Ben, Love your videos and this one is very interesting but PLEASE PLEASE be aware that fuming nitric acid should always be handled either outside or in a fume hood. White fuming nitric (essentially 100% HNO3) and ESPECIALLY red fuming (HNO3 with excess dissolved NO2/N2O4) both give off nitric oxide/dinitrogen tetroxide which is INSIDIOUSLY DANGEROUS. It causes pulmonary edema and death in tiny amounts. An inhaled lethal dose does not manifest itself until days later and then it is often too late. Please be careful and maybe add this warning to your video.
Wow - this is important to know! Guys - please vote parent comment up so that it is more visible because everyone watching this video should know this.
I used to repair equipment for a company that did this, reverse engineer chips for copy write suites....they would do as you did then micrograph the chip and print it out on paper about 50 feet by 50 feet from taped together plots from a 4 foot wide plotter prints spread out on the floor, then a team of people would get on there hands and knees with colored markers and trace out the circuits of interest...
Craig Diamond I Developed a New Intergrated Circuit Subtrate Design Based on " INFUSED "Wide Band Gap " TUNNELING " Utilizing Nanotechnology Doping , How Would I Approach A Technology Company For A Licensing agreement?
Wow! That was interesting. I never dreamed those IC's looked like that inside. I actually thought the circuit was a lot larger. Thanks for posting that.
I usd to look inside early E-PROM chips with a microscope, and we measured the size of the transistors, them counted them! One 64k chip had 65,000 transistors.
I know this is a really old video, but I enjoyed watching the decapping process you are using. It reminded me of a project I played around with in the late 80's. The project used a decapped dynamic ram chip as an image sensor. The very large dip memories of the era had metal lids so they were very easily decapped with a torch or soldering iron. The dynamic ram packages are designed to be totally light proof since the capacitor based memory's charge bleeds off quicker in the light. This effect was used to make a crude image sensor in this project.
I have been trying to pop these things apart for months! Ive never found any useful videos but this finally answered my questions! Thanks! P.S. I've noticed that a lot of older CPU's (early 90's) simply have a metal cap soldered over small panel in which the chip is mounted; apply a bit of heat and you can pop them off, and since older chip architecture is comparatively large, it's very easy and interesting to see with the naked eye!
Very informative and interesting for those who, Like me, are used to do reverse ingeneering. For those who are asking. You don't need to decap microprocessors since the 80286 because they are not encapsulated in epoxy. It is very easy to open them with a sharp cheasel. The same goes to all ics in metal or ceramic case.
but is it possible do with the chip in circuit and without damaging the silicon or rip off the wires conections ? . so that the circuit can still work flawlessly. do heating it make it easier to cut ? do u heat them ? how much ? or heating the blade edge ?
You could increase the magnification of the microscope and look for hidden markings or the so called designer grafitti, which where used to trace the chips in case of industrial espionage, altough many times the designers put this markings there for fun...
One thing you could do id take infrared heat photos of the chip die during operation. That would especially be cool on larger chips such as SoC's or CPU's.
I once paid a guy a lot of money to do this. Why? In order to place the bare chip under nucleus beam, from different elements, from a large van de graaff generator at Brook-haven National Laboratories, while the chip was operating. Why? To determine how the chips (memory) would perform in space. That is - in Space Radiation - solar wind. The packaging plastic would have distorted the results. The actual chips used (on the spacecraft) were then parylene coated to prevent out-gassing.
CNCing a pocket in the die to get the ball rolling looks like a great idea! It reduces the time it takes, the amount of acid you need and the amount of toxic fumes released. Not to mention how frigging cool it is to get a photograph of a die inside the packaging, and possibly still have the chip in working order!
Nice! You can make an ultra fine needle by dipping (at short intervals) a small steel wire in NaOH and running some electricity through it, don't remember the polarity though. This way we get tiny preparation needles for microscopy.
thanks for the tip(pun intended)! last week I was searching online how to exactly do this but unable to find. I just had in mind that I've read it somewhere online and couldn't find any traces when needed.
Very cool to see how it looks inside a microchip. It's more than 30 years since I first wondered how it looks inside 😃 It was a lot smaller than I thought. Now I wonder how they manage to make it so small 😂 Thank you for making this video !!
It's all done with photographic equipment and acid. The patterns projected onto the Silicon slice were originally hand drawn in a much larger sized, then photographically scaled down. Modern factories use special machines from the Netherlands to scale shapes way down so a thin line on the drawing becomes just a few atoms wide. Also they now use computers to do the original drawings for new chips. But the old 74xx and 40xx logic chips had much fewer lines and the inventors could still use hand drawn designs.
But the nitric acid would also eat through the chip and the metal connectors! I know the gold would mostly be spared, but it's not all made of gold. How do you expect that the chip would still work in that case?
But I don't think you could observe anything, it would just look static as it did the calculation. Maybe decap, add liquid nitrogen and *O V E R C L O C K*
Ben, just a small tip from somebody who has done this professionally. After dripping, just rinse the whole IC in the acid then rinse in acetone, after that reheat and repeat the process. This will get a more even decap. We also used a flat mill bit for the initial hole. But be aware, fuming nitric acid is serious stuff.
fft2020 have you tried using a secure VPN from a non static mobile IP address. Make sure none of that stuff is traceable back to you and your golden! Oh shit! I just heard muffled rotor sounds and seen a laser dot on the wall! 😂😂😂
Yes. I think the MEMs part shows up in one of my SEM videos. The package was a metal-capped ceramic body, so I just pried the top off -- didn't need nitric acid.
How about making an inert (ptfe?) socket fixture, dissolving all of the package off to leave just the leadframe and chip, then casting it into a new, clear package with clear casting resin.
I love your videos! You have a great way about you that helps people like me try to understand very complicated subjects. I've been recently playing around with Optocouplers/optoisolators and would love to see the LED inside actually do its work. Thanks for the vids, keep it up!
When I was a kid I just put them in a vise at an angle, and crushed the top open. Mostly worked. They were from a Radio Shack grab bag, and I never did know what they were supposed to do. That was 1970s technology.
Back in... 1986??? ish. I visited the SEC research labs in Richmond Melbourne). One of the instruments they shwed off was an SEM. They had a de-capped chip from a digital watch, powered, and as the circuit switched you could see it in the SEM because the energised paths in the silicone bled a few electrons... totally astonishing to watch silicon circuits in real time using electron imaging. SEM's should be much more accessible these days... It'd be utterly awesome to see one of these circuits lit up in an SEM.
Share this with HUGHES aircraft. I almost took a job, in the late 1970s, that required milling chips to detect electrical faults. Your way seems better because of powering up the device. Great work!
When I worked at Maxwell Technologies in the early 2000's, we tested off the shelf IC's for space use. They would remove the epoxy on the chips to be tested to expose the bare die. They would then be 'Zapped' with particles that came out of a cyclotron at we rented time on at places like Texas A&M. While this was happening, test equipment would analyse them for Latch-ups, and flipped bits if they where memory chips.
Finally a technerd channel with a host that doesn't have a super annoying accent or appears to be a psychopath. Yay :D So, what's the progress on this? A friend of mine actually is part of a project for recreating DIEs for chips long out of print (mainly videogame/computer related custom chips) ... he's managed to find a way to crack them open using heat at a 80% success rate, but this way seems a little more reliable.
I’ve seen this done in an R&D lab once and they dissolved the chip encapsulation and the glass layer over the chip with hydrofluoric acid which is seriously nasty stuff. The chip was then run under an electron microscope and you could see the gates operate as light transitions. Very cool.
Is there anything you could physically see with one of these chips running? Like under the electron microscope, compare the chip pieces when its doing different things?
I worked at GTE Sylvania in the late 1970s designing vertical deflection circuits for color TVs. I watched as several 8 watt audio ICs were decapped to identify the failure mode and location. A powered hood was used to protect against fumes. It is possible to identify the difference between over voltage versus over current faults. Picture tube arcing could produce over voltage, and poor heatsinking could produce over current faults.
Put the chip in a ceramic chip socket then use trichlorethelate (however you spell that) as Japanese audiophiles do with DAC chips so they can solder silver wires on
If you de-cap some DRAM you can focus an image on the die, write all 1's to every memory location and then read the memory back. The brighter the light, the faster the memory location will discharge and you should be able to decide the image. This is how CCD camera sensors work.
That is putting it rather simply, and only focuses on the "capacitive" nature which some designs use. The actual structure of each bit in terms of doping is quite different, and CCDs are better described as photodiodes in varying bias modes (design determined) attached to bit-bucket shift registers, rather than addressable, selectively rechargeable capacitors. The corresponding decay in DRAMs is what corresponds to the dark current and some other factors in the CCD, which when combined with the fact that you want the impinging light to actually transition the cells and that cells for CCDs are read and reset at rates no faster than what the cells for DRAMs are scanned in a read and rewrite process. There are also other factors which cause differences between the two but...
really interesting. I'm trying to troubleshoot some old SNES turbo controllers. One of them had one of those "blob" ICs on it. Now, I really want to get in there!
Back in the late 80's, Micron Semiconductor (memory manufacturer) used a water drill to expose the top of their RAM chips for failure analysis. One of my side projects was using a decapped RAM chip for a vision system. Didn't have much luck with it.
This is how we would hack ASICs, laying needles on the data buses and logging the interaction of the sub structures. With enough patients the IC's would usually cough up there secrets :) As Systemrat2008 mentioned you can use HF to remove the uppermesh layers on the die. These final delayering steps can be very difficult and many of the more advanced ICs has many layers to prevent this.
I saw a de-capped chip being investigated in the old BT research labs. I seem to remember them using polarized light where they could see the pd of various tracks changing. They did not use probes as such. Very interesting though about 28 years ago.
You could use Amine solvents. I've seen Amines dissolving epoxies. You may begin with triethylamine, Ethanolamine. But the downside is that copper may get attacked. What can you do next? You can make the chip work normally in its circuit-only with the silicon visible under a microscope. In the dark you should be able to see the PN junctions & transistors glowing like little LEDs. All this is my conjecture. Perhaps its worth discovering. I suggest you begin with a 555 IC wired in timer mode. This ought to result in some visible flashes.
Fuming acid is a must for fast decap. I had usual concentrated nitric & sulfuric acid, and it was just not working at all. So I had to decap it for 3 days in my concentrated sulfuric acid with a bit of hydrogen peroxide.
WOW, I've wanted to know how to do this for nearly forty years! I always thought that chip packages were made out of some sort of porcelain (in that case, I guess then one could crack open these packages with a hammer). Apparently, I was wrong about the materials. Very cool that this is possible, though the demo makes clear that this is a project beyond my patience level.
Ben if you could decap a static RAM ic it can make a camera type device. As the RAM will change on light. This sensor is better than camera chips as your light data is read parallel and nit sent out serial by hardware. This can make lots if intresting things
Wow! You've re-invented it for us :) I've thought that the chip is hollow inside and you can open it from edges if you really want it... But now I see... Cool look BTW. It would be great to know about it's components in details there... a little "pick a point" video.
I'd love to see you power & exercise the decapped chips and see if activity is visible using an NIR camera. All conducting p-n junctions glow, dissipating that (in the case of Si) 0.7V forward drop. Unfortunately, that's not in the visible region, and I'm not sure if this glow will escape or be hidden under aluminization or something. If the chip is interesting enough, you might be able to see different parts activate at different times like some sort of brain scan. This will only work with old BJT-based chips, CMOS probably doesn't glow much at all since there aren't many p-n junctions doing anything interesting.
Very cool. I've been wanting (for so long it's embarrassing) to decap some NES CPUs and PPUs, along with some RAM chips, and integrate them into a smaller package (along with some other bits and bobs in other packages) in order to form handheld OEM-style NES consoles. Just never had the equipment or money. Good luck to ya, I look forward to more video!
I have always liked the IC's that are light erasable. there is a sticker over the little round window. removing the sticker and letting light fall on the die erases the contents. but it was interesting to see the lexicon inside, even if what ever was on them was gone.
Why doesn't the acid erode the metal layers on the chip? Is the chip passivized in some way, and if so, how would you be able to probe it while running?
Yeah, in that situation, they might actually use aluminium (Aluminium and gold are the materials that wedge/ball bond well). I was referring to this case in particular, that they may be straight gold. Another thing that's common is the use of multiple bonds in parallel for high-current stuff. When they need to handle high-current, they have multiple bond-wires running from the pad on the die to the leadframe.
Keep in mind that if you want to probe the IC under real-world conditions, not only will you need extremely fine probe wire, you'll have to attach the probe wires, then run the thing in the dark, as silicon is light sensitive (it's the same stuff solar cells and phototransistors are made of)!
I had an old PC kicking around with an Intel duo in it, some old worthless thing. I took it out and using a chisel, popped the metal cover off, only to sadly see a plain silicon die,with no circuitry on the top layer. However, with some luck I attempted to chisel the die out, I did break it in half in the process, but I managed to expose the circuitry, and I really wish I had a metallurgy microscope to look at it now, because its quite something and has incredibly small details.
This is apparently how the guy in the UK used to break the encryption of the old analog SKY satellite service... he would use these methods to expose the die on the smart cards and probe them while operating. I think it's great when something thought impossible or crazy idea is carried out and works! :)
okay, did you get offers of industrial espionage after posting this? ;-) of course not, I know... But you are a damn skilled individual and equipped with so much interesting experiences... dude - you are precious to your country! :-) All the best with your experiments, man! I really, really admire your passion for that, and the outcomes!
Every now and then, I find a micro controller while dumpster diving that is one-time programmable. I wonder, is this is feasible for turning OTPROM into PROM to salvage old MCUs and OTPROMs for personal projects?
Maybe the reason the nitric acid disappears is because the boiling point of the 100% pure acid is about 83C compared to the higher boiling point of the water azeotrope at 120C. I think you mentioned you had set the temperature to 100C on the hotplate. Much of the nitric acid would just evaporate into your lab air and cause big problems with corrosion, and edema in your lungs. Be careful. BTW I just discovered your videos and they are most interesting, thanks for sharing with us.
How is even possible to print at that level seriously how? does the little CI circuit (3:28) include any special set of features like semicounductors, resistor, etc. Or is just a single material in the circuit´s track?
They don't print the circuitry on ICs. They use photographic techniques to produce the design from a much larger transparency. Using photo resist (as used in PCB manufacture) they can etch away areas of the silicon substrate with acid and add layers of different materials using either vapour deposition or sputtering to make up the circuit one layer at a time. Depending on where the photo resist is exposed to light or not determines the pattern of areas that are either etched away with acid or added to with other materials, such as metal oxides. Eventually, after multiple processes, the circuit is complete with its various layers. Pretty much any component can be produced on the tiny circuit using the right mix of layered materials. This is how electronic devices have managed to shrink so dramatically over the last 50 years. Components on a present day computer processor are so small that even the width of the light rays (frequency) is crucial to enable them to be produced at the size they are.
Ben, Love your videos and this one is very interesting but PLEASE PLEASE be aware that fuming nitric acid should always be handled either outside or in a fume hood. White fuming nitric (essentially 100% HNO3) and ESPECIALLY red fuming (HNO3 with excess dissolved NO2/N2O4) both give off nitric oxide/dinitrogen tetroxide which is INSIDIOUSLY DANGEROUS. It causes pulmonary edema and death in tiny amounts. An inhaled lethal dose does not manifest itself until days later and then it is often too late. Please be careful and maybe add this warning to your video.
Wow - this is important to know! Guys - please vote parent comment up so that it is more visible because everyone watching this video should know this.
Sadly we have to report that Ben has....
Ha! REAL MEN breath in nitric oxides and don't even blink!!!
I stilkkjlsgk ohh help....... hksbdfsdfajksaf;nsadfksfdkjsdf
accidentalvoyeur m
Sounds like bullshit. How many real people have died 3 days after playing with fuming nitric acid?
I used to repair equipment for a company that did this, reverse engineer chips for copy write suites....they would do as you did then micrograph the chip and print it out on paper about 50 feet by 50 feet from taped together plots from a 4 foot wide plotter prints spread out on the floor, then a team of people would get on there hands and knees with colored markers and trace out the circuits of interest...
what year was this
Nic Skram may be 2015)
Copyright* Suits* their*
Craig Diamond I Developed a New Intergrated Circuit Subtrate Design Based on " INFUSED "Wide Band Gap " TUNNELING " Utilizing Nanotechnology Doping , How Would I Approach A Technology Company For A Licensing agreement?
Craig Diamond That sound super interesting!
I gotta say... this is a cool thing to do. It was really neat seeing the microscopic view of the chip circuits. good channel.
Blows my mind the precision of the IC traces
Wow! That was interesting. I never dreamed those IC's looked like that inside. I actually thought the circuit was a lot larger. Thanks for posting that.
I usd to look inside early E-PROM chips with a microscope, and we measured the size of the transistors, them counted them! One 64k chip had 65,000 transistors.
I know this is a really old video, but I enjoyed watching the decapping process you are using. It reminded me of a project I played around with in the late 80's. The project used a decapped dynamic ram chip as an image sensor. The very large dip memories of the era had metal lids so they were very easily decapped with a torch or soldering iron. The dynamic ram packages are designed to be totally light proof since the capacitor based memory's charge bleeds off quicker in the light. This effect was used to make a crude image sensor in this project.
I have been trying to pop these things apart for months! Ive never found any useful videos but this finally answered my questions! Thanks!
P.S. I've noticed that a lot of older CPU's (early 90's) simply have a metal cap soldered over small panel in which the chip is mounted; apply a bit of heat and you can pop them off, and since older chip architecture is comparatively large, it's very easy and interesting to see with the naked eye!
Very informative and interesting for those who, Like me, are used to do reverse ingeneering. For those who are asking. You don't need to decap microprocessors since the 80286 because they are not encapsulated in epoxy. It is very easy to open them with a sharp cheasel. The same goes to all ics in metal or ceramic case.
but is it possible do with the chip in circuit and without damaging the silicon or rip off the wires conections ? . so that the circuit can still work flawlessly. do heating it make it easier to cut ? do u heat them ? how much ? or heating the blade edge ?
You could increase the magnification of the microscope and look for hidden markings or the so called designer grafitti, which where used to trace the chips in case of industrial espionage, altough many times the designers put this markings there for fun...
Yes, like the famous “Bill Suxx” logo done by intel cpu designers
@@MaverickM1 too bad that was just a hoax and didn't really happen
One thing you could do id take infrared heat photos of the chip die during operation. That would especially be cool on larger chips such as SoC's or CPU's.
That would be pretty cool, nice idea
I once paid a guy a lot of money to do this. Why? In order to place the bare chip under nucleus beam, from different elements, from a large van de graaff generator at Brook-haven National Laboratories, while the chip was operating. Why? To determine how the chips (memory) would perform in space. That is - in Space Radiation - solar wind. The packaging plastic would have distorted the results. The actual chips used (on the spacecraft) were then parylene coated to prevent out-gassing.
Nicely edited!!
Hey you!! 😘😁
👍
This was extremely interesting and informative! Thank you for taking the time to share your experiment with us. Cheers from Australia, Rolf
How about decapping the proprietary blob chips / chips in electronic toys? This is pretty cool!
Bump
Aceatone?
Aceatone?
No, acetone is far too weak.
CNCing a pocket in the die to get the ball rolling looks like a great idea! It reduces the time it takes, the amount of acid you need and the amount of toxic fumes released.
Not to mention how frigging cool it is to get a photograph of a die inside the packaging, and possibly still have the chip in working order!
Nice! You can make an ultra fine needle by dipping (at short intervals) a small steel wire in NaOH and running some electricity through it, don't remember the polarity though. This way we get tiny preparation needles for microscopy.
thanks for the tip(pun intended)! last week I was searching online how to exactly do this but unable to find. I just had in mind that I've read it somewhere online and couldn't find any traces when needed.
Very cool to see how it looks inside a microchip. It's more than 30 years since I first wondered how it looks inside 😃
It was a lot smaller than I thought. Now I wonder how they manage to make it so small 😂
Thank you for making this video !!
It's all done with photographic equipment and acid. The patterns projected onto the Silicon slice were originally hand drawn in a much larger sized, then photographically scaled down. Modern factories use special machines from the Netherlands to scale shapes way down so a thin line on the drawing becomes just a few atoms wide. Also they now use computers to do the original drawings for new chips. But the old 74xx and 40xx logic chips had much fewer lines and the inventors could still use hand drawn designs.
But the nitric acid would also eat through the chip and the metal connectors! I know the gold would mostly be spared, but it's not all made of gold. How do you expect that the chip would still work in that case?
It would be very interesting to decapp a ATTINY/MEGA and let it do an complex calculation, and then view it under the electron-microscope.
But I don't think you could observe anything, it would just look static as it did the calculation. Maybe decap, add liquid nitrogen and *O V E R C L O C K*
Ben, just a small tip from somebody who has done this professionally. After dripping, just rinse the whole IC in the acid then rinse in acetone, after that reheat and repeat the process. This will get a more even decap. We also used a flat mill bit for the initial hole. But be aware, fuming nitric acid is serious stuff.
If I search for "fuming nitric acid" on ebay here where I live, black helicopters will be hovering outside my window in 5 minutes
Well, make sure and put it on RUclips if you do!
Do it!
This post has been recorded by the secret service already, so no worries ;-)
fft2020 have you tried using a secure VPN from a non static mobile IP address. Make sure none of that stuff is traceable back to you and your golden! Oh shit! I just heard muffled rotor sounds and seen a laser dot on the wall! 😂😂😂
@@MrDegsy69 On the wall is OK. On your chest, sorry!
"mostly interested in playing with nitric acid" I appreciate you honesty.
Yes. I think the MEMs part shows up in one of my SEM videos. The package was a metal-capped ceramic body, so I just pried the top off -- didn't need nitric acid.
De-cap an Atmega :0 and then couple it to a liquid nitrogen cooled heatsink and play with clock speeds......................
Do Atmegas have external clock inputs?
They do.
I think almost all Atmel AVR chips have a clock input. Even the 6 pin devices have a clock input.
Use an Attinny85 it's cheaper and have less pins.
A really nice idea. oldschool EPROM chips with clear silica windows have already been used. Needless to say that exposure times are quite long.
Those bond wires are pure gold, so there is a good reason to do this.
Very cool. Very very cool. The decapped logic ic's looked like dead bugs.
another approach is to use rosin (yes, pine resin). when heated to 320-360C the abietic acid eats the epoxy.
How about making an inert (ptfe?) socket fixture, dissolving all of the package off to leave just the leadframe and chip, then casting it into a new, clear package with clear casting resin.
See how the IC reacts to various frequencies and intensities of light. Also try a strobe light to mess with the logic gates.
Would really love to see that!
In the 80's I used to use a screwdriver and a hammer to split them on their seams the long way and had the same results. Perfect separation
That chip looked gorgeous, nice work.
Cool idea! The nitric acid will attack the leadframe, but it might be possible to get it done fast enough to still have everything intact.
Clear and concise, short and sweet, thanks this was great!
I love your videos! You have a great way about you that helps people like me try to understand very complicated subjects. I've been recently playing around with Optocouplers/optoisolators and would love to see the LED inside actually do its work. Thanks for the vids, keep it up!
When I was a kid I just put them in a vise at an angle, and crushed the top open. Mostly worked. They were from a Radio Shack grab bag, and I never did know what they were supposed to do. That was 1970s technology.
I always wondered how is it inside an IC! Thanks for the experiment.
Back in... 1986??? ish. I visited the SEC research labs in Richmond Melbourne). One of the instruments they shwed off was an SEM. They had a de-capped chip from a digital watch, powered, and as the circuit switched you could see it in the SEM because the energised paths in the silicone bled a few electrons...
totally astonishing to watch silicon circuits in real time using electron imaging.
SEM's should be much more accessible these days... It'd be utterly awesome to see one of these circuits lit up in an SEM.
Cool! Yeah, have no idea what to do, but it's amazing to see that clean chip and wires!
Share this with HUGHES aircraft. I almost took a job, in the late 1970s, that required milling chips to detect electrical faults. Your way seems better because of powering up the device. Great work!
When I worked at Maxwell Technologies in the early 2000's, we tested off the shelf IC's for space use.
They would remove the epoxy on the chips to be tested to expose the bare die.
They would then be 'Zapped' with particles that came out of a cyclotron at we rented time on at places like Texas A&M.
While this was happening, test equipment would analyse them for Latch-ups, and flipped bits if they where memory chips.
Great experiment. Thanks for showing.
RODALCO2007 I see you everywhere.
RODALCO2007 especially on photons vids.
Finally a technerd channel with a host that doesn't have a super annoying accent or appears to be a psychopath. Yay :D
So, what's the progress on this? A friend of mine actually is part of a project for recreating DIEs for chips long out of print (mainly videogame/computer related custom chips) ... he's managed to find a way to crack them open using heat at a 80% success rate, but this way seems a little more reliable.
I’ve seen this done in an R&D lab once and they dissolved the chip encapsulation and the glass layer over the chip with hydrofluoric acid which is seriously nasty stuff. The chip was then run under an electron microscope and you could see the gates operate as light transitions. Very cool.
This is proper decapping. We should send these instructions to Linus or JZTwoCents to further improve their decapping methods for CPUs :P
Is there anything you could physically see with one of these chips running? Like under the electron microscope, compare the chip pieces when its doing different things?
I don't have a damm idea what this guy is talking about most of the time, But man; his work keeps me fixated ! 🤪
We decapped IC's to do FA (failure analysis). I always wondered how that was done. Thanks. PS, Nice work on the decap.
You did it, but didn't know how to do it?
WAT?
:-/
I worked at GTE Sylvania in the late 1970s designing vertical deflection circuits for color TVs. I watched as several 8 watt audio ICs were decapped to identify the failure mode and location. A powered hood was used to protect against fumes. It is possible to identify the difference between over voltage versus over current faults. Picture tube arcing could produce over voltage, and poor heatsinking could produce over current faults.
You got some nice handy tools to work with! Intesting video :)
Put the chip in a ceramic chip socket then use trichlorethelate (however you spell that) as Japanese audiophiles do with DAC chips so they can solder silver wires on
That's really cool. I've always wanted to see inside one of those chips.
good idea. i wonder if that would work for parts that are potted.
this is the coolest shit I think I've ever seen in my life. I've always wanted to know how a chip works and what's inside.
Love the sight of the ic
What chip is at 3:30? Not terribly complex, but complex enough. An amazing amount of engineering and science in its own right. Thanks.
If you de-cap some DRAM you can focus an image on the die, write all 1's to every memory location and then read the memory back. The brighter the light, the faster the memory location will discharge and you should be able to decide the image. This is how CCD camera sensors work.
That is putting it rather simply, and only focuses on the "capacitive" nature which some designs use. The actual structure of each bit in terms of doping is quite different, and CCDs are better described as photodiodes in varying bias modes (design determined) attached to bit-bucket shift registers, rather than addressable, selectively rechargeable capacitors. The corresponding decay in DRAMs is what corresponds to the dark current and some other factors in the CCD, which when combined with the fact that you want the impinging light to actually transition the cells and that cells for CCDs are read and reset at rates no faster than what the cells for DRAMs are scanned in a read and rewrite process. There are also other factors which cause differences between the two but...
Thank you very much for the upload. Always wanted to take a look in IC.
Well that was damn interesting... I've always wondered how enthusiasts reverse engineer custom chips for game consoles... Thanks....
Super cool. Always wondered what these look like inside
I agree. If anyone knows of a chip with an Easter Egg, please let me know.
Decapping a Core i7.
+Isaías J. putting it in close to 0k surrounding(liquid nitrogen?) and overclock it to the max
mathias de potter Yes.
+mathias de potter It may not work due to the cold bug unfortunately :/
Ciro Santilli
we're not trying to reverse engineer it though..
Romanator X
it is possible, you don't put the cpu in before starting, you wait till the computer starts.. before plopping it in...
"I was just interested in playing with nitric acid" Love this line!
Interesting stuff does not have to have a goal, it's all about the lesson. Thanks and cheers to you.
So this is how they reverse engineer IC's.
Cool to see, nice video !
really interesting. I'm trying to troubleshoot some old SNES turbo controllers. One of them had one of those "blob" ICs on it. Now, I really want to get in there!
Back in the late 80's, Micron Semiconductor (memory manufacturer) used a water drill to expose the top of their RAM chips for failure analysis. One of my side projects was using a decapped RAM chip for a vision system. Didn't have much luck with it.
dude whoa! that is fascinating- you need some needle tip probes to test those... build some? Good video buddy!
As always, a very informative and well presented video on a subject that I've always wanted to foray into. Thanks!
Are they still functional? At least one of them?
This is how we would hack ASICs, laying needles on the data buses and logging the interaction of the sub structures. With enough patients the IC's would usually cough up there secrets :) As Systemrat2008 mentioned you can use HF to remove the uppermesh layers on the die. These final delayering steps can be very difficult and many of the more advanced ICs has many layers to prevent this.
Awesome job Ben! I was on the fence before, but after this video I had to sub!
I saw a de-capped chip being investigated in the old BT research labs. I seem to remember them using polarized light where they could see the pd of various tracks changing. They did not use probes as such. Very interesting though about 28 years ago.
You could use Amine solvents. I've seen Amines dissolving epoxies. You may begin with triethylamine, Ethanolamine. But the downside is that copper may get attacked.
What can you do next? You can make the chip work normally in its circuit-only with the silicon visible under a microscope. In the dark you should be able to see the PN junctions & transistors glowing like little LEDs. All this is my conjecture. Perhaps its worth discovering. I suggest you begin with a 555 IC wired in timer mode. This ought to result in some visible flashes.
Fuming acid is a must for fast decap.
I had usual concentrated nitric & sulfuric acid, and it was just not working at all.
So I had to decap it for 3 days in my concentrated sulfuric acid with a bit of hydrogen peroxide.
WOW, I've wanted to know how to do this for nearly forty years! I always thought that chip packages were made out of some sort of porcelain (in that case, I guess then one could crack open these packages with a hammer). Apparently, I was wrong about the materials. Very cool that this is possible, though the demo makes clear that this is a project beyond my patience level.
High temperature chips are ceramic.
PDIP chips are plastic and CDIP are ceramic. Most 7400 series chips (or other simple, low temp chips) now are PDIP because it costs less.
everything changes, stay open-minded
I had great success using a pair of side cutters, and popping the top off. That was in the late 70's and early 80's.
Ben if you could decap a static RAM ic it can make a camera type device. As the RAM will change on light. This sensor is better than camera chips as your light data is read parallel and nit sent out serial by hardware. This can make lots if intresting things
Dear Sir thanks for the video, but how do you remove the type which uses ceramic encapsulation from the IC silicon dies?
This is just awsome!
Wow! You've re-invented it for us :) I've thought that the chip is hollow inside and you can open it from edges if you really want it... But now I see... Cool look BTW. It would be great to know about it's components in details there... a little "pick a point" video.
2:22 The nitric acid is very strong, and it reacts with the packaging by popping open the epoxide ring structures in the plastic molecules.
I'd love to see you power & exercise the decapped chips and see if activity is visible using an NIR camera. All conducting p-n junctions glow, dissipating that (in the case of Si) 0.7V forward drop. Unfortunately, that's not in the visible region, and I'm not sure if this glow will escape or be hidden under aluminization or something. If the chip is interesting enough, you might be able to see different parts activate at different times like some sort of brain scan.
This will only work with old BJT-based chips, CMOS probably doesn't glow much at all since there aren't many p-n junctions doing anything interesting.
Very cool. I've been wanting (for so long it's embarrassing) to decap some NES CPUs and PPUs, along with some RAM chips, and integrate them into a smaller package (along with some other bits and bobs in other packages) in order to form handheld OEM-style NES consoles. Just never had the equipment or money. Good luck to ya, I look forward to more video!
Why not use an organic solvent and reflux to dissolve away the epoxy resin?
This is cool. I had no idea how small the chip was under there.
I have always liked the IC's that are light erasable. there is a sticker over the little round window. removing the sticker and letting light fall on the die erases the contents. but it was interesting to see the lexicon inside, even if what ever was on them was gone.
Why doesn't the acid erode the metal layers on the chip? Is the chip passivized in some way, and if so, how would you be able to probe it while running?
Yeah, in that situation, they might actually use aluminium (Aluminium and gold are the materials that wedge/ball bond well).
I was referring to this case in particular, that they may be straight gold.
Another thing that's common is the use of multiple bonds in parallel for high-current stuff.
When they need to handle high-current, they have multiple bond-wires running from the pad on the die to the leadframe.
Did you find out if the chips were still functional after exposing the "innards"?
That's what I've been looking for!
Keep in mind that if you want to probe the IC under real-world conditions, not only will you need extremely fine probe wire, you'll have to attach the probe wires, then run the thing in the dark, as silicon is light sensitive (it's the same stuff solar cells and phototransistors are made of)!
do you try is that still works ?
decap a large CPU. photograph it and sell the pictures. The landscape looks nice and could be a talking point.
There's no cap on a CPU, and they are made on much smaller process (nm scale), you can't see anything with an optical microscope like here.
I think he means like a microcontroller like the AVR chips. Those you'll be able to see cool looking circuitry, and they're only a buck or 2.
I had an old PC kicking around with an Intel duo in it, some old worthless thing. I took it out and using a chisel, popped the metal cover off, only to sadly see a plain silicon die,with no circuitry on the top layer. However, with some luck I attempted to chisel the die out, I did break it in half in the process, but I managed to expose the circuitry, and I really wish I had a metallurgy microscope to look at it now, because its quite something and has incredibly small details.
This is apparently how the guy in the UK used to break the encryption of the old analog SKY satellite service... he would use these methods to expose the die on the smart cards and probe them while operating. I think it's great when something thought impossible or crazy idea is carried out and works! :)
Would it be possible to use fuming sulfuric acid instead? It should be less hazardous and more readily available.
okay, did you get offers of industrial espionage after posting this? ;-) of course not, I know... But you are a damn skilled individual and equipped with so much interesting experiences... dude - you are precious to your country! :-) All the best with your experiments, man! I really, really admire your passion for that, and the outcomes!
Every now and then, I find a micro controller while dumpster diving that is one-time programmable. I wonder, is this is feasible for turning OTPROM into PROM to salvage old MCUs and OTPROMs for personal projects?
Maybe the reason the nitric acid disappears is because the boiling point of the 100% pure acid is about 83C compared to the higher boiling point of the water azeotrope at 120C. I think you mentioned you had set the temperature to 100C on the hotplate. Much of the nitric acid would just evaporate into your lab air and cause big problems with corrosion, and edema in your lungs. Be careful. BTW I just discovered your videos and they are most interesting, thanks for sharing with us.
How is even possible to print at that level seriously how? does the little CI circuit (3:28) include any special set of features like semicounductors, resistor, etc. Or is just a single material in the circuit´s track?
They don't print the circuitry on ICs. They use photographic techniques to produce the design from a much larger transparency. Using photo resist (as used in PCB manufacture) they can etch away areas of the silicon substrate with acid and add layers of different materials using either vapour deposition or sputtering to make up the circuit one layer at a time. Depending on where the photo resist is exposed to light or not determines the pattern of areas that are either etched away with acid or added to with other materials, such as metal oxides. Eventually, after multiple processes, the circuit is complete with its various layers. Pretty much any component can be produced on the tiny circuit using the right mix of layered materials. This is how electronic devices have managed to shrink so dramatically over the last 50 years. Components on a present day computer processor are so small that even the width of the light rays (frequency) is crucial to enable them to be produced at the size they are.
Yes, now there is a MOS fan in the house. Amazing what they did, making chips cheap by being abile to fix the bad ones in a run.