The future of electronic materials
We are on the cusp of a materials revolution – in electronics, health care, and avionics – says guest engineer-scientist Eric Pop.
For instance, silicon and copper have served electronics admirably for decades, he says, but at the nanoscale, better materials will be needed. Atomically thin two-dimensional semiconductors (like molybdenum disulfide) and topological semimetals (like niobium phosphide) are two candidates, but with AI tools to design new materials, the future is going to be really interesting, Pop tells host Russ Altman on this episode of Stanford Engineering’s The Future of Everything podcast.
Transcript
[00:00:00] Russ Altman: This is Stanford Engineering's The Future of Everything and I'm your host Russ Altman. I thought it would be nice to revisit the original intent of this show. In 2017 we wanted to create a forum to dive into and discuss the research my colleagues across the campus are doing in science, technology, medicine, and other areas. My goal was to show you that these are people working hard to improve the world. The university has a long history of doing work to impact the world and it's a joy to share with you the motives and the work of these colleagues as they try to create a better future for everybody. I hope you'll remember that when you think about universities and their role in society. I also hope you'll walk away from every episode with a deeper understanding of the work that's in progress here, and that you'll share with your friends and family as well.
[00:00:49] Eric Pop: You know, there's hope now that at least for sub tens of nanometers of, you know, conventional metals, of which copper is the best, there may be a class of other materials which could carry electricity a little bit better.
[00:01:10] Russ Altman: This is Stanford Engineering's The Future of Everything. I'm your host, Russ Altman. If you enjoy The Future of Everything, please remember to follow it in whatever app you're listening to right now. That'll guarantee that you never miss an episode and you're clued in on the future of everything. Today, Eric Pop will tell us that the materials that we build our electronics with, silicon and copper, may be on their way out as we introduce better and newer materials. It's the future of electronic materials. Before we get started, a reminder to make sure you're following the podcast in your favorite app, so you always get notified of the new episodes.
[00:01:54] You know, electronics are everywhere. They're in our phones, they're in our computers, but they're also in our refrigerators, our ovens. And they're made out of materials that we've been using for sixty years, copper and silicon, plus their friends, to make electronics. Well, you might say, why are electronics getting faster and faster over the last sixty years? It's because of miniaturization. But we're starting to reach limits in that miniaturization. And so there's a need for new materials that can perform the same functions as copper and silicon, but maybe do it in even smaller, tiny, little assemblies.
[00:02:29] Well, fortunately, Eric Pop is a professor of electrical engineering and materials science at Stanford University, and his lab is looking at these new materials. They're finding them, and they're characterizing them, and they're seeing that there is a rosy future for new materials that will help us get to smaller, more miniaturized, better management of heat, materials that will give us the electronics that we all crave for.
[00:02:59] Eric, what is the status of electronics these days and what do you see as the big technical challenges that wake you up every morning?
[00:03:07] Eric Pop: Well, Russ, those are, uh, those are big questions. Um, the status of electronics is, you know, on one hand, um, things have been very, you know, pretty, that's pretty much the same for about sixty years. Everything has been silicon, you know, made essentially out of sand, second most common element on earth. Why wouldn't you build electronics out of it if you could, which we can't.
[00:03:29] Russ Altman: Right.
[00:03:29] Eric Pop: Um, and a lot of this, by the way, started down the street from us on San Antonio Road, uh, with, um, William Shockley, Jim Gibbons, um, Gordon Moore, of course, and others. Um, and so we've been building electronics for about sixty years, um, using silicon, using aluminum at first for the wiring, then copper in the last twenty years or so. Um, recently in the last ten years or so, we introduced hafnium in as an insulator, hafnium oxide actually as an insulator before that used to be silicon oxide, which is a type of glass, um, as well. So that's on the materials side. We've been building these things sort of the same way. And we've also been building them in what's called the von Neumann architecture. Which basically means that, you know, one company makes the logic chip, the chip that actually does the computation. And it's almost like a different division or different company makes the memory chip, the chip that actually stores the data. So these things are physically been separated, you know, on the motherboard, or if you take a part of your laptop or your cell phone or whatever, uh, things have been physically separated for, um, about sixty years, mostly built out of silicon, copper wiring.
[00:04:44] Russ Altman: Let me ask one question about that as a non-expert. So silicon is not actually used for the wires. It's silicon, it's copper for the connections mostly and not silicon itself.
[00:04:55] Eric Pop: That's correct. Yeah, so silicon is a semiconductor, which means that we can build out of silicon, we can build a transistors which we can then switch on and off. So these are essentially one and zero switches, which are switched by a voltage. Which, by the way, coincidentally today, it's about one volt and zero volts. So it's about there.
[00:05:13] Russ Altman: Very nice.
[00:05:14] Eric Pop: In the past, it was a little bit higher, but for low power, sort of general computing, it's about a volt today. Um, and the wiring has been aluminum for many years.
[00:05:26] Russ Altman: Okay.
[00:05:27] Eric Pop: But aluminum gave way to copper about twenty, twenty-five years ago, I want to say, um, it's very hard to beat copper in terms of, you know, a very good metal.
[00:05:37] Russ Altman: Yeah.
[00:05:38] Eric Pop: For just carrying the signal between the transistors. And then of course, you know, if you go home and you take apart the walls, you'll see that all the wiring in the walls is also copper.
[00:05:46] Russ Altman: That I knew, yes.
[00:05:46] Eric Pop: Or kind of the same reason, right? It's a very good conductor, it carries massive amounts of current without getting too hot, you know? And it's not that expensive. Um, and so that's why it's used in chips as well. And so for sixty, well, twenty-five years for copper and sixty years for silicon, it's been pretty hard to think of an alternative partly due to the cost. But also partly because these things are good enough, you know, and all the technological advance, or most of the technological advance, have been about how to make these things smaller.
[00:06:25] Russ Altman: That's amazing, and that's clear. I didn't realize how stable the materials had been, and I was wondering, okay, then why is my current phone so much faster than my phone five years ago? And that's about miniaturization and not material changes?
[00:06:39] Eric Pop: Uh, you know, it's a little bit of both. You know, I don't want to necessarily get in trouble with the material science audience to say it's not materials, because you know, there have been materials added. For example, um, you know, about twenty years ago, one of my colleagues at Stanford, Krishna Saraswat, started researching germanium again. And I think at first people were a little bit dismissive, you know, germanium, we studied that in the 1950s. It's not that abundant. It's not that good. Silicon is good enough, why bother? And, you know, right now the stuff that's in your cell phone uses germanium in what are called the p-type transistors.
[00:07:18] Russ Altman: Okay.
[00:07:19] Eric Pop: Um, and they're not a hundred percent germanium. They're like, I think some fraction germanium, maybe twenty percent or something like that. But germanium is being used, has been introduced the last few years. Um, and so we have been replacing, you know, little pieces here and there. But if you look at the chip as a whole, certainly the n-type transistors are silicon, and the substrate is silicon, and there's a few other insulators thrown in. The metals are copper, pretty much.
[00:07:48] Russ Altman: Gotcha.
[00:07:49] Eric Pop: And so, to your question, why these things have gotten, you know, faster, um, the things have gotten faster since the 1960s because we've been able to cram more of them on a chip. And, you know, in this late 60s, when Gordon Moore first pointed out that we just need to cram more things on a chip, um, you know, since the late 60s, you know, late 60s, probably these things were about ten microns in size. So about the diameter of a human hair, maybe a little bit less, maybe the size of a red blood cell. And then gradually things, transistors, wiring, everything became sub one micron. So well below the size of a red blood cell. About fifteen, twenty years ago, these things became the size of a virus. So about a hundred nanometers. And today, these things are about five times less than a virus.
[00:08:44] Russ Altman: Okay.
[00:08:44] Eric Pop: The individual transistors. So now we can cram billions of these things on a chip the size of my, you know, my fingernail, my thumbnail, a few centimeters across, basically. Um, and in the past, we could only cram a few hundred thousand, a few million, now we can cram billions of these things. Um, but remember what I said, um, maybe three minutes ago, the chips that do the computation and the chips that do the memory, the data storage, are actually still usually separate.
[00:09:13] Russ Altman: Yes.
[00:09:14] Eric Pop: And so that's actually becoming a real bottleneck, because you have to ship data back and forth between the two chips. And it's turning out that we can compute with the data much, much faster if it's on the same chip. Because if you think these signals are propagating almost at the speed of light in the copper wiring between the chips, almost at the speed of light. But if you do the calculation, those delays can be measured in nanoseconds, essentially, or almost nanoseconds. And when a chip is operating at gigahertz, which is what these things today are operating at, delays on the order of even fractions of a nanosecond or tens to hundreds of picoseconds will quickly add up.
[00:09:53] Russ Altman: Gotcha.
[00:09:53] Eric Pop: And it will slow things down. So right now, we're partly bottlenecked by the separation of logic and memory, as we say in my field.
[00:10:02] Russ Altman: Yes.
[00:10:03] Eric Pop: Yes, we are also partly bottlenecked by the fact that we're kind of still using mostly silicon, germanium, copper. And these materials get a little bit harder to work with, and a little bit less, I suppose, optimal, um, when we make them below about the size of ten, twenty nanometers. I think ten nanometers is roughly the agreed upon dimension where things get weird.
[00:10:29] Russ Altman: So I'm, so in prep, in preparing for our conversation, I read about, I read a bunch of your, uh, your work and you talk about universal memory. And I'm now thinking that's what you mean by universal memory. Where the separation between the logic and the storage, or the two words that you use, where that is no longer a separation and they're co resident in the same kind of, uh, electrical environment.
[00:10:53] Eric Pop: Yeah, so that's part of it. That's part of it. A big part of the term universal memory refers to the fact that even if you look at the memory today, you know, we have three or four types of memory. You know, let's start with the most sort of basic ones, you know, you can think of a vinyl record as a type of memory. It's an analog type memory but it's a type of memory. It stores our you know, our Rolling Stones or Beatles records, right?
[00:11:16] Russ Altman: Right.
[00:11:16] Eric Pop: Then we have, you know, magnetic tape. Which obviously is also a type of memory and people still use it even for digital data storage, uh, for large amounts of data. Neither, you know, vinyl records, nor, sorry, nor magnetic tape are very fast to access, as you know. Um, so then we go to magnetic hard drives. Magnetic hard drives have seen a huge revolution in the last forty years. Um, and so, you know, most of the data storage in the cloud is still magnetic hard drives. You know, Seagate, Western Digital, these companies are cranking these things out. And that kind of looks like a vinyl record, except we're storing data in a magnetic medium on a spinning platter.
[00:11:58] Russ Altman: Yeah.
[00:11:58] Eric Pop: Um, and that's, you know, clearly stored, on a physical hard drive that sits, you know, somewhere here and the computers over there and, you know, things are separated. Um, now, so those are all kind of external to, let's say, to our cell phones, right? Here's my cell phone. Inside the cell phone, we also have three types of memory, at least three types of memory. Um, there's the flash memory, which has been slowly replacing hard drives. So we don't have a hard drive in here, we have flash memory. Now, what's special about hard drives and flash memory is that when you turn off the power to your cell phone, or even if you dunk your cell phone in water, uh, that memory is recoverable, it's not dead. So when you turn off the, you know, this is what's called non-volatile memory. So it's pretty well stored even when you remove the battery.
[00:12:47] Russ Altman: Gotcha.
[00:12:48] Eric Pop: Sometimes even when you dunk it in water, like I said, unless it's maybe salt water, things will go weird. Um, but we also have two types of memory in our cell phones and in our laptops. They're called SRAM and DRAM. And these are very, very fast memory types. But they are what's called volatile memory. And so when you turn off the power, those are gone. Basically gone, you know, they're not there the next time you reboot your phone. Um, and, um, you know, and I kind of think about, um, you know, I kind of think about these memories, like even the way the human brain works, right? Um, if you tell me what your phone number is, I'll probably remember it long enough to write it down. But if you ask me your phone number this afternoon, I will have forgotten. So that was stored in the very short-term memory, right? Um, but, you know, things like language, how to ride a bike, you know, all of these things are, like parts of my childhood, are stored in, um, you know, in certainly a long term, reasonably permanent memory, um, you know.
[00:13:53] Russ Altman: Yes.
[00:13:54] Eric Pop: And so, you know, computers and cell phones and all these things sort of work similarly. The trade-offs right now are that the volatile, fast memory, they're fast, right? Um, but they go away.
[00:14:07] Russ Altman: Right.
[00:14:08] Eric Pop: Uh, and the non-volatile stuff, which is called flash or the hard drives, they're slow and they're not that easy to access sometimes, at least not by computing. Um, and so what I meant by universal memory is kind of unifying the attributes of these things into one type of memory.
[00:14:25] Russ Altman: Okay.
[00:14:26] Eric Pop: And so you want something that's fast and non-volatile, so you don't have to, you know, you don't, it doesn't go away when you turn off the power.
[00:14:34] Russ Altman: Yep, yep.
[00:14:35] Eric Pop: But, you're absolutely right, I also want it to basically sit right on top of the CPU. And if it sits right on top of the CPU, the memory and the logic and everything will basically be, you know, a hundred nanometers apart as opposed to a centimeter apart or something like that. These are very, very big distances and signals need time to travel.
[00:14:57] Russ Altman: Okay, so you've been very generous in kind of getting me on board with this. And I want to make sure we get to the stuff that you're doing at the frontier. So what is the stuff you're doing at the frontier? I know it has to do with replacing copper, for example.
[00:15:11] Eric Pop: Yeah, so the way I, you know, the way I'd like to explain my group's research right now is we look at the entire stack of materials that go into building chips today. And so these materials go from the silicon all the way to the copper, to the memory, and even the materials that are used to spread the heat. Because as you know, you know, things get hot when they run. And we need to have a way to, you know, maybe get rid of the heat a little bit better or spread it out, something like that. So as a group, we're primarily electrical engineers, but we're also, you know, people sort of with applied physics bent and material science bent, and we are very concerned with what's next, what's the future of all these materials that go into the chip? Um, and so the bleeding edge, you know, as you said, for what we do is looking at what's next for all of these different components that go on the, go in the chip. From the metals, you know, is there something beyond copper to the silicon?
[00:16:14] Is there something beyond silicon? Or, you know, silicon here to stay. Because it is cheap enough. Uh, to the memory, right, is, you know, we have this weird, complicated memory hierarchy. Can we just collapse it all into one memory type and then build that on top of the silicon, let's say. Um, and then the heat, you know, what do we do about the heat? Right now, um, the heat is basically also spread using copper. Copper is a really good heat conductor. And so if you take apart your laptop or your desktop and even your cell phone, you'll probably see some copper heat spreaders in there.
[00:16:52] Because it's really hard to beat copper in a cost-effective way as a heat spread, not, you know, and electricity conductor, actually, those two are related for copper at least. And so the bleeding edge is all of these things. And over the last, you know, I've been running a group for almost eighteen years. Uh, and so for eighteen years, we've been looking at different components of this that depends on what new ideas we have. Very often we work with chemistry groups and pure material science groups who are looking for discovering new materials. And then as primarily electrical engineers, we ask, you know, Hey, is that manufacturable? Can we bring that into circuits at all? Or is this something that will work, you know, in a physics or chemistry lab, and it may only work at low temperature.
[00:17:41] Russ Altman: Right.
[00:17:41] Eric Pop: Because we do want these things to work at room temperature and a little bit above room temperature, because that's where, you know, all these commercial electronics have to work. Um, so if you want me to get more specific about?
[00:17:54] Russ Altman: Yeah, I mean, I think you, I looked at some of the stuff about replacing, um, uh, the copper with some wires, whose names and chemical names I didn't even recognize from the periodic table. Um, so maybe you could just tell us as an, as a vignette, how do you then begin replacing this very common, very universally used wiring material?
[00:18:17] Eric Pop: Yeah. Yeah. So copper, you know, if you just take the copper wiring in your home, um, we're probably not gonna replace that.
[00:18:24] Russ Altman: Okay.
[00:18:25] Eric Pop: That is really hard to beat. I mean, the only metal that's a better conductor than copper right now is probably silver, and that one is not cheaper. Uh, and it's not as easy to make into wires and it's not that easy to work with and yada, yada yada. So, um, but where we're trying to replace copper is, is in the semiconductor chips. Um, and there the big question is, copper gets weird if you thin it down or narrow it into wires that are less than about ten nanometers across. And it gets weird, not so much because, you know, of pure quantum mechanical reasons, although it's tempting to say, well, ten nanometers, you know, quantum effects, wave nature of electrons, that sort of stuff.
[00:19:06] The real reason actually is a little bit more, I guess, prosaic or classical, if you will. Um, the electrons that carry all the massive current in copper, they travel about tens, a few tens of nanometers before collision, between collisions. So they travel, let's say twenty, thirty nanometers, they collide, travel again, collide. It's basically how they carry current. If you make the copper wiring less than about ten nanometers across, even a little bit more than that, uh, electrons will begin to bump into the edges of the wire. So, you're essentially bringing the edges very, very, very close in, you know, to where electrons like to conduct current. And so now electrons are feeling a lot of scattering, really friction.
[00:19:54] Russ Altman: Right.
[00:19:54] Eric Pop: With surfaces. And so, you know, another way to think about it is, is as you take, you know, a big chunk of copper wire and you shrink it down to a ten by ten nanometer diameter cross section, um, you have a larger surface to volume ratio.
[00:20:08] Russ Altman: Yeah, no, I was just in the London Underground. And when there was a lot of people, I couldn't get, go forward, because I was, they were pushing me into the walls.
[00:20:17] Eric Pop: Right, exactly. Right. So that's kind of what's happening with the electrons as well. They're, you know, they're being basically squeezed more often against the walls of the copper wire. Um, and those walls are not, you know, atomically smooth. Now, somebody might say, well, you know, maybe there's a way to make those walls to be atomically smooth. And maybe if they are atomic is smooth, you know, maybe we will,
[00:20:40] Russ Altman: Just slide, slide along.
[00:20:42] Eric Pop: Things will, electrons will slide along. Um, there's another sort of more subtle problem is that copper these days, they don't just put the copper wire directly in. They have to actually put a little like, kind of like a lining around it. And that lining is made out of something called tantalum nitride, usually. And that's because copper likes to diffuse out a little bit. So they have to protect the copper inside the chips just the way the copper wiring in your walls is protected with plastic, you know, wiring. It's a little bit different so you don't get electrocuted. Um, inside the chips, there's glass around, there's glassy materials that prevent short circuits. But you still need a little bit of a, weird sort of liner. And so we started thinking about this, you know, can we do better than copper at sub ten nanometer sort of diameter dimensions or thickness dimensions? And we started looking at these materials, in this case called niobium phosphate.
[00:21:37] Russ Altman: That was the word.
[00:21:39] Eric Pop: Yeah, which is an alloy of basically fifty percent niobium, a reasonably common metal. And phosphorus, which also by itself is not a particularly uncommon, you know, material, right? It's in match heads, and so on. But niobium phosphide by itself is, was predicted, uh, to be a weird compound, uh, in the sense that the niobium phosphide surfaces were supposed to conduct electricity quite well. So niobium phosphide was predicted to have good electricity conducting surfaces. And so we figured if we can make this, and we can make this small enough, eventually, the surfaces will begin to dominate, right? And you can actually leverage that surface property on this material.
[00:22:23] Russ Altman: Right.
[00:22:23] Eric Pop: And you know, this goes back, I mean, there was a, these types of materials, physicists love to call them topological materials. And topological to me is just a fancy word for, you know, the surfaces behave differently from the block, right?
[00:22:37] Russ Altman: Okay.
[00:22:37] Eric Pop: I mean, there's some geometry arguments and some very elegant math behind this. Um, but what it boils down to from a practical point of view is, in our case, is you know, can the surfaces really carry sufficient current that even if the middle of the wire is just not quite as good as copper, we can still beat copper because the surfaces are better than the copper surfaces.
[00:22:59] Russ Altman: Gotcha. Gotcha.
[00:23:01] Eric Pop: Copper surfaces get worse when you make it thinner.
[00:23:03] Russ Altman: Right.
[00:23:04] Eric Pop: This material's surfaces are pretty good, and they begin to dominate.
[00:23:08] Russ Altman: So you might actually want to be at the edge of the subway walking area. I mean, I'm just trying to keep that analogy going. But it's actually, you move faster if you go to the edge, which is not what you would intuitively expect, but evidently, at this scale, the electrons do better at the edges.
[00:23:24] Eric Pop: That's right. Yeah, and it's not just moving faster. So this, in this recent paper that you referenced that we published a few weeks ago, we did some very careful measurements, and we also learned that the surfaces actually carry a lot of electrons.
[00:23:37] Russ Altman: Okay.
[00:23:37] Eric Pop: So it's not necessarily that they just move faster, it's just there's a ton of them. And so by virtue of just having lots of electrons that are allowed to be at the surface, and they're not particularly slow, they actually carry a good amount of current through the surfaces, which is, you know, which is a cool result. And copper doesn't do this. So, you know, there's hope now that at least for sub tens of nanometers of, you know, conventional metals, of which copper is the best, there may be a class of other materials which could carry electricity a little bit better and not care that it's, you know, ultra-thin, ultra small. You know, there's obviously other questions to be answered, which I'm happy to get into if we have time. But fundamentally, it's about the surfaces carrying electricity better than copper surfaces. And, and these wires being kind of dominated by the surfaces.
[00:24:36] Russ Altman: This is The Future of Everything with Russ Altman more with Eric Pop next.
[00:24:46] Welcome back to The Future of Everything. I'm Russ Altman and I'm speaking with Eric Pop from Stanford University. In the last segment, we learned about electronics, the basic architectures, the issues of heat. We heard about how copper is now, um, at risk of being replaced by some new conducting materials that might have advantages. So let's go to the other big player in electronics, which is silicon. It's reigned for sixty years. You told us about that earlier. Are there any pretenders to the silicon throne that are starting to get attention?
[00:25:25] Eric Pop: Yeah, absolutely. So, you know, like I said, silicon is, has been around for a while, partly because it's cheap, it's sand. It's the second most common element on earth after oxygen. Uh, so why not make electronics out of it? And, but people have been aware for sixty years that this is not, you know, necessarily the best semiconductor all around. Now, there are questions of, you know, what are we exactly using it for? So for low power electronics that go into our cell phones, laptops, and so on, even the stuff powering the data centers, it's all silica. For high power electronics, if you think about the stuff that's driving the electric cars, those things have to carry very large currents, very large voltages. Those things are not necessarily all silicon, and they will probably soon be replaced by things like silicon carbide, gallium nitride, and people are even talking about diamond. But that's in the more distant future. So diamond, you normally don't think of it as a semiconductor.
[00:26:23] Russ Altman: I do not.
[00:26:24] Eric Pop: It's actually a good semiconductor for high voltage applications.
[00:26:27] Russ Altman: Okay.
[00:26:28] Eric Pop: Now, I want to go back to silicon, which is, you know, I think in our daily lives, we probably interact with silicon more than with anything else. Um, and so, you know, ninety-nine percent of semiconductors probably are silicon. And I want to say, um, for low power, low voltage applications, which is ultimately what is driving, you know, this AI revolution, our cell phones, our laptops, all of this stuff. The contenders or pretenders, if you will, are going to be materials which are, by their very nature, nano. Um, and I mean, you know, one or two nanometers across. Not tens of nanometers, but literally a few atoms across. And there are two pretenders, or contenders, if you will. Um, one is carbon nanotubes, which are essentially these little tubes, they're just cylindrical tubes of carbon atoms, which are about one nanometer in diameter. Um, they've been around for about thirty years. Uh, some are metallic, some are semiconducting. If you could have all the semiconducting ones working the same way, it's been predicted they would be better than silicon and you could, if you could integrate them and whatnot because boy, you know, the carbon, the semiconducting nanotubes are very good semiconductors.
[00:27:42] Russ Altman: Okay.
[00:27:42] Eric Pop: And electrons love traveling in them and you can shut them off if you want. So, you know, they're a little bit hard to work with. And that's why, you know, they're still, they're still, you know, being researched, including here at Stanford. Uh, my group has worked on carbon nanotubes in the past, but in the last ten years or so, we got a little bit more excited about what are called two dimensional semiconductors. I really mean two dimensional, like 2D. So think of a material that is literally a sheet of atoms, or maybe a sheet of three atoms in thickness, something like that. And it is a semiconductor, meaning that you can shut the current on and off using a simple voltage nearby. And you can do this with voltages that are one volt or below, so the whole thing works at low power. Um, and so these two-dimensional semiconductors, uh, that we've been looking at for about ten years or so, um, one of, an example of them is molybdenum disulfide. Moly, moly, molybdenum disulfide.
[00:28:40] Russ Altman: Moly. Yep.
[00:28:41] Eric Pop: That's the, you know, that's the thing. Um, and in, it's basically a three-atom thick sheet of, I guess of atoms, moly in the middle, molybdenum in the middle, sulfur above and below.
[00:28:54] Russ Altman: Okay.
[00:28:54] Eric Pop: It's definitely not the only two-dimensional semiconductor, but it's the one that's, let's just say it's easier to work with. It's.
[00:29:02] Russ Altman: Let me just stop you there. In what sense is silicon not 2D? Is it, should I think of it as 1D?
[00:29:08] Eric Pop: No, think of it as 3D, so silicon,
[00:29:10] Russ Altman: 3D. Okay.
[00:29:10] Eric Pop: Silicon, like copper is essentially a 3D material. It's in three dimensions.
[00:29:15] Russ Altman: It's like a lattice of silicon atoms.
[00:29:17] Eric Pop: It's a lattice in all three dimensions. And just like I was mentioning copper earlier, if you take silicon and you make it thinner than a few nanometers, the properties of silicon begin to change. Not in a good way. So, uh, and the most sort of exactly by analogy to copper, uh, the electrons in silicon begin to bump into the surfaces.
[00:29:40] Russ Altman: Oh, okay.
[00:29:41] Eric Pop: The same way and they slow down. And, um, so if you want,
[00:29:47] Russ Altman: So there's a certain bulk in silicon that you have to buy into. And if you want to continue to miniaturize moving it from a 3D to a much less 3D, like a flatter, and that's why we're talking about our molybdenum, I'm going to call it a molybdenum sandwich. 'Cause you have sulfur on the top and the bottom. Okay.
[00:30:04] Eric Pop: That's exactly right. And even today, you know, if you look at the transistors inside our, uh, chips in our consumer devices, silicon, this individual silicon transistors has actually been etched down to vertical fins. They look like little shark fins, that they're actually called fins.
[00:30:23] Russ Altman: Yep.
[00:30:23] Eric Pop: That are about six nanometers across. And maybe about forty. fifty nanometers tall. Um, but people, I've worried that if you make these things any thinner, electrons will significantly scrape along the surface.
[00:30:38] Russ Altman: All right. All right.
[00:30:40] Eric Pop: And so there's this trade off with these materials, silicon and copper, by the way, where surfaces are beginning to dominate.
[00:30:47] Russ Altman: Right.
[00:30:47] Eric Pop: But they're beginning to, to, to.
[00:30:50] Russ Altman: No. And then I can see where the excitement would come when you're talking about layers, like three, four atoms, that, uh, that's much thinner you could fit in, and as long as they're well behaved and I'm sure you're working on all of that stability properties and everything, but as long as that is true, then we could talk about our fins, uh, can become much thinner, much shorter, and there can be a lot more of them.
[00:31:11] Eric Pop: Yeah. And so one thing I want to mention very quickly, even for silicon, people are actually moving away from fins. So if you buy a computer or a laptop in two years or a cell phone in two years, they're actually moving away from fins. They're moving into something called nanosheets. And nanosheets, they're actually taking the fins now, and they're, if you think of a fin like this, they're actually putting them like this, and they're making several sheets of them on top.
[00:31:34] Russ Altman: Oh, like stacks of fins? Stacks of fins.
[00:31:36] Eric Pop: But they're not, I wouldn't call them fins anymore because they're now horizontal.
[00:31:39] Russ Altman: Right. And they're still pretty thick is my guess.
[00:31:42] Eric Pop: Probably, you know, they're not thinner than four or five nanometers, which is incredibly thin if you think that they're making billions of these things, right? TSMC, Samsung, Intel have already made test chips with these things. I don't think they're commercial yet. But these things are coming in silica. So what we need to go beyond is to build these kinds of, you know, nano switches, if you will, out of these little layers, which are actually just a few atoms across right now. And that, you know, like you said, exactly, they have to be well behaved, um, and they have to be, you know, compatible with industry manufacturing. They have to switch at less than one volt, ideally. When they switch off, they really have to be off so that there's no leakage currents. Um, and that, so those are some of the big challenges with the two-dimensional semiconductor.
[00:32:34] Russ Altman: Gotcha. Well, listen, in the last minute I did want to ask you about AI. I have noticed that for many of my guests, AI is changing their field. And I just wanted to do a check-in with you about, is AI playing a role in this discovery that you've been talking about, or is it not a thing?
[00:32:50] Eric Pop: I think it's starting to, I think it's starting to, I think what's exciting to me is that, you know, I mentioned about niobium phosphate as a potential replacement for very thin copper, you know, to the materials as a potential replacement for very thin silicon. Notice, I'm not saying in general replacement, just for very particular, very thin and therefore very dense. Um, I think we're just starting to scratch the surface of these new materials. We are at the beginning of a materials revolution, not only in electronics, obviously in healthcare, avionics, um, you know, probably even buildings and, you know, much bigger things. Uh, but it's certainly in electronics because there's a huge need to replace these sort of conventional materials with things that behave better in certain dimensions and certain scenarios. And the examples I gave you are sort of discoveries that people made in a very random walk kind of way. Now, hopefully, we're hoping AI will, you know, suggest, Hey, did you try this material with these properties, right?
[00:33:55] Russ Altman: Gotcha.
[00:33:55] Eric Pop: Once it tells me that, I'm also hoping it'll tell me how to manufacture it, because manufacturing it is a whole different thing.
[00:34:03] Russ Altman: So this is a, uh, a nascent part of material science, uh, moving, uh, if you'll allow me to summarize, moving from an empirical test, uh, modality to a, maybe the AI can use what it's seen before to actually constructively propose new materials. And I love how you said, and how to manufacture them, which sounds to me like a little bit more tricky.
[00:34:25] Eric Pop: That's really important because, you know, there have been papers emerging in nature and science type journals saying, Hey, we predict this new materials. And those of us who make these materials, we look at it, we go, nobody knows how to make this stuff.
[00:34:38] Russ Altman: Yes.
[00:34:38] Eric Pop: We also need people who know how to make it, but also if there's algorithms and suggestions on, on how to make these things, it'll speed up significantly the discovery. And production, real life production of these things, because we ultimately, we have to make these things.
[00:34:56] Russ Altman: Thanks to Eric Pop. That was the future of electronic materials. Thank you for listening to The Future of Everything. Don't forget, we have more than 250 episodes in our bank of old episodes. And you can find a wide variety of discussions that can keep you entertained for hours. If you're enjoying the show, remember to share it with your friends, family, colleagues, anybody you like, so that they can learn about The Future of Everything as well. You can connect with me on many social media threads such as LinkedIn, also Bluesky, Mastodon and Threads, where I'm @RBAltman or @RussBAltman. You can follow Stanford Engineering on social media @StanfordSchoolOfEngineering or @StanfordENG.
