🚨 *Addendum* 🚨 - I can't electronic my way out of a breadboard, so apologies for any basic EE mistakes! I just want to build the devices, not actually use them in circuits or whatever 😇 - In the PN/MSM animation, the electron/hole move in the wrong direction. Oops! Just pretend I know how to animate things correctly 😅 - Copper (I) == Cu2O and Copper (II) == CuO: This is very confusing! But it relates to the charge of the copper cation. Oxygen has a -2 charge, and in Cu2O there are two coppers per oxygen so they each carry a +1 charge hence (I) in the formula. CuO only has one copper per oxygen, so it carries a +2 charge and (II) in the formula. Very confusing to read out loud though 🫠
There was a time when end-users of radio revivers made their own schottky diodes diy on a daily basis. Each time you move the whisker on a crystal detector you make a new semiconductor diode.
Technically speaking - tech behind 4004 is roughly is in DIY range. There was a guy who managed to pull out a transistor grid, but is yet to go further. But he managed to fab some electronic guitar stuff.
@@samihawasli7408 LN2 is actually pretty cheap and easy to get. Some welding supply stores have it on hand, and commercial suppliers like Airgas sell to individuals. You're also forgetting that milling machine in the background. I wish I had one of those, but they are **not** cheap.
I just love the fact that a high tech corporation created a bleeding edge microscopy technology and said to themselves: "We need an influencer to promote this product." Whaddaya know there actually is such a guy, and I follow him on YT. The internet is beautiful sometimes.
@@jimzielinski946 very few people do. On the other hand, they probably plan to sell just a couple of them every year, so if even only two out of the 400'000 followers work for an institution in the market for such device, then it is a massive win for them.
@@JobyFluorine-ru4bd The ego must be fed. You're smarter than anyone here, and making condescending responses proves it, soon everyone will know about your superior intellect and you will finally be happy.
Transistors are now "free" to any reasonable degree of precision. Blows my mind. Even in 1960, they were about $1 each (~$8 today) which is STILL staggeringly cheap! And yet we found enough room to drop the cost by a factor of over a BILLION.
@@monad_tcp Incorrect. As everyone should know that is only half of the matter. The other half is the demand for the product. Supply and demand is the name of the game, not supply. Seriously, wth? How can you not know this?
It's insane that there are *multiple* people on youtube doing DIY semiconductor fabrication. Not only that, but you're getting sponsors from industry instrument manufacturers, instead of mobile games and VPN providers
I love that there are multiple people working on the tooling too - atkelar rebuilt a curve tracer being used here to demonstrate the elements - that's pretty deep arcana but in a different dimension.
lol I was just thinking the mobile game sponsor spot into would be funny, you know what I do while I'm waiting for my electron beam lithography setup to etch tiled panels? Idleking 9000! But no this sponsorship is amazing and their involvement with the project is great. It's the only sponsor spot I've ever seen where I learnt like 5 different new aspects of physics in 2 minutes.
@@multiarray2320 there is projectsinflight and sam zeloof, and jeri ellsworth, but sam and jeri stopped uploading. so there're only like 2 people on youtube making semiconductors right now.
This honestly insane to me how you were able to just fab a working “Chip” essentially. Im IC Layout Engineer working for a major Semiconductor Company and the amount of things you got correct in terms of the process (especially the double layer deposition for stepwise etching) while just studying a nature paper is honestly impressive. Im guessing your problems with the misalignment at the end have something to do with your process, you need to do some test alignments and see how “offset” from the grid your Cupric Oxide Layer is from your aluminum layer. Then you can just design your “masks” around that offset. We usually have alignment marks for these sort of limitations in Manufacturing. We mostly design Power Management IC’s in our Design Center, so I can’t say for sure if your MSM array will work, but Im hoping to see more like this soon. Great Job!
At an (amateur) astrophotography conference, the keynote talk was given by a "professional" astronomer (e.g., he got paid to do it, rather than spending his own money) who described a sensor and camera his research group built. It could do single photon detection AND also measure the wavelength of each photon's detection event. There is some solid state physics phenomenon whereby they would fabricate a feature on the chip they designed, such that when it interacted with a photon, it's effective capacitance changed by an amount proportional to the wavelength. So they would fabricate an array of LC pairs, organized as pixels on the wafer. The intention was that each pixel had a different, unique resonant frequency. Then they would pump the device with a broadband noise spectrum and then look at the resulting spectrum with an SDR receiver. They would observe a spectrum with a bunch of notches in it, corresponding to the resonant frequency of each pixel. When a photon interacted with one of the pixels, the notch in the spectrum would shift according to the wavelength. How fast you could run the FFT in the SDR software gave you the time-tagging precision of the detection event. He described the many iterations of fabricating his detector (with a 64x64 or maybe just 32x32 array). Some pixels would have the same resonant frequency as others, sort of aliasing them. Other pixels were, of course dead. Etc. I don't recall if the detector needed to be cryogenicly cooled in the camera body or not.. Many things were learned the hard way. I chatted with him after the fact and told him he had the most interesting job in the world, comprising: astrononmy, solid state physics, semiconduction fabrication, software defined radio and, of course, grant writing 🙂 At the end of all this, they did get to observe the pulsar in the Crab Nebula and time-tag those observations with simultaneous radio telescope observations to confirm the direction relationship which hadn't been done previously. What a long, long road to that result. Thanks for your video, it was really quite amazing to see all these fabrication steps actually done right before your eyes. And it gives me just a hint of the "art" beyond the "techniques" that goes on in modern semiconductor fabs.
I do hope he is able to find time for a trim for the next one. Only because people will judge the content from the creators looks. Also I hope he's able to take a nap ;-) delay the video is fine man, or make them in smaller chunks? We'll be here. We appreciate the effort you put in but don't blow yourself up doing this for us yaknow?
Seriously. This is IQ 160+ range. Just requires being able to learn such a broad range of subjects well enough to really do them, actually anticipate, RECOGNIZE, and verify the actual epistemic uncertainty ... and device solutions to overcome those. And then explain it all so my far lower IQ can grasp it all.
I know your lab is significantly better than a lot of home labs out there, but it still gives me hope that one day we'll have DIY electronics from scratch. I love the thought of just plonking down a box in the corner and sending it a file and a few hours later a new chip.
I mean, it is simply indicating the charge of the cation (copper in this case), when you are dealing with stuff like Sulfates or Phosphates (-3/-4 charge anions) it becomes a lot more clear why this convention is important, as the number of atoms for the cation isn't going to be the opposite of the I/II
Some elements can form more than one ion others can only form one ion generally, because copper can form multiple ions the number of missing electrons must be indicated.
Its SOO cool to see topics I studied in university applyed in practice - learning just the theory and doing it all on paper made it seem so bland and boring. But every time I see someone actually DO the theory, I am just stunned and miss this part of my life ... thanks for bringing it back to me
Absolutely incredible, even before demonstrating the sensor arrays, this is one of the most impressive single-person projects I've seen - that's a significant fraction of the technology required for a usable semiconductor process (and not even a huge node, 800nm is totally functional), just developed by one person! I can't wait to see where this goes :D (Yes, I know about Sam Zeelof's ICs, but unlike his approach, this video includes making the semiconductor as well - basically a SOI process, but not silicon)
@@DaveEtchells The fact that anything measured in microns even exists is crazy enough. That it is considered large is just insane if you really think about it.
Exactly. To include people making EDM machines at home ... this is still SO much bigger. A compendium of knowledge required ... and yet, it's still but a subset of this genius's generalist knowledge. Un. Real.
Seeing how much work it takes to make such a tiny sensor really makes me appreciate just how much work goes in to modern computers. It's remarkable that we can make the chips we do.
Can't believe how much technology has evolved. Like there are millions of similar stuff in your phone cameras right now and don't even get me started on processors
DIY semiconductors are my favorite type of videos. Want to say that your attempt at a full device was a great satisfactory ending for this work, and I'm really looking forward to you finishing it
I work at a major semiconductor fab on some cutting-edge lithography tooling. Intel High-NA team if you want to see the machines themselves. I love seeing semiconductor tech start to show up in the DIY space. I'm hoping we see somebody try to home-brew a 4004 in the near future.
There was a thread on twitter where people in the industry were discussing how expensive oscilloscopes when turned on create unexpected voltage on the probes with some destroying the test object and how there are recall campaigns due to this issue.. Fascinating work as usual, mate!
In case you all wanted to know, here is a list of metal-to-oxide adhesion layers: Ti, Cr, Al, Ta, Mo, Nb, V, & Hf. I have used Ti, Cr, Ta, and Mo in my chips so far, for adhesion.
A+ work here! In high school I made circuit boards by using toner and etching with HCL, this is truly several steps up from that. One day I dream of having my own lab to be a mad scientist like you
I thought I was gonna be completely lost during these explanations but the way you explain the different processes and especially along with your animations made it so much easier to follow along. I know it must have been a lot of work to put the animations together but I really appreciate it because it helped so much.
Wow, you've come so far from your early videos of rebuilding a mill to this! I can't wait to see your 0.000016 mega pixel camera. On a completely different note, am I the only one who sees a Trojan Horse in the short circuit at 27:13 ? I guess it's a bit of a Rorschach test.
Great video. And this is only the first stage. After that there is all the electronics to actually transfer the pixels, manage exposure time, HDR, serial communication with external IC, etc. Makes you realize the complexity of these devices that we take for granted.
I'm in the Semiconductor industry and I'm amazed by the level of details in your video. Awesome to see that deep tech is becoming available on YT. Nice job!
DIY microlitography is certainly not a thing I’d expect to see but how I’m I glad I did. Amazing work, love all the referencing and explanations, beautiful pictures, good 3D animations. Great work man
Man this is the first sponsor spot I have ever been *invested* in. I learnt a bunch of stuff and that tech is AMAZING. Kudos to them for being so open with how it all works and for sponsoring you to do this cool project. Only suggestion might be a little blurb on why this copper semiconductor photo sensor is special other than being fabricatable? Unless that is the prime attraction? I just saw "high performance" in the title of the paper was all.
This nano stuff is fascinating. Even though these scales are so tiny that they escape my comprehension, it's so fascinating to look at. 27:55 now I wanna see a collab with the SloMoGuys to see what the hell is happing there at 100k FPS.
In high school I took a printing a publishing course, and part of it was on screen printing. We used photo lithology to make the stencils. You would lay out the pattern in light sensitive paper, as a negative, and then shine laser on it for 20 minutes. After placing the cued light sensitive paper on the cloth screen, you could wash away with a forced water the non-protected areas and therefore create the stencil to screen print through. I'm sure it's not the modern method for screen printing stencils, but it was cool to do by hand, and so many of those principles of lithology carry over.
Man this is the video that I've wanted to see you make for a while now. Very cool stuff, can't wait to see the stuff you do next with diy image sensors!
This is the first video of yours I have watched/come across and I just want to commend the detail and patience in describing not just the processes and intent, but the tools in use, and dude... way to save 30 grand! You burned over that so quickly but seriously that alone was so impressive!
1:10 This may be a stupid question, but how can the semiconductor be naturally p-type? What is acting as an acceptor in undoped CuO? 3:40 If the copper - after solidstate dewetting - follows the underlying Silicon crystal structure, then what happened to your Oxide layer? 3:55 looks particularly interesting. What as the big blobs and why are they not merged with the little blobs? A thicker SiO2 layer may have helped as well. And last but not least, label your KLayout layers god damnit! So we can be even more hyped for the next project! KLayout also has a macro interface. If you want to process variations of the same layout (e.g. with different spacing) you (or I) can easily write you some code to do that. Hit me up it that would help.
- Grain of salt in that I'm a bit hazy here too, but my understanding is that both copper oxides are p-type due to intrinsic defects, namely copper vacancies in the lattice (just happens naturally as it grows, since the energy required to form a vacancy is pretty small for this material). These missing copper cations make the neighboring oxygens eager to gain a few electrons, so it acts as a positive charge charrier. But my understanding is that they aren't very mobile since they are tied to the oxygen orbitals, and have lower performance than impurity doped semiconductors. There's also some contribution due to adsorbed oxygen on the surface, a few papers show effects based on oxygen concentration/vacuum/etc - Good question, I'm not sure! Possible I didn't grow a thick enough oxide, or it wasn't very high quality (pinholes, etc)? It was closer to a dry oxidation than wet, so I don't expect the oxide was very thick. - I didn't think to EDS the sample at the time, so unfortunately I'm not sure. But just based on the BSD signal (not shown in the video) I think it's all copper. So likely not a contaminant. There are protocols out there which show you can grow copper oxide nano-wires by long duration furnace treatment. Perhaps this is the start of a similar process? Really not sure! - KLayout keeps deleting my layer labels 😭 I'm probably doing something wrong though, I'm very new to klayout. Will take a look at the macro capabilities, and thanks for the offer! Might take you up on that! :)
This video was uploaded 5 hours ago. To think about the precious time, I have wasted doing something else besides watching your video... Thank you for uploading something just in time to make this day an enjoyable one, after spending countless hours on a task that could have easily been automated. You literally saved my day!
Possible food for thought : when I get discouraged I try to remember that it is the act of attempting projects and trying to figure things out that I actually enjoy more than the completed project itself . Any way awesome video brother man can't wait to see what you come up with next.
It would be great to have a light wavelength sensor that could be used to monitor a variable LED output source. Accurate LED wavelength calibration could help identify variations in color blindness. It also could be used to measure the color phase shift with those suffering from macula degeneration. (green LED light phase shifted to blue inside the eye thru the macula)
This really makes you appreciate the tech, we take for granted. It's insane how much the semiconductor industry has improved, that we can get a precisely manufactured chip for only a few cents. If someone from like 500 years ago saw this, they wouldn't believe it. The fact, that we literally have miniature supercomputers in our pockets, that are much much more powerful, than the first computer, that took up an entire room, just wow
You, Sir, just earned a subscriber! Doing this as "a random youtuber" is NUTS! I thought "Breaking Taps" would be a machinist or automotive channel, but looking through your back catalogue tells a very different story. Can't wait for more!
This is awesome in every way. To do this and film it at the same time is over the top. To DIY making something that small that precisely is remarkable. Filming it complicates it 100x. Thank You! For putting so much of your time into this, I appreciate the education, and its delivery made it enjoyable and easy.
Hats off to the excellent quality production of a high tech journey. The obscure drama that unfolds during technological invention typically escapes the mainstream audience. Thanks for putting the effort in to share your journey.
There were two papers I found a while back that cover creating transistors w/ cupric oxide. The bandgap is narrower w/ CuO which enabled them to make transistors that are lower power and able to switch faster than silicon. If we could even get CuO integrated circuits at the semiconductor fab scales we had a decade or two ago, we'd likely still be ahead of the Moore's Law curve because we could overclock those suckers way harder! We could make CPUs larger to have the number of cores we're accustomed to today, and run them on less power.
Curious about the stoichiometry or equilibrium of the deposit -- does seem like a mix of oxides is possible, given the spectra. Looking forward to more detail! Regarding layer thickness, what was special about this, was it just kind of taken from the paper? The polycrystalline deposits look quite rough; I wonder what contribution if any the grain boundaries have? Seems like there could be some quite narrow boundaries; but perhaps it averages out over the area of a pixel (let alone a macroscopic device like a CdS photocell, to name an equivalent application/example).
Agreed! Hoping we can refine some of the analytical results for future videos. In theory, at elevated temperatures Cu2O -> CuO conversion is reaction-limited so should be pretty complete. But at lower temperatures, Cu -> Cu2O is diffusion limited and will likely end up with a mix of Cu, Cu2O and CuO. I've seen some papers that control time, humidity, oxygen and temperature more carefully to fine-tune the results. Might try that in the future. Re: layer thickness: just following the paper! Their general thesis is that higher temperatures leads to large grains with smaller neck regions between the grains, and this improves the photoresponse. So I think they were wanting a thin monolayer to help prove that out. (Their hypothesis is that the outer shell is more conductive due to easy oxygen exchange, while the inner core is more resistive. So increased resistance and decreased dark current scales relative to grain and neck sizes). In theory a thicker layer would yield better photon efficiency since the absorption depth is a few hundred nm if I remember correctly (~500nm I think?). But yeah, I didn't want to deviate too far from the paper to start since I just wanted to get something working 🙂
Seriously! I had to constantly look them up while writing the script for this video. Madness.
7 หลายเดือนก่อน
Copper (I) has donated 1 electron per atom to oxygen, and oxygen wants 2. Cu(I) is Cu2O, because each Oxygen atom reacts with two copper atoms. Cu(II) donates two electrons, so 1 Cu atom and one O atom reacts. Also it's Cu2O, not Cu20. The capital O means Oxygen. Same thing with water, H2O, with H2 being two hydrogen atoms.
Im a Student in a German University of Applied Sciences and studing Elektrical Engineering and i had a course last semester about semiconductors. This is so amazing. Thanks allot
I assume you got into a situaiton with SpaceX or NASA therefore removed the last video (the one about the tiles), anyway great to see you are back with another great video.
Might've even been the DoD. Remember that the other big usage case for heat shields is dropping nukes and other large explosive devices on people from space, without them melting in the process.
I could watch that copper film peel for ages. So relaxing, but also kind of stimulating. Wouldn’t mind a separate video of just the copper film peeling. Keep it up!😃
It does look really neat under the microscope! Sometimes the film is stable until it is scratched. The point of damage becomes a nucleation site and the whole thing starts peeling from there. Pretty fun to watch. Less soothing when you spent the last two days making it and then you watch it peel off though 🫠
15:29 interesting, I've seen graphs like this, especially where it comes to diodes, but I never knew this are actual graphs produced on an oscilloscope. I thought there we constructed only theoretically or by just measuring specific points on that graph.
You can build a curve tracer yourself with a signal gen, oscilloscope, resistor and class AB DC audio amplifier. Using the volume knob for adjusting the voltage. You may need to use two channels and set the scope to differential, maybe float the output, and create a star ground for probing to eliminate noise.
The greenish color you see on corroded copper is actually copper chloride! The oxides range from yellow to black depending on how much CuO vs Cu2O there is (and how thick)
I love your diy probe station mover. Reminds me of grad school where we did a similar hack to move about entire microscope using some ball bearing drawer slides and an x-y manipulator
Absolutely amazing from one end to the other. The AFM-FTIR thing is super cool! I made a simple AFM in grad school, back in the early days, so interesting to see how many different directions that idea has gone.
I've been trying to explain how silicon manufacturing works to my dad and this and your MEMS video will help him understand some of it, thanks. If you ever find the time or need to make a short'ish (10-20 minutes) video about it, please do so. You're explanations and graphics make it much easier to understand.
Very interesting!! You could update your furnace by putting a modern PID temperature controller on it. They support multiple different kinds of thermocouples, support hysteresis settings. Not sure about the internals of what controls the filament in your furnace but people wire in PID controllers to their espresso machines with great results and even some data logging functions. Keep these awesome videos coming! I learn so much from them.
FYI, you can somewhat better points on the probe needles if you translate the points up and down in the NaOH; that results in a rounded, convex tip vs. the typical triangular shape. We used to wind up curling the points when probing because there's a bending moment from the geometry of probing. The rounded tips would last 10x longer, because there's a larger cross-section. That's something I learned during an internship at a semiconductor lab during college, many, many, eons ago.
I am amazed at the quality and effort behind the video, pure eye candy animations and the sponsor is actually participating in the project with their amazing technology. Since it's so sensitive, does it run in a vacuum or just clean air? The box doesn't look like a vacuum chamber.
I believe it's just open, clean atmosphere (albeit in an enclosure to help limit drafts and vibrations). My understanding is that part of the extreme sensitivity is due to the "depth" which it detects molecules. Only the very surface molecules interact with the tip, so you read the top ~20nm of the material. If there's a thin layer of contaminant you'll mostly pick that up (which is great if you're looking for contaminants! Less good if the contaminant is accidental from packaging). They also have a "bulk" mode that looks a few hundred nanometers below the surface. I'm not entirely sure how it works to be honest, but I think they can combine that result with the "surface" results to get a picture of how materials are positioned.
If TH-cam had existed when I was in high school and if I'd watched this, I imagine I'd have decided to study whatever you did instead of my current profession.
Great content as always, but the real gem was introducing me to PhysicsInFlight. Finally understanding the underlying principles makes watching videos like yours so much more enjoyable.
Happy to share a cool new channel! It was a very helpful video to me as well, helped solidify some questions I had about how all the energy levels and bands work. Really excited to see what he does in the future, each video is building out a nice little semiconductor fab!
Nice work! Love the alignment mark pic 🤣. I feel your pain brother. This work is teaching many valuable lessons. The most important: Never bet against silicon.
These videos really do a great job of inspiring me. These really help quantize intimidating topics like "DIY semiconductors". Amazing content, keep it up!
Try to create p-Si/SiO2(20-50 nm)/Al structure. p-Si - GND, Al grid contact - positive bias about 1 - 10 V. (negative bias on Al in case of n type silicon substrate). The depletion region is formed under the bias. Minority charge carriers are injected into the SiO2 upon illumination. The current can increase 100 times if intense illumination is used. You can use ITO or ZnO instead Al to increase this effect. Also you can use ZrO2, TiO, Ta2O5 dielectrics instead SiO2, and also overstoicheometric SiOx, ZrOx, etc to increase current density
🚨 *Addendum* 🚨
- I can't electronic my way out of a breadboard, so apologies for any basic EE mistakes! I just want to build the devices, not actually use them in circuits or whatever 😇
- In the PN/MSM animation, the electron/hole move in the wrong direction. Oops! Just pretend I know how to animate things correctly 😅
- Copper (I) == Cu2O and Copper (II) == CuO: This is very confusing! But it relates to the charge of the copper cation. Oxygen has a -2 charge, and in Cu2O there are two coppers per oxygen so they each carry a +1 charge hence (I) in the formula. CuO only has one copper per oxygen, so it carries a +2 charge and (II) in the formula. Very confusing to read out loud though 🫠
you better make the readout circuits from scratch as well
NP. (I couldn't pass that up :)
Your animations are great man. I never really *got* how the junction stopped the conduction or how to pronounce schotty lol but this really helped.
It’s ok, we still like you 😊
Thank you for erratum
"DIY" and "semiconductor" are not words I normally see in the same sentence. tbh it is pretty awesome.
There was a time when end-users of radio revivers made their own schottky diodes diy on a daily basis. Each time you move the whisker on a crystal detector you make a new semiconductor diode.
Wel sure. Whose DYI lab doesn’t have a SEM, sputter tool, photolithography bay… etc.. speaking of, my LN2 delivery is here. gtg
Wait till you hear about Sam Zeloof
Technically speaking - tech behind 4004 is roughly is in DIY range. There was a guy who managed to pull out a transistor grid, but is yet to go further. But he managed to fab some electronic guitar stuff.
@@samihawasli7408 LN2 is actually pretty cheap and easy to get. Some welding supply stores have it on hand, and commercial suppliers like Airgas sell to individuals.
You're also forgetting that milling machine in the background. I wish I had one of those, but they are **not** cheap.
I just love the fact that a high tech corporation created a bleeding edge microscopy technology and said to themselves: "We need an influencer to promote this product." Whaddaya know there actually is such a guy, and I follow him on YT. The internet is beautiful sometimes.
It is the right crowd of people to advertise to
@@linecraftman3907something tells me I couldn't afford such a toy for a hobby.
@@jimzielinski946 well I believe it's aimed at people working in a lab with big bucks
@@jimzielinski946 very few people do. On the other hand, they probably plan to sell just a couple of them every year, so if even only two out of the 400'000 followers work for an institution in the market for such device, then it is a massive win for them.
@@jimzielinski946 who's saying you're their target audience
"I hope it was interesting...", really? This is just mindblowing... yet again
@@JobyFluorine-ru4bd You must be fun at parties.
@@JobyFluorine-ru4bd The ego must be fed. You're smarter than anyone here, and making condescending responses proves it, soon everyone will know about your superior intellect and you will finally be happy.
@@PJ-oe6eu LOL. Savage. But appropriate.
@JobyFluorine-ru4bd I don't think this is sixth grade level.
The average selling price of a CMOS Image Sensor is currently just above $3/sensor. The semiconductor industry is insane.
its amazing that the price of anything really does only depend on the amount of it we're able to produce
Transistors are now "free" to any reasonable degree of precision. Blows my mind. Even in 1960, they were about $1 each (~$8 today) which is STILL staggeringly cheap! And yet we found enough room to drop the cost by a factor of over a BILLION.
@@monad_tcp Incorrect. As everyone should know that is only half of the matter. The other half is the demand for the product. Supply and demand is the name of the game, not supply.
Seriously, wth? How can you not know this?
@@whyjnot420it's quite obvious that he is assuming a given level of demand.
@@DisorderedArray They stated "only".
They are objectively wrong because of that word.
It's insane that there are *multiple* people on youtube doing DIY semiconductor fabrication. Not only that, but you're getting sponsors from industry instrument manufacturers, instead of mobile games and VPN providers
I mean, given the type of people watching these videos, it's probably effective advertising.
I love that there are multiple people working on the tooling too - atkelar rebuilt a curve tracer being used here to demonstrate the elements - that's pretty deep arcana but in a different dimension.
what other youtubers do this type of content? asking for a friend :)
lol I was just thinking the mobile game sponsor spot into would be funny, you know what I do while I'm waiting for my electron beam lithography setup to etch tiled panels? Idleking 9000! But no this sponsorship is amazing and their involvement with the project is great. It's the only sponsor spot I've ever seen where I learnt like 5 different new aspects of physics in 2 minutes.
@@multiarray2320 there is projectsinflight and sam zeloof, and jeri ellsworth, but sam and jeri stopped uploading. so there're only like 2 people on youtube making semiconductors right now.
This is absolutely insane to be able to do this DIY outside of an industrial setting. I've never seen anything like this attempted, amazing!
🥰
This honestly insane to me how you were able to just fab a working “Chip” essentially.
Im IC Layout Engineer working for a major Semiconductor Company and the amount of things you got correct in terms of the process (especially the double layer deposition for stepwise etching) while just studying a nature paper is honestly impressive.
Im guessing your problems with the misalignment at the end have something to do with your process, you need to do some test alignments and see how “offset” from the grid your Cupric Oxide Layer is from your aluminum layer. Then you can just design your “masks” around that offset. We usually have alignment marks for these sort of limitations in Manufacturing.
We mostly design Power Management IC’s in our Design Center, so I can’t say for sure if your MSM array will work, but Im hoping to see more like this soon. Great Job!
Your comment should be stickied.
Once every few years someone drops a new diy semiconductor vid and its the my favorite thing
100%
At an (amateur) astrophotography conference, the keynote talk was given by a "professional" astronomer (e.g., he got paid to do it, rather than spending his own money) who described a sensor and camera his research group built. It could do single photon detection AND also measure the wavelength of each photon's detection event. There is some solid state physics phenomenon whereby they would fabricate a feature on the chip they designed, such that when it interacted with a photon, it's effective capacitance changed by an amount proportional to the wavelength.
So they would fabricate an array of LC pairs, organized as pixels on the wafer. The intention was that each pixel had a different, unique resonant frequency. Then they would pump the device with a broadband noise spectrum and then look at the resulting spectrum with an SDR receiver. They would observe a spectrum with a bunch of notches in it, corresponding to the resonant frequency of each pixel. When a photon interacted with one of the pixels, the notch in the spectrum would shift according to the wavelength. How fast you could run the FFT in the SDR software gave you the time-tagging precision of the detection event.
He described the many iterations of fabricating his detector (with a 64x64 or maybe just 32x32 array). Some pixels would have the same resonant frequency as others, sort of aliasing them. Other pixels were, of course dead. Etc. I don't recall if the detector needed to be cryogenicly cooled in the camera body or not.. Many things were learned the hard way. I chatted with him after the fact and told him he had the most interesting job in the world, comprising: astrononmy, solid state physics, semiconduction fabrication, software defined radio and, of course, grant writing 🙂
At the end of all this, they did get to observe the pulsar in the Crab Nebula and time-tag those observations with simultaneous radio telescope observations to confirm the direction relationship which hadn't been done previously. What a long, long road to that result.
Thanks for your video, it was really quite amazing to see all these fabrication steps actually done right before your eyes. And it gives me just a hint of the "art" beyond the "techniques" that goes on in modern semiconductor fabs.
If there is a video on youtube i would like to watch it.
@@wal-3732 Unfortunately, the presentation wasn't recorded.
Was that a Microwave Kinetic Inductance Detector by chance?
Your hair cut speaks volumes about the complexity and frustration of this project :D
I do hope he is able to find time for a trim for the next one. Only because people will judge the content from the creators looks. Also I hope he's able to take a nap ;-) delay the video is fine man, or make them in smaller chunks? We'll be here. We appreciate the effort you put in but don't blow yourself up doing this for us yaknow?
thats the first thing i noticed when starting this video
immediately told me this is going to be a crazy one
I shan't cut my hair until a working camera is complete! 😆
@@BreakingTaps it's a good look imo!
@@BreakingTaps We shall call you Gandalf from here forth !
I can't believe you're able to do DIY semiconductor fab, AND do it while making a great video. Incredible stuff as always dude!
Seriously. This is IQ 160+ range. Just requires being able to learn such a broad range of subjects well enough to really do them, actually anticipate, RECOGNIZE, and verify the actual epistemic uncertainty ... and device solutions to overcome those. And then explain it all so my far lower IQ can grasp it all.
I know your lab is significantly better than a lot of home labs out there, but it still gives me hope that one day we'll have DIY electronics from scratch. I love the thought of just plonking down a box in the corner and sending it a file and a few hours later a new chip.
I’ve thought about this too, surely something like 400 nm wouldn’t be too difficult to automate, then just package like a bamboo printer.
Your ability to go past the part where Copper II is Cu and Copper I is Cu2 without snark is one of the many ways you're better than me.
I am super annoyed at that naming scheme smh IS IT THAT HARD???
@@jacobblotkamp2945 it's about some property with electrons
I mean, it is simply indicating the charge of the cation (copper in this case), when you are dealing with stuff like Sulfates or Phosphates (-3/-4 charge anions) it becomes a lot more clear why this convention is important, as the number of atoms for the cation isn't going to be the opposite of the I/II
Some elements can form more than one ion others can only form one ion generally, because copper can form multiple ions the number of missing electrons must be indicated.
Its SOO cool to see topics I studied in university applyed in practice - learning just the theory and doing it all on paper made it seem so bland and boring. But every time I see someone actually DO the theory, I am just stunned and miss this part of my life ... thanks for bringing it back to me
Absolutely incredible, even before demonstrating the sensor arrays, this is one of the most impressive single-person projects I've seen - that's a significant fraction of the technology required for a usable semiconductor process (and not even a huge node, 800nm is totally functional), just developed by one person! I can't wait to see where this goes :D
(Yes, I know about Sam Zeelof's ICs, but unlike his approach, this video includes making the semiconductor as well - basically a SOI process, but not silicon)
Back when I was in the semiconductor research biz, DARPA was pushing to get to a 1.25 _micron_ fab node (dating myself 😂)
@@DaveEtchells The fact that anything measured in microns even exists is crazy enough. That it is considered large is just insane if you really think about it.
Exactly. To include people making EDM machines at home ... this is still SO much bigger. A compendium of knowledge required ... and yet, it's still but a subset of this genius's generalist knowledge. Un. Real.
Seeing how much work it takes to make such a tiny sensor really makes me appreciate just how much work goes in to modern computers. It's remarkable that we can make the chips we do.
Can't believe how much technology has evolved. Like there are millions of similar stuff in your phone cameras right now and don't even get me started on processors
DIY semiconductors are my favorite type of videos. Want to say that your attempt at a full device was a great satisfactory ending for this work, and I'm really looking forward to you finishing it
I work at a major semiconductor fab on some cutting-edge lithography tooling. Intel High-NA team if you want to see the machines themselves. I love seeing semiconductor tech start to show up in the DIY space. I'm hoping we see somebody try to home-brew a 4004 in the near future.
There was a thread on twitter where people in the industry were discussing how expensive oscilloscopes when turned on create unexpected voltage on the probes with some destroying the test object and how there are recall campaigns due to this issue..
Fascinating work as usual, mate!
In case you all wanted to know, here is a list of metal-to-oxide adhesion layers: Ti, Cr, Al, Ta, Mo, Nb, V, & Hf. I have used Ti, Cr, Ta, and Mo in my chips so far, for adhesion.
What about Sn, In and Zn?
Felt like a three minute video. It has genuinely been a while since I've been so immersed and attentive during a video, thank you.
he's got that medieval blacksmith look locked down, give him a leather apron and a hammer
Maybe some of the Taiwanese descendents of them are working at TSMC right now?
roses aRE RED violet blue.. there is and asian beter than you.. liek hat
A+ work here! In high school I made circuit boards by using toner and etching with HCL, this is truly several steps up from that. One day I dream of having my own lab to be a mad scientist like you
I thought I was gonna be completely lost during these explanations but the way you explain the different processes and especially along with your animations made it so much easier to follow along. I know it must have been a lot of work to put the animations together but I really appreciate it because it helped so much.
Wow, you've come so far from your early videos of rebuilding a mill to this! I can't wait to see your 0.000016 mega pixel camera. On a completely different note, am I the only one who sees a Trojan Horse in the short circuit at 27:13 ? I guess it's a bit of a Rorschach test.
Great video. And this is only the first stage. After that there is all the electronics to actually transfer the pixels, manage exposure time, HDR, serial communication with external IC, etc. Makes you realize the complexity of these devices that we take for granted.
Most interesting ad I've seen on youtube
I'm in the Semiconductor industry and I'm amazed by the level of details in your video. Awesome to see that deep tech is becoming available on YT. Nice job!
Nice DIY photosensitive fuses. Now I can finally know if the lights were on when the fuse blew! 💡
DIY photosensitive ^ _thermite_ fuses. (maybe)
@@trumanhwa device that shits itself and melts everything when exposed to light is finally possible 🤩
DIY microlitography is certainly not a thing I’d expect to see but how I’m I glad I did. Amazing work, love all the referencing and explanations, beautiful pictures, good 3D animations. Great work man
This video is awesome. Form the lighting to the rendered diagrams. And the subject is super cool
Man this is the first sponsor spot I have ever been *invested* in. I learnt a bunch of stuff and that tech is AMAZING. Kudos to them for being so open with how it all works and for sponsoring you to do this cool project. Only suggestion might be a little blurb on why this copper semiconductor photo sensor is special other than being fabricatable? Unless that is the prime attraction? I just saw "high performance" in the title of the paper was all.
This nano stuff is fascinating. Even though these scales are so tiny that they escape my comprehension, it's so fascinating to look at.
27:55 now I wanna see a collab with the SloMoGuys to see what the hell is happing there at 100k FPS.
In high school I took a printing a publishing course, and part of it was on screen printing. We used photo lithology to make the stencils. You would lay out the pattern in light sensitive paper, as a negative, and then shine laser on it for 20 minutes. After placing the cued light sensitive paper on the cloth screen, you could wash away with a forced water the non-protected areas and therefore create the stencil to screen print through.
I'm sure it's not the modern method for screen printing stencils, but it was cool to do by hand, and so many of those principles of lithology carry over.
Really cool. Been really fascinated with how far copper semis can be taken since I saw keystones videos on a copper oxide solar panel.
this was so mind-boggling, it took me 3 sessions to go through the video. Amazing stuff there, sir
Very cool , i can't belive what you achived in basically a home lab . Can't wait for the next video
Man this is the video that I've wanted to see you make for a while now. Very cool stuff, can't wait to see the stuff you do next with diy image sensors!
10:16 that fluid simulation must have taken a long time
It shouldn’t be too bad anymore in blender with a decent computer.
It was really coarsely simulated so it wouldn't take nearly as long as you would think.
This is the first video of yours I have watched/come across and I just want to commend the detail and patience in describing not just the processes and intent, but the tools in use, and dude... way to save 30 grand! You burned over that so quickly but seriously that alone was so impressive!
1:10 This may be a stupid question, but how can the semiconductor be naturally p-type? What is acting as an acceptor in undoped CuO?
3:40 If the copper - after solidstate dewetting - follows the underlying Silicon crystal structure, then what happened to your Oxide layer? 3:55 looks particularly interesting. What as the big blobs and why are they not merged with the little blobs?
A thicker SiO2 layer may have helped as well.
And last but not least, label your KLayout layers god damnit! So we can be even more hyped for the next project!
KLayout also has a macro interface. If you want to process variations of the same layout (e.g. with different spacing) you (or I) can easily write you some code to do that. Hit me up it that would help.
- Grain of salt in that I'm a bit hazy here too, but my understanding is that both copper oxides are p-type due to intrinsic defects, namely copper vacancies in the lattice (just happens naturally as it grows, since the energy required to form a vacancy is pretty small for this material). These missing copper cations make the neighboring oxygens eager to gain a few electrons, so it acts as a positive charge charrier. But my understanding is that they aren't very mobile since they are tied to the oxygen orbitals, and have lower performance than impurity doped semiconductors. There's also some contribution due to adsorbed oxygen on the surface, a few papers show effects based on oxygen concentration/vacuum/etc
- Good question, I'm not sure! Possible I didn't grow a thick enough oxide, or it wasn't very high quality (pinholes, etc)? It was closer to a dry oxidation than wet, so I don't expect the oxide was very thick.
- I didn't think to EDS the sample at the time, so unfortunately I'm not sure. But just based on the BSD signal (not shown in the video) I think it's all copper. So likely not a contaminant. There are protocols out there which show you can grow copper oxide nano-wires by long duration furnace treatment. Perhaps this is the start of a similar process? Really not sure!
- KLayout keeps deleting my layer labels 😭 I'm probably doing something wrong though, I'm very new to klayout. Will take a look at the macro capabilities, and thanks for the offer! Might take you up on that! :)
This video was uploaded 5 hours ago. To think about the precious time, I have wasted doing something else besides watching your video... Thank you for uploading something just in time to make this day an enjoyable one, after spending countless hours on a task that could have easily been automated. You literally saved my day!
How about those graphics!
Haha thanks! Been quite a journey learning Blender. Still long way to go, but starting to feel more comfortable with it :)
Possible food for thought : when I get discouraged I try to remember that it is the act of attempting projects and trying to figure things out that I actually enjoy more than the completed project itself . Any way awesome video brother man can't wait to see what you come up with next.
The age of home lithography is here and I am all here for it
Every time I see one of your videos I'm more and more impressed with your knowledge and skill with this kind of thing!
It would be great to have a light wavelength sensor that could be used to monitor a variable LED output source. Accurate LED wavelength calibration could help identify variations in color blindness.
It also could be used to measure the color phase shift with those suffering from macula degeneration.
(green LED light phase shifted to blue inside the eye thru the macula)
This really makes you appreciate the tech, we take for granted. It's insane how much the semiconductor industry has improved, that we can get a precisely manufactured chip for only a few cents. If someone from like 500 years ago saw this, they wouldn't believe it. The fact, that we literally have miniature supercomputers in our pockets, that are much much more powerful, than the first computer, that took up an entire room, just wow
God, I love this channel. This is obscenely good.
You, Sir, just earned a subscriber! Doing this as "a random youtuber" is NUTS! I thought "Breaking Taps" would be a machinist or automotive channel, but looking through your back catalogue tells a very different story. Can't wait for more!
man, what am I doing here
Fr
Sleeping
I have a whole new perspective of digital sensors watching you build one pixel. My phone has 64 MP Sony sensor. Huge thanks my friend.. keep at it.
The video idea: 😀
Materials required:😵😵😵
This is awesome in every way. To do this and film it at the same time is over the top. To DIY making something that small that precisely is remarkable. Filming it complicates it 100x. Thank You! For putting so much of your time into this, I appreciate the education, and its delivery made it enjoyable and easy.
Day 1 of learning how to make a computer from scratch if civilization falls
Hats off to the excellent quality production of a high tech journey. The obscure drama that unfolds during technological invention typically escapes the mainstream audience. Thanks for putting the effort in to share your journey.
This is astounding. Thank you so much for sharing! ❤
This is at the same level as that very old video of this person making vacuum tubes in his garage. Amazing. Thanks for posting.
Awesome stuff, thank you!
Unbelievable 😳 The amount of equipment, skill and deep knowledge required to do this "DIY project" completely blew my mind.
There were two papers I found a while back that cover creating transistors w/ cupric oxide. The bandgap is narrower w/ CuO which enabled them to make transistors that are lower power and able to switch faster than silicon. If we could even get CuO integrated circuits at the semiconductor fab scales we had a decade or two ago, we'd likely still be ahead of the Moore's Law curve because we could overclock those suckers way harder! We could make CPUs larger to have the number of cores we're accustomed to today, and run them on less power.
ur on to something .
I feel like I actually learn stuff watching and listening to you. I love the way you present the information. Quality content.
Curious about the stoichiometry or equilibrium of the deposit -- does seem like a mix of oxides is possible, given the spectra. Looking forward to more detail!
Regarding layer thickness, what was special about this, was it just kind of taken from the paper? The polycrystalline deposits look quite rough; I wonder what contribution if any the grain boundaries have? Seems like there could be some quite narrow boundaries; but perhaps it averages out over the area of a pixel (let alone a macroscopic device like a CdS photocell, to name an equivalent application/example).
Agreed! Hoping we can refine some of the analytical results for future videos. In theory, at elevated temperatures Cu2O -> CuO conversion is reaction-limited so should be pretty complete. But at lower temperatures, Cu -> Cu2O is diffusion limited and will likely end up with a mix of Cu, Cu2O and CuO. I've seen some papers that control time, humidity, oxygen and temperature more carefully to fine-tune the results. Might try that in the future.
Re: layer thickness: just following the paper! Their general thesis is that higher temperatures leads to large grains with smaller neck regions between the grains, and this improves the photoresponse. So I think they were wanting a thin monolayer to help prove that out. (Their hypothesis is that the outer shell is more conductive due to easy oxygen exchange, while the inner core is more resistive. So increased resistance and decreased dark current scales relative to grain and neck sizes).
In theory a thicker layer would yield better photon efficiency since the absorption depth is a few hundred nm if I remember correctly (~500nm I think?). But yeah, I didn't want to deviate too far from the paper to start since I just wanted to get something working 🙂
Having a background in material science, this channel hits close to home. Instant sub! Really cool to see this fabrication process on a small scale
That naming convention is crazy, copper (II) oxide is CuO but copper (I) oxide is Cu20 like alright buddy.
Seriously! I had to constantly look them up while writing the script for this video. Madness.
Copper (I) has donated 1 electron per atom to oxygen, and oxygen wants 2.
Cu(I) is Cu2O, because each Oxygen atom reacts with two copper atoms. Cu(II) donates two electrons, so 1 Cu atom and one O atom reacts.
Also it's Cu2O, not Cu20. The capital O means Oxygen. Same thing with water, H2O, with H2 being two hydrogen atoms.
Im a Student in a German University of Applied Sciences and studing Elektrical Engineering and i had a course last semester about semiconductors. This is so amazing. Thanks allot
This is definitely how sponsorships should be done
Hello. I'd like to thank you for documenting all of this !
You are making TH-cam a smarter, better place !
I assume you got into a situaiton with SpaceX or NASA therefore removed the last video (the one about the tiles), anyway great to see you are back with another great video.
elon musk ruins everything he touches
Might've even been the DoD. Remember that the other big usage case for heat shields is dropping nukes and other large explosive devices on people from space, without them melting in the process.
@@Ithirahaddoubt it, all the info was public and the tiles are readily available on eBay
I could watch that copper film peel for ages. So relaxing, but also kind of stimulating. Wouldn’t mind a separate video of just the copper film peeling. Keep it up!😃
It does look really neat under the microscope! Sometimes the film is stable until it is scratched. The point of damage becomes a nucleation site and the whole thing starts peeling from there. Pretty fun to watch.
Less soothing when you spent the last two days making it and then you watch it peel off though 🫠
this guy DOES NOT need TAIWAN.
If smiling through that level of frustration isn't worthy of a sub, I don't know what is.
15:29 interesting, I've seen graphs like this, especially where it comes to diodes, but I never knew this are actual graphs produced on an oscilloscope. I thought there we constructed only theoretically or by just measuring specific points on that graph.
You can build a curve tracer yourself with a signal gen, oscilloscope, resistor and class AB DC audio amplifier. Using the volume knob for adjusting the voltage. You may need to use two channels and set the scope to differential, maybe float the output, and create a star ground for probing to eliminate noise.
Why is copper oxide not green?
The greenish color you see on corroded copper is actually copper chloride! The oxides range from yellow to black depending on how much CuO vs Cu2O there is (and how thick)
I love your diy probe station mover. Reminds me of grad school where we did a similar hack to move about entire microscope using some ball bearing drawer slides and an x-y manipulator
My mom warned me about you......
Absolutely amazing from one end to the other. The AFM-FTIR thing is super cool! I made a simple AFM in grad school, back in the early days, so interesting to see how many different directions that idea has gone.
a video of throwing watermelons into a shredder got 320 million views. how was that more interesting than yours?
Do you have a link?? 😂
Thanks! Very educational! ❤
Always a pleasure when this channel has new content.
This is AWESOME!!! 20/10, thanks for the great video! ALSO, thanks for introducing me to the PiFM technique! that is AMAZING!
I've been trying to explain how silicon manufacturing works to my dad and this and your MEMS video will help him understand some of it, thanks. If you ever find the time or need to make a short'ish (10-20 minutes) video about it, please do so. You're explanations and graphics make it much easier to understand.
Just came across this channel, and I must say I'm blown away by your knowledge and skills! I shall continue to watch your other videos now!
Very interesting!! You could update your furnace by putting a modern PID temperature controller on it. They support multiple different kinds of thermocouples, support hysteresis settings. Not sure about the internals of what controls the filament in your furnace but people wire in PID controllers to their espresso machines with great results and even some data logging functions. Keep these awesome videos coming! I learn so much from them.
Oh man... this makes me hopeful for more home fabrication tools in the future. Really amazing work you've done here!
As usual, absolutely amazing and lots of relatable moments 😅 keep it up!
FYI, you can somewhat better points on the probe needles if you translate the points up and down in the NaOH; that results in a rounded, convex tip vs. the typical triangular shape. We used to wind up curling the points when probing because there's a bending moment from the geometry of probing. The rounded tips would last 10x longer, because there's a larger cross-section.
That's something I learned during an internship at a semiconductor lab during college, many, many, eons ago.
Will give that a try on the next batch of needles, thanks!
Very cool! You're rapidly approaching Applied Science level of projects.
I am amazed at the quality and effort behind the video, pure eye candy animations and the sponsor is actually participating in the project with their amazing technology. Since it's so sensitive, does it run in a vacuum or just clean air? The box doesn't look like a vacuum chamber.
I believe it's just open, clean atmosphere (albeit in an enclosure to help limit drafts and vibrations). My understanding is that part of the extreme sensitivity is due to the "depth" which it detects molecules. Only the very surface molecules interact with the tip, so you read the top ~20nm of the material. If there's a thin layer of contaminant you'll mostly pick that up (which is great if you're looking for contaminants! Less good if the contaminant is accidental from packaging).
They also have a "bulk" mode that looks a few hundred nanometers below the surface. I'm not entirely sure how it works to be honest, but I think they can combine that result with the "surface" results to get a picture of how materials are positioned.
@@BreakingTaps very interesting stuff! Thanks!
If TH-cam had existed when I was in high school and if I'd watched this, I imagine I'd have decided to study whatever you did instead of my current profession.
wow - I can barely fathom most of the science in this, but I need to see where this project goes. Subscribed, and godspeed!
Great content as always, but the real gem was introducing me to PhysicsInFlight. Finally understanding the underlying principles makes watching videos like yours so much more enjoyable.
Happy to share a cool new channel! It was a very helpful video to me as well, helped solidify some questions I had about how all the energy levels and bands work. Really excited to see what he does in the future, each video is building out a nice little semiconductor fab!
Nice work! Love the alignment mark pic 🤣. I feel your pain brother. This work is teaching many valuable lessons. The most important: Never bet against silicon.
At this point your videos are my favourite thing to watch. Incredible effort, and surely an entertaining outcome. Thanks for all that you're doing!
I honestly can't believe this content is available on the internet for free. I mean DIY semiconductor fabrication...!? What a time to be alive!
These videos really do a great job of inspiring me. These really help quantize intimidating topics like "DIY semiconductors". Amazing content, keep it up!
Try to create p-Si/SiO2(20-50 nm)/Al structure. p-Si - GND, Al grid contact - positive bias about 1 - 10 V. (negative bias on Al in case of n type silicon substrate). The depletion region is formed under the bias. Minority charge carriers are injected into the SiO2 upon illumination. The current can increase 100 times if intense illumination is used. You can use ITO or ZnO instead Al to increase this effect. Also you can use ZrO2, TiO, Ta2O5 dielectrics instead SiO2, and also overstoicheometric SiOx, ZrOx, etc to increase current density
Will look into it, thanks for the tip!
@@BreakingTaps These oxiides would be tohard to reliably work with in my opinion