The future of electronics: How small can we go?

Tiny Matters

The first computer was created in 1945 and came in at double the size of a one-bedroom apartment. Just 50 years later, the architecture of the computer on a chip that measured just 7.44 by 5.29 millimeters in size. And now, computers have gotten smaller and smaller [looks down at Apple Watch]. So how did we go from apartment-sized calculators to the tiny devices we use to look up cat pictures when we’re bored? And just how much smaller can we go?

Transcript of this Episode

Deboki Chakravarti: You’ve probably seen pictures of those old computers. Not like the bulky ones that Sam and I used when we were younger. I’m talking about computers like the Electrical Numerical Integrator and Calculator, also known as ENIAC. Finished in 1945, ENIAC had a footprint of about 1,800 square feet. For scale, the average 1 bedroom apartment in the US is somewhere around 700 square feet.
 
But by 1995, to celebrate the 50th anniversary of ENIAC, a group of students at the University of Pennsylvania were able to recreate the architecture of the computer on a chip that measured just 7.44 by 5.29 millimeters in size.

Sam Jones:
And since then, as I’m sure you know, computers have gotten smaller and smaller. So how did we go from apartment-sized calculators to the tiny devices we use to look up cat pictures when we’re bored? And just how much smaller can we go?

Welcome to Tiny Matters. I’m Sam Jones and I’m joined by my cohost Deboki Chakravarti. Today we’re talking about transistors, the electron-passing, signal-amplifying switches that make so much of our technology today possible. We’ll be talking about what they are, how they were invented, and why their future is about size, but also so much more.
This episode was suggested by one of our listeners, Mattias from Sweden. Thanks so much Mattias.

Deboki: Sam, I think we’ve been very upfront about our backgrounds, which are very much not in electrical engineering.

Sam: Yes, they are definitely not.

Deboki: So going to the basics on transistors meant that we needed some expert help, which is why we turned to Sameer Sonkusale, who is a professor of electrical and computer engineering at Tufts University.

Sameer Sonkusale: You can think of transistors as switches, which you turn on and off using a voltage that you apply at one of the terminals. And with just switches you can do a lot of computation. But in terms of what I do, I try to use them as analog devices. These are analog transistors. So you want to amplify signals, you wanna amplify signals that are really weak. And so the transistors allow you to essentially perform those kinds of signal amplification electronically.

Sam: To understand how having switches in the form of transistors works in our technology today, let’s go back to the ENIAC so we can understand how computers worked before transistors.

A lot of that 1,800 square feet that the ENIAC took up was occupied by vacuum tubes. These were our pre-transistor transistors, acting as switches or signal amplifiers in the ENIAC’s circuitry. These tubes were about the size of a human thumb, and they relied on heat to release electrons from a cathode, which would then travel through the tube when you applied a voltage to the plate at the other end.

Deboki: That movement of electrons creates a current. And importantly, you could modify the current through the voltage you apply to a control grid that those electrons travel through. If you increase that voltage by even a little bit, you increase the current at the plate by a lot, which is how you get your amplified signal or your on/off switch.

Using these vacuum tubes enabled ENIAC to do a lot of math that was very impressive at the time, but the vacuum tubes also had some disadvantages. As we’ve mentioned, they took up space. They also relied on a lot of heat and maintenance, so by the end of World War II, scientists were trying to find another way.

Sam: Scientists were excited by a relatively new type of material called a semiconductor, which can either act as an insulator or a conductor of electricity depending on different variables, for instance the temperature. Semiconductors are made up of atoms that might not typically conduct electricity, but they have these loose electrons that aren’t bound to other atoms and are easy to knock off. And when they do get knocked off, they wander through the material and create a current.

Deboki: In 1945, a team of scientists at Bell labs consisting of William Shockley, Walter Brattain, and John Bardeen began working together to see if they could use semiconductors to make transistors. It took some time, but eventually Bardeen and Brattain were able to assemble one out of the element germanium, as well as plastic and gold foil.

William Shockley had been excluded from this work, and he was not happy about it. So he went off and holed himself up in a hotel room for several weeks to produce a different design that would ultimately prove to be more reliable.
Sam: It’s one of those stories that highlights some of the pettier moments that make science happen. But we should also note that Shockley would go on to be known for his very racist and very eugenicist views, which does a lot to take away from his legacy as the person whose work essentially started Silicon Valley.

Deboki: And speaking of silicon…

Sam: Very smooth.

Deboki: Well, the thing about silicon is that it’s great for transistors. It has 4 electrons that are available to bond to stuff, which means it’ll easily bond to 4 other silicon atoms. But the real magic happens when you make silicon impure, through a process called doping.

Basically, instead of having all those silicon atoms bound to each other, you introduce another element. And depending on what you choose, you can get a few different outcomes.

Sam: In some cases, you’ll end up having an excess of electrons floating around, creating a current. That type of silicon is called n-type.

In other cases, you have a sort of void of electrons instead, creating a positively-charged hole that can move through the material. And that type of silicon is called p-type.

And it’s important to note that while we’re talking about electrons and positive holes floating around, the n-type and p-type silicon are still electrically neutral—they don’t have a charge. But when you put them next to each other and apply a voltage, those electrons start moving and you start to get a current.

Deboki: Maybe this is starting to sound a little familiar because it sounds a bit like the vacuum tube we talked about in the ENIAC, where the current going from one end to the other was controlled by some kind of voltage that you’re applying to the tube. The big difference is that instead of heat and plates and vacuums, we’re now working with different types of silicon arranged next to each other.

Sam: There are a lot of different types of transistors out there, and it can get a little overwhelming keeping track of them all. But for silicon transistors, the differences generally lie in the configuration of these n- and p-type silicon. And how you arrange them all comes down to a really basic set of things.

Sameer: There is a part that controls the switch. And the switch is a semiconducting material of some sort. So in silicon, silicon is the semiconductor. And then you have a gate, which we call an electrode that turns it on and off. And then there are two ends to it. And the two ends of switches, we call them a drain or a source. Now in silicon, the way these transistors work is the gate, the one that's controlling the switch, by applying a voltage at the gate, will turn on and off the semiconducting channel. So it'll basically dope it with more electrons or extract electrons, or dope it with more holes, to control when the switch turns on and off.

Deboki: So our computers are filled with chips that are in turn filled with these transistors, essentially etched into a wafer of purified silicon crystal. It’s a very cool and involved process that uses lasers and even diamond cutters.
But what’s been especially exciting over the past few decades is that, every year, it seems like people get better and better at figuring out how to make these transistors smaller and smaller.

Sameer: Silicon is the work horse behind all the transistors we use today inside our cell phones, in your computers, in your smartwatches. And they have gone through, you know, this Moore’s law of scaling, which is basically making transistors smaller and smaller and they're more efficient, they run faster, and they're basically taking computing to the next level.

Sam: Sameer is talking about Moore’s Law, which I had not heard about before we chatted with him. So Gordon Moore was one of the founders of Intel, and one of his famous predictions was that as companies learned to both make smaller components and get better at arranging those components on their silicon wafers, the number of transistors on a chip would go up, doubling every two years.

And so far, that prediction has largely held, which is why we have the technology we have today.

Sameer: The entire world is trying to make transistors smaller, more efficient, and that has to continue. That dimension is necessary because we are addicted to, you know, high speed computing. We can't wait for that video to download any faster. You want that video there in a microsecond as soon as you press the button. So we want high speed computing, we want processing, and with artificial intelligence and data machine learning, we want these computing engines to run even faster so that they can essentially compute much faster. So that's gonna happen. And I think that dimension is the reason why the silicon transistors are getting better and smaller and more efficient.

Deboki: These are incredible accomplishments, but the question is how long can we continue at this rate of progress? While announcing their latest products in 2022, Intel and Nvidia took opposing views on whether or not Moore’s law is dead. There’s a lot to be said about whether Moore’s Law should be taken as a law of physics or as a reflection of economic progress, but the fact is that there are physical limitations that scientists are confronting when it comes to making transistors smaller.

For one, smaller transistors means that we get closer and closer to having to confront properties of electrons we don’t have to deal with at larger scales. Like quantum tunneling, where electrons start to pass through thin walls and become difficult to control. So we asked Sameer the question we all want to know:

Sam to Sameer: How small do you think a transistor could be one day?

Sameer: Single atom, right? A drain and source separated by a single molecule or a single atom. I think we are very close with the next generation of transistors that Intel or TSMC, which is Taiwan Semiconductor Manufacturing Corporation, some others who are building…the transistors are getting really, really close to that limit. So during my whole career as an electrical engineer, every few years someone would say, okay, we have reached the limit. There's no way we can go down below that. This is it. This has been happening my entire career and I have been around for many years and we have never hit that limit. We've always been able to figure out a way to get that small. So then you have to ask, what is the most fundamental limit? Well, the most fundamental is the single thing between the two ends of a transistor, and that could be as small as a single atom or a molecule. That's the end, right? That's where you basically stop. So we can get there, I'm sure.

Deboki to Sameer: Is that something special about hitting that limit beyond the fact that it's a limit? Would that look like instant computing or anything like that, or it's just that is the limit and whatever speed we are when we're there, whatever power we are when we're there, that is the limit of the power we'll get. Or is that a design challenge? How do we integrate these transistors? Like what do we do when we hit that limit?

Sameer: You hit on all the points that are all relevant. So you mentioned all of this, and they're right on the money. So first is, they're gonna be unique transistors, there are going to be quantum tunneling effects happening because they're so small at that level that the transistor behavior is going to be unique. And so now you, as a circuit designer, you have to exploit or explore architectures of your circuit stack or how you design circuits using these interesting properties of these transistors that you probably never had to use before. You will reach a limit to how much power and how much computing and how much speed you can get based on the underlying limits of physics. So, it’ll be interesting work. There'll be lots of very interesting opportunities for designing new kinds of architectures. It will definitely be better than the ones we are using today. It should happen in our lifetimes for sure.

Deboki: So we started working on this episode thanks to Mattias’ question about how small transistors could be. But as we were working on it, we got more curious about the bigger question: what does the future of transistors look like?
Obviously we all love our phones and tablets and laptops, and we all want them to be faster so we can download more Tiny Matters episodes while watching TikToks about obscure hobbies while also messaging our friends about a show we just watched. But the future of technology is more than our computers, which brings us to Sameer’s work.

Sam: Sameer works with biosensors, but that wasn’t always his focus.

Sameer: I am trained as an electrical engineer, my PhD was in electrical engineering. All the way back when I was a kid and I loved tinkering with radios and phones and, and things like that. And my dad was an electrical engineer. I did my PhD at UPenn in electrical engineering. I was a circuit designer, I designed lots of integrated circuits for very high performance devices using silicon.

But when I started my faculty career, I realized that I wanted to find out and see if I can use some of my background and expertise in the problems that need solving. And what I realized soon enough was that most of the interesting problems are at the boundaries of various disciplines. They're not inside electrical engineering or they're not inside, let's say biology. The problems which needed addressing were at the boundaries. But when I did my first sabbatical at Brigham and Women's Hospital, I was looking at problems that, you know, hadn't been solved. And we quickly realized that there was an area or a problem that needed solving was this whole area of wound healing. And the reason being that most of the wounds are very complex and people, the way they were being treated was using bandages that had really no smartness or no sensors or nothing very intelligent about it.

So we came up with an entire toolkit of sensors and platforms that are flexible, that can be used for monitoring how the wounds are healing and if you can intervene at the right time so that it can deliver drugs in real time. So we have basically a closed loop, smart bandage kind of platform that would allow us to monitor wounds in real time.

And so that has taken me down the track now where we are developing biomedical platforms for many different problems. So, you know, starting from ingestibles for studying the cut microbiome to doing smart sutures for surgical wounds, to smart bandages.

Sam: The work that Sameer’s lab and other groups like his are doing is still in the early stages, looking at developing tools to monitor things like oxygen level or other aspects of wound healing that might be important to different patients.

But developing sensors that can monitor a wound comes with some challenges you don’t encounter as often with computers.

Sameer:
I looked at the biomedical world and realized that what we need for this is a different version of the transistors that would be more amenable to interacting with the wet biological world. And silicon is a dry, rigid universe. So there is a little bit of a mismatch in terms of where we are in the transistor world when it comes to interfacing with biology and life sciences.

Deboki: So they figured out something unique: transistors that are built into a thread.

Sameer:
And what a thread transistor does, it actually does the same thing, but instead of relying on silicon, it uses some other semiconducting materials that are more suitable for a thread, like individual fiber or textile threads as substrates. So you still have the two ends of the switches. So instead of a dry silicon dioxide or some of these oxide dielectrics that are used in silicon, which can break if you flex them, we use electrolyte. So at the very basic level, it's basically just a single thread fiber that is behaving like a transistor.

Deboki to Sameer: Is the size of these fibers, are they like what? In terms of diameter? In terms of length? What would it look like?

Sameer:
So these are not nanoscale devices. These are what I like to call micron scale or even macro scale devices. Because we are coming from the world of the sort of wearables or wound healing or smart textiles, there is no need for those electronics devices to go that small. We are not trying to pack everything in a single piece or crystal of silicon so that you can fit inside a smartwatch. You have much more room to expand. So, what I like to say is thread based transistors are ideal for what I call tissue-embedded bioelectronics. They can literally be embedded into the living tissue. They're much more amenable for this bidirectional interface with the biological world. But they do not need to be that small. They only need to be at the level of the environment you're trying to sense.

Sam: The reason we were so intrigued by our conversation with Sameer is that it’s really about how varied the future of transistors might be. Smaller transistors are going to be amazing, and it’s gonna be so cool to see what future versions of our favorite electronics do. Like when I think about the computers I used when I was a kid, they seem so bulky and slow. So what is it going to be like for future generations to look back on our state of the art tablets and phones?

Deboki: But size is just one property of transistors, and engineering is really about adapting to different environments and circumstances and constraints. Which means there’s probably a lot of very interesting transistors out there still waiting to be invented. For Sameer, that future is in the melding of electronics with biology.

Sam to Sameer: Transistors have gotten so small, right? But do you think that there's something that we can't do with them right now that maybe in the future we will be able to do?

Sameer: I envision a world where electronics and biology are biointegrated at much more intimate levels, where computing is happening close to the cells and tissues, and we are communicating both ways. And for that to happen, the transitions need to be biocompatible made with biologically relevant materials that do not cause adverse reactions and at the same time do not foul the transistors themselves. So there is a long way to go if you think of biointegrated electronics. We are not there yet. We are getting there. Thread based transistors are just one pathway of getting there. But we need to look at materials and systems that are biocompatible that can be engineered and fashioned with living cells and tissues around and fill this world of tissue embedded by electronics in a much more meaningful way.

Sam: Shall we Tiny Show and Tell?

Deboki: Yeah.

Sam: Cool. I can go first this time.

Deboki: Yeah, go for it.

Sam: Awesome. So for those of you who do not know, bears hibernate in the winter, at least many types of species of bear hibernate in the winter. I feel like a lot of people know that. I feel like as a little kid you learn about hibernating species and the one that stands out to me is bears. So in humans, even temporary bouts of immobility can actually be deadly because they can lead to blood clots and stroke. So this research team was trying to understand how bears are able to hibernate for months, where they're very much immobile and don't develop thrombosis, which is blood clots in their veins.

So what they did was they collected blood samples from 13 hibernating bears in their dens during the winter, and then in the summer they collected blood from those same bears. To do this in the summer, they had to use tranquilizer darts from a distance, from a helicopter actually, because, well, you probably don't want to try to collect blood from a bear who's awake.

So then they tried to identify different proteins and other factors that might vary between the two time periods, so hibernation versus summer months. And they were able to whittle it down to proteins that are involved in clotting and one in particular called heat shock protein 47 or HSP47.

That protein is involved in recruiting an enzyme called thrombin that helps platelets in your blood stick together forming a blood clot, which is very important when you are injured. Less HSP47 would mean less likelihood of blood clotting. And it turned out that when the bears were hibernating, they on average produced 55 times fewer of these HSP47 proteins than when they were active, meaning their blood was less likely to clot when they were hibernating.
And then the researchers also looked in people as well as in mice, and they found something similar. Although in people it seems like reduced levels of HSP47 takes time. So if someone is immobile more temporarily, it seems like they're still at a higher risk for clotting.

I thought this was really cool because it provided some insight that researchers could work off of to potentially develop treatments that would target this HSP47 protein and help prevent clotting and stroke.

It also reminds me a bit of the episode we did last year on regeneration and how scientists are taking notes from organisms that have this incredible ability to regenerate arms and legs and heads, to try to figure out maybe how they do it and then apply that to humans, and ideally improve our own regenerative capacity. It's another example of how we are looking to the rest of the animal kingdom, trying to learn from different organisms, to see how we might use that information to improve human health.

Deboki: We narrowly avoided catastrophe today. Because I was debating talking about this article too. My dad had a really bad blood clot a while ago now and luckily came out of okay, but it was a very scary thing. And so when I saw this it was like, "This is fascinating." It makes sense to study hibernating bears. It's just such a fundamental… everything about hibernation. It's like, "How do they do it?" And also, can I do that? Can I just curl up and sleep for a few months?

Sam: I know.

Deboki: It's so interesting. My Tiny Show and Tell is also a thing that I think we both are very interested in because we've talked about this, not on the podcast, but we both really love science archives. I saw this article in the New York Times that I got super excited about, which is about a group of natural history museums that got together and basically put together an inventory of more than a billion objects that are in 73 museums across 28 countries, to basically see just like, what do we all have? It's really hard to know. And the thing is a lot of these collections are still getting digitized and that's a really long and involved process.

I thought this was really interesting. Instead of relying and waiting on these digitized collections to fully happen and come together, people asked curators to fill out a survey about what they have in their museum and there's actually a dashboard that you can see online about what they have and it's super cool you can see where these things were collected from. There's both the terrestrial locations, different continents, but also different oceans because that's also very important. I was just thinking about things you're collecting on land and it's like, "No, obviously we've gotten things from different oceans." And they have different stats for all of these things.

And so one of the reasons why this is super useful for us to have, even if we don't get to see the full collections, is it's really useful for seeing what the holes in our collections are. So one of the things they notice is we don't have a lot of collections from polar regions, which makes sense. But these are regions that are, especially now in terms of understanding how climate change is affecting things. The fact that we don't have a lot of collections from these areas means that we may not get to have the full picture compared to, or not full picture, but we won't get to compare them across time the way that we get to do with other places that we have a lot of collections from.

Insects are also apparently underrepresented, which I think makes sense in a way because there are so many of them that it's difficult to fully represent insects in your collection. But yeah, I just thought that was very interesting.

Sam: That's really interesting. And this is publicly accessible.

Deboki: The dashboard is, I'm not sure what other parts of the collections are. But we'll include it in the show notes for anyone who wants to see it. Because I thought it was pretty cool.

Sam:
I'm excited. I want to check this out. I'm definitely going to log onto that dashboard. Thanks for tuning in to this week's episode of Tiny Matters, a production of the American Chemical Society.

Deboki: This week’s script was written by me and was edited by Michael David and by Sam Jones who is also our exec producer. It was fact-checked by Michelle Boucher. The Tiny Matters theme and episode sound design are by Michael Simonelli and the Charts & Leisure team. Our artwork was created by Derek Bressler.

Sam: Thanks so much to Sameer Sonkusale for joining us. If you have thoughts, questions, ideas about future Tiny Matters episodes, send us an email at tinymatters@acs.org. And remember, we have coffee mugs now! Which is very exciting. So I’ve left a link to where you can buy them in the episode’s description. You can find me on social at samjscience. And if you get a mug, definitely tweet at me—I want to see it.

Deboki: I also want to see it. You can find me at okidoki_boki. See you next time.

 

LISTEN AND SUBSCRIBE