The first high-yield, sub-penny plastic processor | Hacker News Reader

The first high-yield, sub-penny plastic processor

spectrum.ieee.org

148 pointssohkamyung69 comments

The 2000 Nobel prize in chemistry was awarded for the discovery of organic/polymer semiconductors. The field's come a long way and there has been lots of innovation happening since. Prototypes of polymer-based ICs is nothing new (I've seen plenty of them over the last decade). Metal is still preferred for interconnects, but that's pretty cheap to process in roll-to-roll. I can vouch for this being legit. That said, polymers (soft matter) doesn't do well in heat (or light), so applications are (and will forever be) restricted to low-cost throw-away devices IMO.

Shameful plug: I'm first author of this paper reporting a world record for transistor transconductance (all technologies combined). https://onlinelibrary.wiley.com/doi/pdf/10.1002/advs.2015003...

> That said, polymers (soft matter) doesn't do well in heat (or light), so applications are (and will forever be) restricted to low-cost throw-away devices IMO.

I can think of plenty of sterile single-use 'use cases' in Medical Devices where a throw away polymer processor would be useful but an existing semi-conductor processor would not make financial sense to use.

> I can think of plenty of sterile single-use 'use cases' in Medical Devices where a throw away polymer processor would be useful but an existing semi-conductor processor would not make financial sense to use.

Could you provide some examples?

any sight of 3d-printable semiconductors I can use to make 3d bulk circuits with?

Don't be ashamed. Transconductance is cool as fuck.

The technology used by PragmatIC is not based on organic semiconductors though. The only organic material is the substrate (Kapton, I assume).

The transistors are based on IGZO (Indium Gallium Zink Oxide), which is a relatively new material and is an amorphous high-channel mobility wide band gap material. The transistors are not as fast as LTPS, but have much lower leakage. Since the material does not have to by crystallized, it can be deposited at fairly low temperatures on plastic subtrate.

You may very well regard IGZO as the plastic electronics killer, because it does everything that PTTT/Pentacene etc. was supposed to do. Noble prize? Let's see.

Sadly, IGZO is not a rival to silicon when it comes to power efficiency, speed, and cost per gate. So these MCUs are nice demonstrators, but the use case is doubtful.

IGZO is great for display, large area sensing and possibly memory.

p.s.: Congrats on the research, very interesting paper! I am long out of the field, but still amazed about the progree being made. I wonder, what does the high ionic conductivity mean to device stability, especially if used with a field effect device?

>That said, polymers (soft matter) doesn't do well in heat (or light), so applications are (and will forever be) restricted to low-cost throw-away devices IMO.

since I have no knowledge in this area I will ask a question that probably doesn't make sense - would these be useful in objects that were very sturdy and shielded from variations in heat and light - or will shielding cause other problem? I'm pretty much considering high cost outdoors equipment. Or given that it's probably high cost the cost of actually using metal ones would be negligible so it wouldn't make sense to skimp?

My mind immediately went to Bell Labs Jan Schön superconducting plastics almost landing him a Nobel https://www.youtube.com/watch?v=nfDoml-Db64 (spoiler alert - it was all a scam)

We discussed a different project from the same source a little over a month ago[1].

Yes, they are making transistors out of plastic[2]. ”We coat everything on that, then peel it off and reuse the glass a carriers which means we can use silicon equipment.”

That they manage to get this to work at all is amazing. I view this as a technology demonstrator more than anything else. It looks like there's a pent up demand for low end processing that this might fill.

Still no comment about this from Sam Zeloof[3] 8)

[1] - https://fuse.wikichip.org/news/6648/reincarnating-the-6502-u...

[2] - https://www.eenewseurope.com/en/first-300mm-fab-for-the-uk/

[3] - https://twitter.com/szeloof

I kind of feel it is going to end up being used in food packaging, marketed along the lines of smart-packaging. Not sure how well the sensor market is and if some could be built in if not already, but it sure does seem to be that such cheap plastic chips only lean themselves being a cheap disposable item.

What value they will add, may well vary, but a few pennies upon many items they would be targeting, well those items mostly borderline latent-landfill at best and whilst it is cheap upfront. Is it cheap to recycle or more costly? Equally the enviromental risks of the ones not recycled is often overlooked and easy to say something is recyclable when that depends upon every single one being recycled. many recyclable plastic waste in the environment already, so I'm somewhat of mixed feelings about this in the respect of how it may well pan out being utilized.

There will be a mixture of beneficial and wasteful uses. Ppl will say the good outweighs the bad bc they don't know how to measure environmental and personal health outcomes. Ppl who question it will be villified and called luddites for besmirching the godly triology of Innovation, Science, and Dopamine. But the true crazies will be the ones connecting it to defending our country and saving children.

Now they can put digital cooking for tracking onto physical cookies you buy.

Hopefully just the packaging but i would not be surprised if someone puts them inside clothing or food.

What's the advantage of a processor that costs less than a penny? I would think that the connectors, battery, I/O, etc. would cost at least 10-25 cents for pretty much anything you can think of. So not sure how much you gain from making the CPU so cheap, unless there are other benefits like power consumption.

There are ways to print solar panels on a plastic substrate too, and led's look possible. NFC antennas are already doable with the same process.

When all the devices can be 'printed', you might be able to put electronics on pages in books/newspapers or on labels of products for zero cents because you're already putting that paper through a printing press.

Ah, the future of 3D home printers as we all dream of just took another step closer.

I love the teaching potential here - program this 4 or 8 bit processor. They are so cheap you can give them to pupils in volume that you don't care how they experiment with them, and there is potential that they are simple enough to literally fit in your head.

More of these please - the opportunities must be there

I'm curious what they will be able to hook up to one, and how it will be programmed - if you add a tiny amount of flash this could definitely be interesting, maybe have a dollar processor though.

The use cases for a student are more difficult to imagine - turn a light on an IO port is fun and somewhat education but you will need to do something more informative, like generate a coded signal, i.e. simulate an RFID chip (what I presume the commercial goal is here as well), but then vary based on some input(s)?

Its the same issue as with the current crop of low-cost computers - Raspberry Pi is great, but its most useful applications are due to its ethernet port and ability to drive a display, both of which drive up the price and board size.

It is a fair point - I just got excited about heading back to a few thousand transistors that we can understand idea. I doubt there are more than ten people who can reason about the latest things coming out of Intel.

One presumes that there must be some general purpose processor else this is just a RFID specific die (It does say ALU, PC and branch logic) so one assumes the idea is to be able to flash something some how. Then again, the cost of hooking up a die to flash it must exceed 1c so ... oh I am just talking in circles.

simple silicon micros are also ridiculously cheap, its the packaging that kills you and makes them rarely go below $1.

My reading of this seems to be a matter of them simply reducing the chip complexity (down to 4 bit) so they can obviously have fewer components on the die. Aside from that simplification, I can't find anything actually new techniques.

It also continues to annoy me that semiconductor-on-plastic gets labelled as a plastic processor.

Is the cost of the substrate a huge barrier in the field? I recall when gallium arsenide first came to market and when whatever was magic about ibm "copper" and diamond/graphite. Each time, it's wound up highly context specific and frankly not very informative to the market as I saw it.

A cheap plastic substrate for low gate count functions might be a niche which fills supply chains in smart disposables, rfid and other contexts: cans of soup which tell you when their subject to a recall.. (another dream in rfid which never happened btw)

GaAs is a workhorse in the RF world and let people replace a lot of vaccuum tubes.

Also switching power supplies I believe.

It seems like they changed a lot and got a hugely different niche.

> Instead of adapting an existing microcontroller architecture to plastic, Kumar’s team started from scratch to create a design called Flexicore. “Yield goes down very quickly as you increase gate count,” says Kumar. Knowing that, they came up with a design meant to minimize the number of gates needed. Using 4-bit and 8-bit logic instead of 16-bit or 32-bit helped. As did separating the memory that stores instructions from the memory that stores data. But they also cut down on the number and complexity of the instructions the processor is capable of executing.

> The team further simplified, by designing the processor so it executes an instruction in a single clock cycle instead of the mulit-step pipelines of today’s CPUs. Then they designed logic that implements those instructions by reusing parts, further reducing the gate count. “In general, we were able to simplify the design of FlexiCores by tailoring them to the needs of flexible applications, which tend to be computationally simple,” says Nathaniel Bleier, Kumar's student.

> All of this resulted in a 5.6 square millimeter 4-bit FlexiCore made up of just 2,104 semiconductor devices (about the same as the number of transistors in an Intel 4004 from 1971) versus some 56,340 devices for PlasticARM.

By reducing gate count on the processor, they effectively will make the programs longer (more instructions to do the same thing), but it doesn't seem like they are counting memory here..

It does raise the question... Why didn't they just use an Intel 4004... The patents have long expired and plenty of other companies make very old intel clones to out in modern products (eg. Most optical mice use a 50 year old intel to control a custom DSP to do image processing to figure out which way you're moving the mouse)

One of their (presumably motivated) choices was to use a Harvard architecture; which a 4004 isn’t.

Probably because NMOS masks don't translate well to this technology. And the chip has big standard cell auto placed and routed energy, so it's probably not that big of a deal for the benefit to be working in a sane HDL before hardening the design.

And the ubiquitous 8051s don't typically use Intel masks AFAIK for the same reasons. You really want to optimize for whatever process node currently makes sense.

>It also continues to annoy me that semiconductor-on-plastic gets labelled as a plastic processor.

It's pretty common to call current processors "silicon chips", what's wrong with calling ones on a plastic substrate plastic chips?

In silicon ICs, the silicon is actually part of the transistors.

Yes, but my impression (maybe wrong) is that the reference to silicon chips comes from the use of the substrate.

I've never heard of this before, and was confused and thought the processor itself is made of plastic. If I understand correctly it's actually a metal processor like any other on top of a flexible plastic film.

if only they could put a little red led in the core.

I realize the wafer has no bus, but I have an irrational desire for an all plastic 4 bit connection machine. I can handle failed nodes on the front machine. I want to allocate a zector of bites and make the lights blink.

Does anyone have a good idea for use cases for these? It seems a lot of use cases for "small, cheap electronics" are already covered by RFID, where you keep the processing "smarts" in the device that reads the RFID. So when do you need actual computation to take part on each tiny chip? Some kind of "smart dust" mesh network? (assuming you can pair these chips with an antenna and power source)

These devices are something like one tenth the price of a regular passive RFID tag.

That may well be down to the cost of the metal to make the inductive coil. Now if you add the ability to do comparable RFID comms using this chip, then how would the costs compare is what we do not know.

Copper is definitely not cheap but I doubt there's more than 0.5g in most passive RFID tags.

Sure... replace any barcode with one and it can remember its history of scanning.

What is the use of that, then? Also, how does that require computation?

Just brainstorming, if the chip is large enough you could just plop a reader down on it.... not as convenient as RFID but then you could sell a reader which reads info off of bananas, e.g.

Why the consumer would want to read info off of a banana, great question, but.

You had me chuckle:) There are however serious uses for very cheap sensors with brains, for example in food boxes to check that none of the packages was exposed to critical temperature unintentionally, or even prevent frauds in those places where they relabel expired food as fresh: a small chip that senses everything including air pressure besides carrying the manufacturing date and duration information,could also signal that the contents are non edible as soon as it detects that the vacuum bag has been tampered with.

What's the environmental impact for adding more cheap plastic to all of the things? Serious question. The onus should be placed on the producer to ensure they are sustainable (the opposite of how it's always been done, by the consumer). The article makes no mention of this.

I get that we're still at the laboratory phase, but I think it should still be part of the discussion.

Would it be possible to 3d print with this? That could be a game changer.

Sounds to me they are very close to being able make a plastic 6502 processor. That would be huge. There is a ton of code (and programmers) out there that could make use of it in short order.

Implementing security with this little state space will be a challenge. Especially with "macro" probe-able circuits.

Why would anyone actively work towards "ubiquitous electronics"? That sounds like a nightmare.

Some people say electronics of any sort are a nightmare. Where do you draw the line?

I'd draw the line by trying to evaluate the utility of adding electronics vs. not adding it.

RFID tags on clothing and other such items - good and reasonable, because it aids with logistics and has no major downsides for the customer (and potentially adds value, e.g. by including washing instruction for supported washing machines that could warn you if the settings don't match the clothes the put in).

Adding internet connectivity to most household appliances - questionable at best, actually reducing utility at worst (see [0]).

AI-powered fridges [1] - bad value, gimmicky idea that drives up prices, makes the system more prone to failure (added complexity), more expensive, and adds little to no value:

So you want to know what's in your fridge while standing in front of it?

Good idea: just open it to take a peek inside.

Bad idea: adding an LCD screen to the door that activates if you knock on the fridge door [2]

So you want to know what's inside your fridge at home while grocery shopping, so you don't accidentally forget to buy something or buy something you already have?

Bad idea: add AI powered item detection and internet connectivity to your fridge that then communicates with your smartphone, may get bricked remotely [0], get infected by malware, or become expensive to fix.

Good idea: add the same functionality in the form of a smartphone app that either scans barcodes or uses (local!) AI to identify items (e.g. by taking a snapshot of the fridge's content) and that can auto-generate a shopping list for refilling the fridge or by matching its content against a weekly plan of meals (feel free to implement this idea and make millions of it, I'll allow it).

There's certainly more and better examples to be found if you look for them, but these are just some quick examples illustrating what a quick cost-benefit-analysis could look like.

[0] https://www.electromaker.io/blog/article/electrolux-accident...

[1] https://www.theverge.com/2020/1/2/21046822/samsung-lg-smart-...

[2] https://www.youtube.com/watch?v=U4aeGzafgbQ

Some people say stabbing someone with a knife is bad. Some people say you should ban all knifes, including your butter knife. Where do you draw the line?

Not a good analogy. Tech can be used for good or bad. Cheap processors are not the equivalent of stabbing someone.

Indeed they aren't. The "stabbing someone" equipment was this:

> a world where absolutely every object you encounter—bandages, bottles, bananas—will have some kind of smarts

Because there is little what those smarts could do for the user but a lot of ways how they could work against the user.

> Where do you draw the line?

I would draw it around the user. Devices aren't inherently bad, unless they're used against their owners. Also, removing the concept of ownership creates a new set of problems. Cellphones would be a wonderful technology if their main purpose wasn't to profile and track users pushing them to micropay or be exposed to advertising to use functions that they could easily find, download and install on more open platforms. Ditto for Smart TVs, car electronics, home assistants, etc. they all are faces of the same problem: manufacturers went past that line and the user lost all ownership and control in exchange for low cost and ease of use, but this came at a price that some of us aren't willing to pay.

A person from the other side of the globe can bother you at any time of the day and you are expected to make yourself available for the conversation on moments notice. And it's not just for kings and queens. It's for everybody, like your annoying aunt.

I'm sure someone in the 18 century, when informed, that people are working towards that future would conclude that it's a nightmare.

I don't get your point. You mean we should just keep on building a dystopia because someone positive aspects will probably show themselves along the way?

My point is, we should always progress in all aspects of science and technology we can. And decide what to use and how to use it after we make it possible to use.

No need to go so far back, this comic calling mobile phones a horror is from 1919.

https://boingboing.net/2018/01/31/century-old-comic-accurate...

Do we not have ubiquitous electronics already? What hasn't got electronics shoved it that could meaningfully use them?

If the cost is low enough, my bathroom mirror, my shower faucet, my locks and keys, my pens and paper.

Might be interested in getting a smart (e-ink?) tattoo, if it doesn't look obvious when it's in a suitable state.

Countless Ray Kurzweilian microscopically small computers the size of blood cells a billion times more powerful per dollar in 25 years coursing through your circulatory system maintaining perfect health and interfacing with your nervous system.

2122 headlines: New Gigantic Botnet Built on Western-Atlantic Plastic Waste Agglomeration Causing Major Outages in Critical Infrastructure!

Asking the dumb question. How does plastic conduct electricity...

From the article: "The processors his team built were made using the flexible thin-film semiconductor indium gallium zinc oxide (IGZO), which can be built on plastic and continues to work even when bent around a radius of millimeters."

It's not the plastic that conducts, instead it is the foundation which the conducting material is built upon.

And what's the medium for current semiconductors

Great, a thinking banana.

We are all thinking bananas (with a little rounding) [0]

[0]: https://www.pfizer.com/news/articles/how_genetically_related...

Speak for yourself, I haven't had a thought in years. I'm allll banana.

Would a bananna have enough power to power this I'm now wondering, and a quick search yeilds a yes - https://www.thejakartapost.com/news/2017/11/07/malang-studen...

Soon it'll run EC2 instances

Hard to imagine how shuttling data thousands of kilometers to/from subpenny 4-bit CPU can make any sense. But I'm certain someone will try this because !cloud!.

Reply on news.ycombinator.com