Full Transcript

This transcript was prepared by a transcription service. This version may not be in its final form and may be updated.

Alex Ossola: In the early days of computers, it took a whole room of machinery to perform even simple calculations. And then in 1961 came a huge tiny revolution.

Speaker 2: Here is a packaged integrated circuit. Inside this package is a chip of silicon which provides the electrical equivalent of many transistors, resistors, and dials, all interconnected to provide the desired function.

Alex Ossola: Silicon chips were a breakthrough. They were small and incredibly efficient. 60 years later, they've only gotten better. I mean, I have a supercomputer in my pocket right now. It unlocks by recognizing my face. It tracks my location. It can make a whole movie. And I use it to call my mom.

Speaker 3: Hi, Alex.

Alex Ossola: Love you. Bye. I had seen pictures of microchips, but I had never actually held one in my hand. Recently, though, I decided to change that. Ooh. My boyfriend let me crack into one of his old phones to take a look at what's inside. Are we going to know the chip when we see it?

Speaker 4: I know what a chip looks like.

Alex Ossola: You do? All right. The phone's processor looks like a small black square about the size of my fingernail. And like most chips in the devices we have, this one is still made out of silicon. But in the past few years, scientists have started hitting the physical limits of that material.

Deep Jariwala: We have reached a point that even though you can keep shrinking silicon, it has reached a point where it is no longer energy efficient.

Alex Ossola: Dr. Deep Jariwala is an engineering professor at the University of Pennsylvania.

Deep Jariwala: Even though silicon works at these extremely small dimensions, the energy efficiency required to do one calculation has been going up. And this makes it highly unsustainable. Energy-wise it doesn't make sense anymore.

Alex Ossola: Especially as the demands on chips grow. Forget about making a movie. Researchers want phone-sized devices that can discover new vaccines or search an animal's genome, or predict the path of a hurricane. That future will only be possible with chips that are even smaller, even faster, and even more efficient. To meet this demand, scientists are devising new ways to improve chips made out of silicon, and they're exploring new materials that could replace it. From The Wall Street Journal, this is The Future of Everything. I'm Alex Ossola. Today, we're exploring the future of silicon chips and the world beyond.

President Biden: Today is a day for builders. Today, America is delivering, delivering.

Alex Ossola: In August, President Biden signed a $280 billion bill to boost U.S. semiconductor manufacturing and research.

President Biden: Today, I'm signing into law the Chips and Science Act, a once-in-a-generation investment in America itself.

Alex Ossola: A big feature of the law is to create incentives for chip manufacturers to build facilities in the U.S. Mostly, they'll make silicon chips. Silicon is a semiconductor, meaning that it sometimes conducts electricity and sometimes doesn't depending on how much energy is applied to it. It wasn't the only contender to be the go-to for chips. It won out decades ago because it's easy to find, it's cheap, and it's reliable. But Dr. Stephen Forrest, an electrical engineering professor at the University of Michigan, points out that another reason silicon is so pervasive today is because it's been around for so long.

Stephen Forrest: So it turns out that silicon, although it's a really good electronic material, or it's a pretty good electronic material, it's fantastic because of what we've done with it, right? We keep perfecting it and perfecting it. And sometimes you perfect the thing that in particular aspects is not the best stuff. But when you put all of the capability that people have focused on since the 1960s, in totality it becomes the best stuff. I mean, there's nothing in, I think, human experience which has ever made anything so perfect, so repetitively, because we all own these things, and so reliably.

Alex Ossola: But there's a limit to what silicon can do. And we're already hitting it. Silicon chips have transistors, switches that turn on and off to perform logical functions. The first silicon chip had four transistors. Today, they regularly have billions, and engineers want even more. In an effort to boost processing power, they're packing so many transistors onto each chip that they're pushing up against the laws of physics.

Stephen Forrest: The size of the transistor is getting very close to an atomic spacing in silicon.

Alex Ossola: Atomic spacing, like the size of an atom.

Stephen Forrest: So for example, the current node is probably around seven nanometers. The atomic spacing in silicon is roughly .5 nanometers. We keep getting to smaller and smaller lengths. So eventually, you just run out of silicon.

Alex Ossola: That's a problem because those chips use a lot of energy. Keeping them cool becomes a challenge. Like when you're running a bunch of apps on your phone all at the same time, the phone can get too hot and just shut down. And more complex operations mean more transistors on the chip, which means even more heat. But we're not going to stop asking chips to do more. So one way engineers are trying to get more out of silicon chips is by spreading out the processing power over multiple chips and linking them together. Some are expanding them horizontally from the size of a dime or quarter to a playing card or a dinner plate. Some are going vertical. My colleague, Christopher Mims, writes a tech column for The Wall Street Journal.

Christopher Mims: So they have started doing more so-called three- dimensional chips, chip stacking, where you put one silicon chip on top of another, on top of another. More three-dimensional features, and that gets pretty complicated. But you look at them under a electron microscope and they're little bumps or fins or whatever, and that helps the gates and the transistors function better.

Alex Ossola: These chips function better because even though they take up more space, they're better connected.

Christopher Mims: So if you think about, it's really not so different than if you're laying out a city, right?

Alex Ossola: If a city has tons of sprawl with lots of suburbs and exurbs, bringing everybody into the downtown area is a recipe for disaster.

Christopher Mims: It's going to be snarls and it's going to be a nightmare.

Alex Ossola: Similarly, a chip spread out over a flat area is going to have a harder time sending information from one side to the other.

Christopher Mims: Or like most mega cities, you can go up. And when you go up, of course, the distance between where people live and work or where they want to shop is shorter because you're taking advantage of that third dimension. And in the exact same way, the more transistors you can pack into a more cubicle space as opposed to just spreading them out on a big flat chip, the faster they can communicate with one another.

Alex Ossola: But there's a risk to this kind of stacking. The parts can interfere with each other and make the whole thing less efficient. Engineering Professor Deep Jariwala at the University of Pennsylvania says these kinds of processors will need other materials around them to work properly.

Deep Jariwala: So I think this may be the first thing you'll see within a decade. It may be a new insulator on top of silicon. And then what will happen is that newer layers of semiconductor may get added on top of it on the time scale of maybe 15 or 20 years.

Alex Ossola: And the new materials won't just be around the silicon chips. They'll be in the chips themselves. And in some cases, they might replace silicon altogether. We'll have a look at the contenders after the break. My current phone weighs a little less than seven ounces. It can take photos of my cat while I'm streaming a TV show. But in the future, we're going to be asking phone-size devices to do even more. And since we're already hitting silicon's limits, engineers are looking elsewhere for the next big little thing. So what material could steal silicon's crown as the go-to for chips in all of our devices? There are at least a dozen possible candidates, but they kind of fall into three groups; materials with wide bandgaps, we'll talk about what that is, two-dimensional materials, and materials that can work in photonics, a chip based on light. Yeah, light. Let's get to know the candidates.

Speaker 9: First up, they can handle the heat, but can be a little hard around the edges. They include diamond, gallium nitrate, and silicon carbide. They're the wide bandgap materials.

Alex Ossola: So what's the wide bandgap? Remember earlier when I talked about how silicon is a semiconductor that can both conduct electricity and not conduct it depending on how much energy is applied? Well, for wide bandgap materials, the distance between conducting and not conducting is big. So when you think of a semiconductor like a door that opens and closes, when a wide bandgap material is closed, it's really closed. That makes these materials perfect for things that involve a lot of power or a lot of heat, like electric vehicles or power stations. A company called Wolfspeed is already making chips out of some of these materials, gallium nitride and silicon carbide. Silicon carbide sounds a lot like silicon, but it is not the same thing.

John Palmour: So it's 50/50 silicon and carbon. Silicon carbide is an extremely hard material.

Alex Ossola: That's Dr. John Palmour, Wolfspeed's chief technology officer.

John Palmour: Silicon carbide is uniquely suited for power management. So anytime where you need to pull energy out of a battery or out of the wall and convert it to make it useful for whatever the application is, that's where our silicon carbide power devices operate.

Alex Ossola: Palmour has been working on silicon carbide since 1983 when he was a grad student. Back then, he says, it was really hard to grow silicon carbide in the lab without too many defects. Figuring out how to do that was Wolfspeed's big innovation. But the science wasn't the only roadblock.

John Palmour: There was a couple of challenges, the biggest of which was cost because we were on relatively small diameter wafers. So they were quite a bit more expensive than silicon. And the general reaction was, "Yeah, come back and talk to us when it's the same price as silicon." But what we had to preach and work with the customers a lot is teaching them that even if it was more expensive at the component level, at the system level you could actually save money even though the silicon carbide was a lot more expensive component. And once you finally get those first couple of people, everybody else wakes up and starts to follow suit. But getting those first early adopters is the hard part.

Alex Ossola: And it paid off. Wolfspeed's silicon carbide chips, as well as those they make from gallium nitrate are now in thousands of types of devices, from electric vehicles and solar panels to power plants and cell phone towers. Now, those applications and more are really taking off. Last fiscal year, Wolfspeed says it's revenue increased by 40% over the year before. Wolfspeed is also building a new chip factory specifically for silicon carbide. When it's done in 2030, Wolfspeed will be able to produce 10 times the number of chips it makes today.

John Palmour: That is just going to be an enormous factory, but we feel very firmly that it's going to be needed with the rapid adoption that we're seeing in both the automotive space and in the industrial space.

Alex Ossola: Wolfspeed isn't the only company already using wide bandgap materials in chips. Companies like Belkin, Anchor, and even Apple have started using gallium nitride in their phone and laptop chargers to make them more compact and efficient. So wide bandgap materials are already starting to beat out silicon when it comes to high-power applications. But that's not the only area where silicon is falling short. Engineers are trying to create devices that we can wear comfortably on our clothes or even directly on our body, wearable tech. Doctors could use them to monitor patients after surgery. Parents could stick a sensor on their kids to use as a baby monitor. But my colleague Chris Mims says that's hard to do with the silicon chip.

Christopher Mims: What they are actually doing generally is making teeny tiny chips like seeds on the surface of a strawberry, and then connecting them with flexible circuits. But you could potentially create actually flexible circuits with graphene and other things like it.

Alex Ossola: That brings us to our next contestant.

Speaker 9: It may be thin, but it still packs a punch. It puts the two in 2D. It's two-dimensional materials.

Alex Ossola: The materials in this group have one big thing in common. They're just one to two atoms thick.

Deji Akinwande: So two-dimensional materials. The best way to think about them is to think of a sheet of paper. So they're sheets of atom equivalent to a sheet of paper. Just like a sheet of paper, it just has length and width, it doesn't really have depth.

Alex Ossola: That's Dr. Deji Akinwande. He studies two-dimensional materials at the University of Texas at Austin. One of these two-dimensional materials is graphene. It's made up of a single layer of carbon atoms that's just barely visible to the naked eye. You could also roll that material into tubes. They call that carbon nanotubes. There's also black phosphorus, a single layer of phosphorus atoms which under normal conditions come together in kind of a puckered structure. As you can imagine, these atom-thin two- dimensional materials can be hard to work with. But they have clear advantages over silicon. They're extremely conductive and they interact well with light.

Deji Akinwande: Our cameras now can provide visible imaging only when there's light. So that's based on silicon technology. Now, putting graphene there as the camera element, you can also do night vision. And this is something that is just not possible with silicon because it doesn't have those fundamental properties, while graphene has the possibility to do both night vision and daytime imaging all with the same material.

Alex Ossola: And going back to Chris Mims, two-dimensional materials can be used alongside silicon to finally make chips that are more flexible and bendable.

Christopher Mims: That's a big deal because if we want to start putting more sensors and circuits on our bodies, we're out in the natural world, that flexibility, literally and metaphorically, is a big advantage.

Alex Ossola: But Akinwande says 2D materials need a bit more work before they're ready for prime time.

Deji Akinwande: The challenge is being able to make these devices very high yield and being able to make them reliable for long term. So they need to be able to resist environmental effects like oxygen, light, normal daily wear and tear. And so they just need a little bit more development to get to that kind of level of maturity that is needed for real product technology.

Alex Ossola: You know, we've been talking a lot about speed, and silicon chips, even though built with superconducting 2D materials have a limit to how fast they can go because they're sending information with electrons. But you know what's a lot faster than electrons? Light.

Speaker 9: And our final contestant shines bright at the smallest scales. The third group of materials helps chips send information with light.

Alex Ossola: The transistors on today's silicon chips turn on and off with electrons. But electrons actually don't move that fast. Light on the other hand is the fastest thing in the universe. So getting that to the scale of a chip would definitely speed it up, by 300 times some scientists estimate. They've been exploring compounds like lithium niobate and barium titanate to make this happen. Deep Jariwala, who researches nanoelectronics at Penn, says that could be a big deal.

Deep Jariwala: So not only will you have semiconductors and electronic materials at various layers, some doing memory, some doing computing, this, that, but then you could also have photonic layers, so one's doing comms or one's doing sensing and routing and things like that.

Alex Ossola: But so far at least, light-based chips haven't been easy to engineer.

Deep Jariwala: Turns out it's incredibly hard to do it at a chip level because shrinking light is much, much, much, much harder than shrinking electrons.

Alex Ossola: So which of these contenders will take the title of go-to material from silicon? Lots of them sound really great, and each has some advantages over silicon, but they also have limitations. Right now, it doesn't seem like any one material is coming out on top. The future of chips is likely a bunch of different materials depending on the purpose. And everyone I talk to agrees that our old standby silicon isn't going anywhere.

Deji Akinwande: So from a sustainability point of view, it makes sense to continue to use silicon to some extent.

Deep Jariwala: I think silicon will remain as the base. It's because the entire semiconductor industry is based on it.

Stephen Forrest: That will not be the end of silicon by any stretch of the imagination. Long after we're both gone, silicon will still be here.

Christopher Mims: I think we are stuck with silicon for the next 1,000 years. It's too useful, right?

Alex Ossola: The Future of Everything is a production of The Wall Street Journal. Stephanie Ilgenfritz is the editorial director of The Future of Everything. This episode was reported and produced by me, Alex Ossola. Our fact- checker is Aparna Nathan. Jessica Fenton is our sound designer. Scott Saloway is our supervising producer. And Kateri Jochum is The Wall Street Journal's executive producer of audio. Thanks for listening.