel lens.
This is incredible !!!
Kwam uw video tegen via het appliedscience Channel, interessant project!
can we use this technology to reduce the exit pupil size in a telescope with a large aperture and focus all the light into the eye pupil, to bring out the color of galaxies and nebulae, so that no light is wasted outside the eye pupil?
Great video! Thank you. Did the fill factor of the DMD chip cause any problems when projecting images?
It was hard to get the outer edges at the same intensity as the center so I generally use the central area of 1080x1080 instead of the total 1920x1080px.
First thanks for the videos, But I'm a bit confused about the UV light source. Was it a UV laser or an array of UV LEDs? and if you don't mind how powerful should the light source be?
It is just one 3W diode of 405nm with a surface area of a few mm2. The beam is collimated to approx. A few cm2 at the dmd-chip. Only a few tens of milliwatts makes it into the microscope, however the image is projected into a very small area so the wattage per area is still high.
@@HuygensOptics Thanks man, i really appreciate it. Keep up the great work. :)
Congrats for this amazing content. Can you share some info about lenses in the CCD Camera side?
Wow really cool project! You hear now also in the cinemas e.d. about laser projectors provdiing better contrast. Would using a 405nm laser have any advantages for your system?
The high resolution (4K for some systems) could be advantageaous. Also it would probably be quite easy to replace one of the lasers with a powerfull 405nm laser. I have mixed experiences with using laser light in photolithography. One of the problems I encountered is the fact that you can get "grainy" images when reflecting monochromatic light on a DMD-chip.
You can make a decoherence filter for lasers by shining them through a weak milk solution. This removes the graininess.
great job!!
Very cool!!!!
Regards, laser physicist!
The big company’s should pump millions to help you
There is little else on earth as thrilling as being able to produce mirrored Poynting vectors. Can you hint at your plans for the end result?
"producing mirrored Poynting vectors", is that a euphemism for "watching naked women on motor cycles"? Just kidding, I don't know what the end result will be. For me this project is about the fun of builing and making. And although the general plan is to make micro-optics, I think the detailed design of meta surfaces and writing the software for that is way over my head. So if you have any suggestions in this field, they are very welcome. The first things I will try to make will be miniaturized versions of conventional lens designs.
@@HuygensOptics Following a normal design process could result in all the fun being drained out of this very enjoyable series! But I believe that you could pick up quickly on design by following some of the examples available on github of resonator analysis, some of which utilize the python 'meep' module, others use Matlab. I'd even bet that you could duplicate a published long wave IR resonator and observe results if you have access to a bolometer imager. This is just my contribution to the universal internet world of lousy advice from strangers, as in 'caveat emptor'. I really enjoy your channel!
@idical idical, thanks for the references and advice. So IR is probably the only application that I can make actual devices for considering the rather limited resolution of setup. To make devices for visible light you need to go far beyond the 400nm resolution limit and I'm not sure if I can extend the resolution to let's say 50nm by just using multiple exposure steps. Using e-beam lithography would be a more logical choice for making devices in the visible wavelength range.
Can you please share some details about the UV source and the condensing optics as I understand you need something close to a point source for DMD chip.
What you need is a highly colimated beam of light incident on the DMD. you can use a point source and a collimating lens or mirror to achieve this. What I used was an LED, then a short focal distance lens and then an additional collimating mirror that reflectes the light onto the DMD at the right angle and focusses the light into the microscope objective. I need to get as much light into the narrow opening of the microscope objective. It is a bit like I explained in this video: ruclips.net/video/E9uU2h1Ldcg/видео.html (starting at around 8 minutes)
@@HuygensOptics Thanks for the explanation, its really helpful. I ran into this problem a few years ago when I tried to convert a DLP projector for mask-less PCBs . I might revisit it.
@@HuygensOptics Thanks for the explanation, its really helpful. I ran into this problem a few years ago when I tried to convert a DLP projector for mask-less PCBs . I might revisit it.
Wow. I'm clearly no optical engineer at all, but I heard that certain 3D optical structures can redirect light around objects. Could you make structures like that with this machine, or would that require smaller nanometer size objects?
Another difficulty might be that the structures may require a 3D shape to gradually shift the light around. I'm not sure how you'd get that smooth stepping. Varying intensity levels or exposure times?
Haha. I'm probably missing something, but either way- it would be super cool to see.
I'm curious about the projector teardown. Seems like a cheaper and less stressful way that having to assemble all the various hardware needed for a TI dev kit. Do you have any images of the projector that you can share? Was it easy to get access to the DLP mirror?
Unfortunately, I do not have any photos of the disassembly. However, getting access to the dmd chip was pretty easy because generally these chips can be replaced if they break. In the case of this projector, I had to modify the part holding the optics to get free access to the dmd chip.
@@HuygensOptics thank you. I’m also trying to get a dev kit from DLi but my application is fairly straightforward so this seemed to be a quick solution.
can this be used to make solid state lidar based on optical phased array?
Are you using an interferometer
Do you have various design files, or was this mainly "Hands On" ? If so, do you have a github?
So the project was really made mostly on the fly, no master plan, just trial and error. Sorry, I do not have Github.
@@HuygensOptics That makes sense, just wanted to see. Thanks for being great with all the comment questions. Keep up the good work!
how can you possibly see such a fine details (half a micrometer) using optical microscope and visible light? i thought such details, being comparable to the size of the light wavelength should be invisible
The camera works with the UV light.
Hey, love your video, just a quick question. With the Microscope objective how did you get it close enough to the DMD While equally allowing the light source to hit the DMD at the same time? Or what kind of "Condensing Optics" did you use? When ever I use my DMD it detracts orders all over the place, even with collimated light, so how did you get the image to focus so far away from the DMD itself?
Any info will be greatly appreciated I've been racking my brain with this problem!
The objective is actually infinity corrected, so you need a "tube lens" as a corrector between dmd and microscope objective. Unfortunately, that part was introduced later in the machine so it is missing in the video. Generally this is a lens with 150-200mm focal distance, with that distance to the dmd .
@@HuygensOptics Do you happen to have the make of thew tube lens you used? Was it from Thorlabs or something similar? sorry for questions but thank you for answering me first question your video is great!
@@danielblackmore4718 yes, the Thorlabs or Edmund tube lenses work fine. But in fact, more "regular" lenses also work well. Just as long as they have limited spherical aberration, generally meaning a fairly high f-number (say F/6 or higher), I guess you are safe.
@@HuygensOptics I'm really sorry but I'm still a bit confused, so did you manage to make a tube lens system with individual lens? if so what kind did you use? I've currently got a 100mm to a 50mm bi-convex in a kinda 4f system to demagnify the image down to the microscope. But the DMD Diffracts all the orders around too much so I never get a clear image, so did you have a similar set up of lens after the DMD before the Microscope? Do you remember what lenses you used?
Sorry again for all the questions thats my last one I promise!
@@danielblackmore4718best look on this page and look for infinite conjugate. It shows how the tube lens works in a microscope. In the dmd projections it works in reverse where the dmd is in the focal plane of the ocular. A tube lens eliminates the requirement for a fixed tube length, because of the infinity projection. www.edmundoptics.eu/knowledge-center/application-notes/microscopy/understanding-microscopes-and-objectives/
Hello! @Huygens Optics I am doing a similar project on maskless lithography like yours. But I really lack the knowledge of programming. I am wonder if you had uploaded your control program on GitHub or could you please teach me how to project the picture with the projector. (is it just using an hdmi wire or by program)
😃really appreciate it！！！
Best send me an email using the email address on the about page.
Awesome. What is the minimum feature width you think you will be able to achieve? Also I would be interested in your thoughts on thermal stability, particle contamination and vibration considerations.
Hi Nick, regarding the resolution: With 193nm it is now possible to produce dimensions below 50nm using multiple exposures and very advanced mask design. So if you would use a similar factor to 405nm, a 100nm should be achievable. But of course, my "mask" is very crude and it is not possible to deliberatly introduce interference effects to enhance resolution, so it would be quite hard to achieve. Bottom line is: I don't know at this point. I guess we just have to find out. Thermal stability and vibrations will be discussed in the next video on mechanics. Please be patient, it will take a few weeks before that video is ready.
@@HuygensOpticsyou don’t have masks. You only would need to change the pattern. So easy. Mount 4 polarised light sources around your DLP (lasers are polarised). Turn on top, bottom for features in one direction, change pattern, turn on left right lights. Or rotate around and animate DLP ?? Oh I see. All needs to be off axis.
Add vote taking concept of making 50 partly overlapping images with different data.
How is going this project?
I did this in my lab a few years ago. I will be interested in how well you can do field stitching. I actually made it so I could swap optics and align though the lens with red light before switching to nUV. I had much more field distortion than a real stepper though.
Nice!
You Wizard!
For me the most obscure part of this system is how exactly "some motors" can give a 100nm precision.
They can't, at least not without a very accurate encoder feedback loop. It's one of the issues that is not really addressed in this video series
You can always stick and use piezo for the fine range. Use that light source in an interferometer
Awesome!
Genius!
I wanna make meta lens too
I'm not sure that "goof around" is quite the right term for what you're doing :)
The real inventer is you
cooooooool
DIY EUV stepper anybody?
super
Just whow
Alien tech