While possibility of creating ICs in a "homelab" remains years or decades in the future, it is an innovation that is very important to strive for. The world of information has become increasingly dangerous, and trust is in short supply. We need stronger guarantees that our computational hardware is secure and backdoor-free and the best (but not perfect) way of doing it is to manufacture it yourself.
Libre Silicon [1] is one organization that is striving forward and I am hopeful they will make great progress.
Edit: when a high schooler can fabricate ICs in his garage [2], you know with enough effort, a lot of progress can be made on the problem
> researchers at the Air Force Research Laboratory and American Semiconductor recently 3D printed microcontroller SoCs from polymers on a flexible silicon substrate.
Then in the source:
"we took silicon IC chips and thinned them until they became flexible but retained their circuit functions."
So no, they did not 3D print an SoC from polymer.
Also, the chip manufacturing process is already, in a sense, additive (you deposit layers of material on each other). Unless the author is proposing printing billions of gates and wires one by one, I don't see where this is going. And if that's the case, the cost structure would exceed the traditional process many times, even if you are trying to make just a single chip.
The best actually-printed-by-an-inkjet-printer circuits I know of are simple ring oscillators of at most ten transistors or so (e.g., [1]---and they're pretty bad transistors at that), and even the simplest microcontrollers require thousands of transistors.
The paper [2] is a good review of the difficulties. Inkjet printing has at best a ≈20um minimum feature size (vs. 0.01um current transistor sizes), and the material choice is really hard: instead of silicon, you need to make semiconducting inks using funky organic molecules like 6,13‐bis(triisopropylsilylethynyl) pentacene (TIPS‐pentacene). [3] is a good recent paper trying to work around some of these limitations. So everything is still very much in the research phase.
Although inkjet printed circuits won't be anywhere near current silicon circuitry anytime soon if ever (and inkjetting transistors is the best (and kinda the only) method of 3D printing circuits we currently have), the different form factor may be useful. Circuits---even if only a few thousand transistors---could be printed on 3D geometry for interesting microfluidics capabilities, flexible circuits might make good healthcare sensors, and [4] is even studying flexible spacecraft via printed electronics for surprisingly economical space debris removal. And combining all this with 3D printed mechanical parts (via, e.g., the impressive PolyJet [5], which is basically inkjetting layers and lacks only the right materials to print electronics) will be fun.
(as to the source's Air Force Research Lab printed chips, yeah, I can't find further info, either, and agree it was probably sandpapering away most of the spare silicon bulk of an integrated circuit [impressive, useful, but not printing])
> Unless the author is proposing printing billions of gates and wires one by one ...
Just to point out, it wouldn't necessarily have to be in the "one by one" serial kind of approach of FDM 3d printing.
Something like the "one whole layer at a time" patterning approach of Stereolithography (SLA) 3D printers might turn out to work.
The closing piece of the article does say:
The future of 3D printing integrated circuits will likely
adapt a photolithography process or functional self-assembly
process to produce integrated circuits with competitive
resolution.
In a sense it is "additive", they're adding a CPU ... but then that's just what a pick and place machine does - it's how we've made PCBs for decades now
Well, two layer anyway. People have printed overlapping traces into 3d printed volumes before, but it's a very flawed process. High currents are basically out of the question (resistance too high), analog voltages are very hard (inconsistent resistance), complex circuits struggle with connecting traces, and high speed circuits struggle with grounding/shielding. It's possible to achieve shockingly good results by electroplating (https://youtu.be/Tx2B5hI4w1U) but there are heavy caveats.
There isn't really a process that approaches pcbs in a single area, much less all of them. Unfortunately that's more limiting than you'd expect- even very cheap ARM processors will probably be unstable without a ground plane and low resistance traces.
I hear there's some interesting stuff coming in 2021 (can't tie rumors to my real name) and lots of people are working on this kind of thing, but the solutions are non-obvious. SLA is having a lot of trouble because it's hard to include a lot of conductive powder without degrading resin quality (and worse, it blocks the light), sintered stuff struggles with mixed powders, and resin stuff struggles with density and continuous lines/planes.
Thomas Sanladerer had a video on this, essentially using the 3d printer as a plotter for PCBs. It creates an mask but the usual etching process is still there.
LPKF sell a funky machine which will deposit 3D traces onto a printed part. We have one of these in a lab at our uni, along with a lot of other expensive LPKF kit, but it's so specialised that I doubt anyone ever uses it. They have lots of interesting demo parts made with it, the obvious use case is antennas that are embedded in the enclosure.
I was thinking more on the line of cheap one-off PCBs that you send from your computer and take from the machine in an hour or two, like what hobbyists are doing.
This one is so much cooler but more expensive too. I hope they find some large market to grow into.
Yeah LPKF's stuff is very high end. I'm 99% sure the only reason we have any of their machines is that someone had a massive budget to blow with a tight deadline. Or they had no idea what they were buying and got sweet talked by a rep. Or they absolutely knew what they were buying and someone high up just signed off on it. We also have stuff like Metal 3D printers which are six-figures, so why not? The problem with using is that it's so expensive, so nobody is allowed to use it without training. Everyone else just submits jobs to the mech workshop and they'll etch it for you.
But.. when you think that a one-day turnaround on a small run of PCBs can cost easily $500+ if you need weird requirements, these sorts of machine can start to save money. They can do tolerances below most of the cheap fab houses (like BGA fanout, RF parts, etc). PCBTrain will charge you £400 for a 2-day turnaround on a 50x50mm 2 layer board!
At home I don't know why you'd want to 3D print a circuit versus milling or laser exposing it. You can get cheap mills with decent tolerances these days [1]. Of course LPKF also sell these sorts of machines with absurd tolerances and high prices!
In theory you can use conductive filament and a multi-material printer. One use case for that is actually making custom RF shields. Otherwise the only 3D printed circuits I've seen are kind of cheating - either using the printer to make the etch mask, or by printing channels which you can fill with conductive paint or epoxy (those are nice though).
> At home I don't know why you'd want to 3D print a circuit versus milling or laser exposing it.
I have absolutely no requirement on an additive process. I also don't think "home" will net enough of a market, it's more for "design house". What I think is missing is some box you can buy where you just add the material, send the design from a computer and an hour later you have a usable PCB.
CNC mills do kinda solve the problem, on its most basic form. But it is very common that one would need at least two layers, and it would be a huge add-on if it could apply masks and silk. The mills are also not that well packaged for the job - it would make a lot of difference if it would just be enclosed.
But anyway, that's in no way a rant about the market or anything like that. It's just that it's unsettling to see somebody jump all the way into ICs when solving the same problem on an easier level is already a useful product.
Electron beam lithography for direct writing ICs has been around for decades. Put down a layer of resist, expose with a steered electron beam. It works fine, down to at least 7nm, but it's slow. It's never been cost effective for making ICs, except for occasional one-offs. It's mostly used to make masks.
Sure. Heidelberg's recently made [1] some cool tools ("maskless aligners") that can expose a ≈150mm wafer in an hour or so down to ≈500nm resolution, which is very convenient when rapidly iterating through one-off low-resolution designs or something in a research lab (for comparison, electron beam lithography does the same to ≈10nm resolution in maybe a day). It's still part of the whole fabricate-everything-as-2D-layers-on-a-wafer paradigm, though, and has the same limitations (e.g., temperature limits) as standard silicon circuitry made with conventional photolithography.
> To increase the printing resolution, the additive manufacturing industry may need to devise a completely new printing process. Currently, inkjet 3D printing provides among the highest resolution features for 3D printing PCBs, but it remains to be seen if this process can be improved to provide resolution less than 1 micron.The future of 3D printing integrated circuits will likely adapt a photolithography process or functional self-assembly process to produce integrated circuits with competitive resolution.
I thought the whole appeal of 3D printing is tool-less manufacturing. Photolithography typically involves using photomasks, which is not exactly suitable for prototyping or low-volume production.
There's been lots of progress on DIY/academic maskless photolithography by using TI digital micromirror devices from projectors to reflect light into a microscope for exposing photoresist. They even sell a UV version that is just for this purpose (maskless light manipulation in general). As usual, the bigger issue is alignment between steps and the etchants.
Still don't know what it has to do with 3d printing, except the article sounds like GPT3 mindlessly connecting lithographic resin 3d printers and semiconductor photolithography, with a Google Scholar search on those terms as a data set.
> maskless photolithography by using TI digital micromirror devices from projectors to reflect light into a microscope
I just looked this up. It's incredible that maskless photolithography is even possible. Are these devices used in any high volume processes for manufacturing PCBs, for example?
The more common maskless route is laser direct write. That is capable to micron level dimensions, and multicolor variants have been proposed to allow submicron.
Ebeam direct write is capable to NM level features. Just slow.
formlabs released a repo to use their old model printer (form 1 / 1+) to do stuff along these lines. i've tested it as far as exposing a photoresist (after forking and fixing some issues) and it is definitely a cool thing to not have intermediate steps like transparencies or toner transfer.
I was curious too, and it turns out that Nano Dimension is publicly traded, has seen its market cap collapse over the last four years, and oddly seen a massive spike in trading volume over the last four months. I suspect this is another Microvision-like scenario where a stock on the verge of dropping into the pink sheets is getting gamed by people trying to convince others that it is the next big thing.
Another possibility: just loading the page pegs my cpu even with javascript disabled for as long as the tab is open or until I zap a bunch of elements. Could it be a new innovation in cryptocurrency mining?
3D printing is hype but beyond a broad assertion that it would be cheaper for high performance LRIP devices I don't feel like a case for it being better than current manufacturing techniques is made in the article. They also completely neglect the fact that you can 3D print complex sub micron structures today using a FIB (a common semiconductor tool) and the technology to do so has existed for over 30 years. It hasn't taken over yet for a reason.
Both FIB deposition and its cousin Focused Electron Beam-Induced Deposition (FEBID) are pretty nifty; they can cut/add materials down to maybe 1nm (!) resolution. However, they're extremely slow: they can only affect a single spot at a time and move at maybe only 100nm/s movement, so patterning a useful circuit may take days; it's mostly useful for research work. I don't think I've seen anyone figure out the chemistry to use it to print doped semiconductors, so transistors might not be doable anyway.
Really the only other sub-micrometer 3D printing category right now is two-photon lithography (shine a laser to cure a liquid resin into a solid, like common resin 3D printing), which can generally only use a single, even-more-specialized-than-required-for-inkjet material---almost always an insulating polymer---for an entire structure.
What quantity? With a traditional fab process it costs a million dollars to create the setup for your chip, but once you have that the cost of each chip is five cents. I suspect that there is basically no setup costs for this but each chips probably runs in the range of 100 each to make (I'm guessing here, but it looks to me like there is a lot of labor watching the printer). If you need only a few chips the printed process is cheaper, if you need millones the traditional process is cheaper.
This article is really too lightweight. It's not as much a review of current research as it is something that would fill whitespace in an industry mag.
Libre Silicon [1] is one organization that is striving forward and I am hopeful they will make great progress.
Edit: when a high schooler can fabricate ICs in his garage [2], you know with enough effort, a lot of progress can be made on the problem
[1] https://libresilicon.com/
[2] http://sam.zeloof.xyz/first-ic/ https://news.ycombinator.com/item?id=20657398