Majorana qubits with Chetan Nayak
Welcome back to the New Quantum Era.
Sebastian Hassinger:I'm your host, Sebastian Hassinger. This is our first episode in 2026, so happy New Year. My guest today is Chetan Nayak, technical fellow at Microsoft and one of the central figures behind the company's efforts towards topological quantum computing. Chetan's career threads together some of the the deepest ideas in condensed matter physics, topological phases of matter, non abelian anyons, and Majorana zero modes with the very practical question of how to build a scalable fault tolerant quantum computer. I met with Chetan initially at APS, the APS summit back in March, and we've been talking about doing this episode for a while, so I'm I'm really excited about this one.
Sebastian Hassinger:We talk about you'll hear in the interview, we talk about how he went from studying the fractional quantum hall effect and exotic quasiparticles initially with his thesis adviser, Frank Wilczek at Princeton, and then eventually, UCLA and then at Microsoft, where he cofounded Station Q, which is their quantum research center at UC Santa Barbara, and, started working on topological quantum computing. And that that whole topic really went from a thought experiment, and and a bunch of theoretical speculation on a whiteboard to eventually the announcement last February of, finding pretty compelling proof of actually experimentally realizing myurana zero modes on a device of their design and and manufacture and also, the launch of the Majorana one chip, which they're working on, in their r and d labs at this point. So we we get into the original proposal for the topological qubit and gallium arsenide quantum hall devices and then their pivots to semiconductor, superconductor, nanowire wires, and and now gate based devices and how those ideas sort of crystallize into this tetron architecture, which is what they're working on now in the the Majorana one chip architecture. Chetan also explains why topological protection is just part of the story.
Sebastian Hassinger:They're gonna be implementing surface code style error correction on top of the the the Tetrons. And so we get into that a bit. And then we talk about some of the advantages of topological qubits in particular, the sort of more binary style control rather than, microwave based or RF based control, and the advantages to scaling that that will provide once they get to that stage. We talk about Microsoft's roadmap, how they think about their material choices, where their focus is right now, where they think they're gonna need to go in the coming years, and why they they believe that their their microsecond scale tetron sort of sits in a a Goldilocks zone for integrating an entire utility scale quantum computer into a single dilution refrigerator. Along the way, we touch on some of the key papers that shaped this field from early theory of non abelian anyons to recent device level blueprints and Tetron experiments.
Sebastian Hassinger:And we close with when and how outside researchers might start to get hands on time with their topological qubits. So I hope you enjoy.
Sebastian Hassinger:Alright. Thanks for joining me, Chetan. I'm really excited about this conversation. I saw your talk at the APS Summit last March, when you were sort of summarizing the work that you've been doing with Microsoft on the Majorana one topological qubit chip. This is a a modality of qubit that we have not dug into on the on the podcast, so I'm super excited to get into it.
Sebastian Hassinger:But if you could start by telling me a little bit about sort of your journey to how you got to this this point, that would be great.
Chetan Nayak:First off, thank you very much. It's a pleasure to be here on on your podcast and to have a on the New Quantum Era to have a chance to talk to you, Sebastian. Know that we've we've run into each other a couple of times and said it would be fun to do this. And so it's nice to to to have the opportunity.
Sebastian Hassinger:Yeah. I appreciate you making it happen.
Chetan Nayak:Yeah. Thank you. So yeah. So I I'm not sure exactly where to begin, but we can go really far back. But I can say I first heard about quantum computing when I was in graduate school, that probably a good place to start.
Chetan Nayak:And Peter Shor came and gave some it was actually shortly after his second big paper. So people understand that Peter Shor revolutionized and rebooted the field of quantum computing with the discovery of his algorithm. Quantum computing goes back to the work of Feynman in the early 80s and actually parallel work in The Soviet Union by Manin. But Feynman was very much motivated by actually something that very much excites me to this day and is exciting for the whole field, which is that if you want to simulate the real world, if you want to simulate nature, materials, chemistry, so on, you actually can't really very effectively do it on classical computer because the real world
Sebastian Hassinger:nature isn't classical, damn it.
Chetan Nayak:Exactly. Know, it's fine. It's colorful way of putting it. And but what Shor did was he found an application, a potential application for a quantum computer, which is factoring numbers that is something that is exciting even to non scientists, right? And so that, of course, generated an enormous amount of excitement.
Chetan Nayak:But very shortly thereafter, he wrote his second revolutionary paper in the field, which is showing that error correction is possible. Because people thought right after Shor's algorithm, there was this feeling like, well, isn't a quantum computer actually kind of like an analog computer and some power is coming from the fact that it's analog and we know that analog isn't going to work, it's going to have problems. So it was shortly after that second paper, the error correction paper, that he came and gave a series of lectures at the Institute for Advanced. I was a graduate student at Princeton. My PhD advisor Frank Wilczek was on the faculty at the Institute for Advanced Study at the time, which is the institute across town in Princeton, formerly independent of Princeton University, but it's where Einstein spent the latter part of his career journal, you know
Sebastian Hassinger:And and Von Neumann built the the system sort of contemporaneous with ENIAC. It's one of the very first operating computers. And and the Von Neumann architecture obviously still is, in our laptops and iPhones.
Chetan Nayak:Yep, yep, exactly. So Shor came and gave these lectures there and that was the first that I heard really started learning about quantum computing. And like completely coincidentally, like unrelated to completely unrelated to this, I was right around that time first learning about topological phases and non abelian topological phases of matter and starting to understand that. And part of my PhD thesis, the work I did with Frank Wilczek was about actually understanding Majorana's zero modes and their topological properties. What happens when you braid, when you exchange and measure them and uncovered this kind of rich mathematical structure, which in modern language, we would say that as you take Meijer on a zero mode, if you take topological defects, these particle like excitations in a emergent state of matter that as you start bringing these things around, you start doing essentially the Clifford gates.
Chetan Nayak:We didn't have that language at the time. Right. But because we weren't thinking in terms of quantum computing, but we were actually working with exactly the kind of unitary transformations that people look at in quantum computing. And after listening to Sean's lectures, I thought this seems like a real exciting area. I need to work on this, but I didn't connect it to what I was working on at the time.
Chetan Nayak:Am
Sebastian Hassinger:I right in in I I think I was reading that you were working on like quantum hall effects is where
Sebastian Hassinger:you were doing the That's right.
Sebastian Hassinger:The right. Okay.
Chetan Nayak:Correct. So there's a there you know, the quantum hall effect is this effect that occurs when you take a two dimensional electron gas. So basically you make a semiconductor structure where you have electrons that move in flat land, you make a quantum well, so the electrons move in two dimensions, put a big perpendicular magnetic field and cool it down to very low temperatures. And what happens is there are these emergent states of matter and these quantum Hall states. I'll come back to this, around Majorana one, we talked about discovering a new state of matter, this topological superconductor or I should say, the real way to say it is discover and actually engineer because we really engineered the state of matter.
Chetan Nayak:But there are topological states of matter that have been known about since the discovery in the early 80s by Dan Tsui, Horst Stormer and the theory due to Laughlin that under these very extreme conditions of low temperatures and high magnetic fields in two dimensional, very clean two dimensional electron gases, there are these self organized emergent quantum states of matter that have these interesting topological properties. One of these states all these states are characterized by these fractions, one third, two fifth. Well, there was one state five halves that was kind of mysterious and its properties, we still don't fully understand, to be honest with you, although our simulations and theory give us a lot of insight into it, I would say not fully nailed down experimentally. But what we did is we took the kind of what was not really quite the leading model yet at the time, but would later become the leading model and understood its underlying structure. So it was in the context of the quantum Hall effect that we first started figuring out basically the topological properties of myron and zero modes and which has a very rich structure.
Chetan Nayak:And then as I said, I was working on this at the time and then there was I was trying to learn about this other thing, not really seeing the connection between them and then this paper of Alexey Kitaev.
Sebastian Hassinger:Right. I was gonna bring up That
Chetan Nayak:would combine these two things that I was, you know, trying to work on at the same time. And, you know, that, of course, turned, you know, turned the light bulb on, in my head. And and
Sebastian Hassinger:And and Kitaev was just he was I think it was the toric code was the first paper. Right? So he was just talking in mathematical terms about top topologies. So he hadn't made the connect he wasn't bringing up Majoranas or Majorana zero modes at at that stage. Yeah.
Chetan Nayak:He's probably thinking about that a few years later. So he was using a very abstract mathematical sense. But since I was working on this thing in the quantum hall effect, I realized that, yes. I mean, he was
Sebastian Hassinger:There it is.
Chetan Nayak:Specific case. And talking about the specific case, but if you generalize it, you know, we're talking about very similar things. And And then I got a phone call out of the blue from Michael Freedman, who was at Microsoft at the time. And he basically comes to the same conclusion. He's like, well, I've been reading these papers of yours.
Chetan Nayak:And I'm thinking about topological quantum computing. And I think we're talking about the same thing. And he came down and visited me. I was a professor at UCLA in the physics department at the time. He came down and visited me.
Chetan Nayak:And we spent a lot of time talking. But he noticed on my desk, there was a picture of Mont Blanc, which is a mountain on the border between France and Italy, which I climbed a few years earlier. And Mike is also into mountaineering. And he said, you know what, you should come to Microsoft and let's build a quantum computer and let's climb Mount Rainier.
Sebastian Hassinger:I've heard a number of stories of Michael Freedman's mountaineering So
Chetan Nayak:that was the initial hook. And so I went to Microsoft Research first as a visiting researcher. He was in Microsoft Research at the time. So we didn't there wasn't our quantum program yet. It was just one of the things that was happening.
Chetan Nayak:It was early 2000s. One of the it was just sort of one of the areas within theory of computation, within theory group, within Microsoft Research. And you know, we climbed Mount So we climbed Mount Rainier a couple of times.
Sebastian Hassinger:That's the easy part. Turns out.
Chetan Nayak:We did the easy part a couple of times over the next few years. And but we also kick started this journey that we've been on now for twenty plus years. And I can say, what was remarkable is just how much those ideas have over time gotten richer and more connected to the real physical world and to materials and into devices and to systems. And so that's why we're talking about, Majorana 1 and quantum computing systems now from those early days where we were really thinking of very basic conceptual ideas. Like it was the kernels of ideas about topology and states of matter and quantum computation and complexity.
Chetan Nayak:And over time, we've come to this juncture where we're now thinking about materials, right? Materials and
Sebastian Hassinger:Well, then in fact, that wasn't that sort of the biggest initial hurdle was was largely a material search for for this, you know, the the the nanowire, the structure that'll actually create those those Majorana zero modes. Right?
Chetan Nayak:Yeah. Well, yes. So that's skipping ahead in the story a little bit. Right?
Sebastian Hassinger:Okay. Sorry.
Chetan Nayak:That became that did become. But in the in the very earliest days, it was okay. We we we realized, okay. There's an idea here, but which physical system? Like, yes, you want to talk about phase, know, it could tie with that about Tourette code.
Chetan Nayak:There was, you know, the Majorana zero modes, which we threw out in the quantum Hall effect. But what's the material system and how you do it? It all sounds great on paper. You know, you take you it's like, oh, great. You take any arms and you break them and that's your computation.
Chetan Nayak:And, you can draw lots of pictures, but, you know, how do you actually make something like this?
Sebastian Hassinger:Well, in a way, it almost seems like it's it you were your challenge was reverse of of the superconducting qubits that we're we're used to, which are based around a a Josephson junction because the the junction preexisted quantum information science, and it was there and then could be exploited. Right? And you you had to figure out what that that technology was, the material, and the and the design was that would actually create this this set of dynamics. Right?
Chetan Nayak:Exactly. Or to take another example, like, with trapped ions.
Sebastian Hassinger:Yeah.
Chetan Nayak:People who have been doing experiments with trapped ions, the ground stayed exciting to the first excited state.
Sebastian Hassinger:Keeping time with them. I mean, Dave Wineland and Chris Monroe both say, basically, you know, they when they read the paper from from Cirac and and Zoller, they were like, oh, well, I think we could do that. And then they just sort of implemented it, right, because they had the device already. Yep.
Chetan Nayak:Yep. Exactly. So in our situation, you know, it you're right. It's the reverse. And part of the reason we took the reverse strategy was at a fairly early point in time.
Chetan Nayak:This was I will say this was something that Craig Mundie, who was senior executive at Microsoft at the time, who ended up being a sponsor for the Quantum program as he developed the Quantum program really insisted, you need to think about the full system and you need to design around, let's say, a workload or a final goal, right? You guys have a great idea, but you need to think about like sort of soup to nuts and be the expression And you would so, from a pretty early time, we thought, okay, look, we want to do something like Shor's algorithm. And now we, of course, have a lot of resource estimations that we can put make We these numbers very didn't have that level of precision in the early days, but we had some understanding of, look, we want to do computations with thousands of logical qubits, which probably means hundreds of thousands of physical qubits, which this tells us something about the overhead and what kind of resources we need and what kind of control architecture. And I'll come to all that. So we had some understanding that look, you're right.
Chetan Nayak:You could take the approach of, well, we've got these systems already and now we can just build these things with them. But that may not get you to the scale that you need for a really useful quantum computer or we can work backwards from, well, here's this is what we need, all the things we need to be at scale. If we work backwards, what does the underlying physical system need to be? And so that was a little bit more the approach we took which has a very different kind of ramp up profile. So in that context, we thought like, okay, so now we need to make a device.
Chetan Nayak:We were talking about all this very abstract stuff, but we need to make an actual concrete device. And the first place we did that was actually in the quantum Hall effector. So we thought about these gallium arsenide heterosuction gallium arsenide devices. Gallium arsenide is a semiconductor that has some industrial applications. It is where the quantum Hall effect was discovered.
Chetan Nayak:And nowadays it's also observed in graphene, for instance. That's where it was discovered and that's where these interesting states that are hypothesized to have Meier on a zero modes exist. And so we said, okay, well, let's make an actual device where we say this is the device layout and if you do this with the voltages and do this measurement, then you're doing this is how you do a computation. And that's a paper that Mike Friedman, Shankar Das Sharma and I wrote. I think, yeah, published in 2005.
Chetan Nayak:So we wrote it in late two thousand and four. And these came out of conversations where we're talking to people and they were saying, okay, look, the idea sounds great but what are you actually going to do? Like how are going to make this thing? And we wrote that paper and that attracted a lot of interest from various experimental groups who are making devices that kind of in that same general category, but they call quantum dots and tunnel junctions and sort of the basic building blocks of semiconductor devices, nanoscale semiconductor devices. So with that, as we started to get all this interest, that's really where we decided, okay, now it's time.
Chetan Nayak:We're no longer just a small, a few people doing quantum computing within Microsoft Research Theory Group. We're actually it's actually now a program. It's actually a project, it's quantum. We need to actually start a quantum program. That was twenty years ago, that was 2005 when we launched it.
Chetan Nayak:It was only about six of us. We founded Station Q here in Santa Barbara And it was all people doing theory of quantum computation and working with academic groups. We funded a bunch of academic groups that essentially we're designing devices and proposing experiments. And then the academic groups that were funded by us were trying to make these devices and do these experiments. And that was quite fruitful.
Chetan Nayak:We learned a lot in those early days. Got to and then we but we also within the same idea, we made a lot of detours and not detours, not the word, but we made a lot of pivots. So we started out doing quantum Hall effect. Then the nanowire idea came out that also in Catatios early work, you had some basic aspects of the idea that nanowires also could Supernan nanowires could also support minor on a zero mode.
Sebastian Hassinger:And and the the nanowire, it's it's it's a like a grown it gets actually, you know, you're laying down layers of and you're you're combining like two a layer of semiconducting material with a layer of superconducting material. Is that right?
Chetan Nayak:Yeah. That's right. So an important point is what is a nanowire? It's a good question. What what is a nanowire?
Chetan Nayak:It's it's basically a wire or channel within a semiconductor device. It could be going as a wire, it could be a channel that you've patterned within a semiconductor device through gates or through hatching either way. And it's a channel which is much longer than it is, than its lateral dimensions. And its lateral dimensions typically are in the 10 to a 100 ish nanometers. That's what nanowires and its length can be hundreds of nanometers or microns in the case of all the devices we make, it's three or more microns long.
Chetan Nayak:And so it's several microns long, tens to hundreds of nanometers in its lateral dimensions. And importantly, the lateral dimensions, the quantization of the energy levels and the lateral dimensions are such that you have only a few transverse modes in this wire or that you can tune the device into a density and temperature regime in which there really are only a few modes so that it effectively quantum mechanically is like a one dimensional wire. There's really no or very little motion in the transverse directions. That's what a nanowire is. And you can either what are called VLS nanowires, which are grown, a vertical wire that's grown from a seed and then you break it off and you place it onto a substrate and make a device.
Chetan Nayak:Or you can grow like starting with a quantum well as we do in the quantum hall effect, starting with flat land with a quantum well. And then some people sometimes etch down, so they just remove parts of the wafer and then you're left with something that looks like a net that is quasi one dimensional. Or what we do, which is we make our wires by putting gates on top. So just put some metallic. Well, we first put down dielectric and then put down some pattern metal, which are called gates.
Chetan Nayak:And by applying voltages, you can basically, you can deplete all the electron density in the two dimensional electron density in the two day or in the quantum well except where you want it to be in the channel and similarly make a a nanowire. Okay. So as of the motion detector turns off the lights in my office by I don't don't move enough. So
Sebastian Hassinger:You have to get more more gesticulation then.
Chetan Nayak:Yeah. Exactly. So
Sebastian Hassinger:So so that nanowire then the the on the ends of that is where you're actually measuring the the Majorana zero modes. Right?
Chetan Nayak:Right. So so when when this is superconducting, so when this so and the and the way to make it superconducting is to is to, you know, coat it with a superconductor and it inherits the superconductivity of the superconductor. So as I said, there was this work of Kitayev suggesting that superconducting nanowires. There was a beautiful work of Fu and Kain saying, well, look, you can use a proximity effect to induce superconductivity. They were thinking about actually surface states on something called a topological insulator.
Chetan Nayak:And then there was work from the Maryland group, Das Sharma, Lutchyn, Sao, Tewari that where they say, well, actually we can actually do that in a semiconductor with strong spin orbit coupling and then further work built on that, the nanowire idea and then there was an explosion of ideas around how to do this with nanowires. And the thing that makes that such a great idea is implementation of this topological concept is two things. Number one, there's a I mean, this the double edged sword, I suppose, but you have a lot of choices of semiconductor and a lot of choices of superconductor, right?
Sebastian Hassinger:So The search space is large.
Chetan Nayak:Yeah. The search space is large, which means you can really tailor it to have the properties you want. So with the quantum Hall effect, one of the difficult things it was at that time, it was the early days of graphene, but at that time, one material, very hard to grow. And if you want to change some parameter a little bit to get sort of a qubit that was optimal in one way or another, there wasn't a lot of freedom, right? There weren't a lot of knobs to change.
Chetan Nayak:And as you think of designing a device with millions of qubits, there are things you're going to want to optimize and right? You care about the size of your qubit, you care about the speed at which you're doing it, you care about you're ultimately going to want to do rapid measurements and rapid operations. And so you want to be able to read out fast and so then the RF problem
Sebastian Hassinger:You want to optimize everything in the cube.
Chetan Nayak:Right? There's so many things you want to optimize. And so the good thing is number of choices of many choices semiconductor, major is a superconductor. And there was a and the ability to really, you would think to like design the material you want. Now, the flip side I'll say is, are a lot of choices, semiconductor, lot of choice superconductor.
Chetan Nayak:So it's a large search space. And so that can really lead you down the rabbit hole. And the idea of putting together a different superderm, different semiconductor gives you a lot of freedom. And it means that you can make the semiconductor have the properties that you really want, which is large orbit coupling, very high mobility. So, make it very clean, very tunable density.
Chetan Nayak:So, those are things that semiconductors can be really good for without having to worry that much about superconductivity because you're to get the superconducting, you're going to inherit it from some superconductor, right? And you can make the superconductor optimize those properties. But then there's an integration challenge, is does that something semiconductor is optimal and that superconductor that you think is optimal separately. Do they play well together? Can you actually grow them?
Chetan Nayak:Is the interface going to be good? Because you could get the best of both worlds, get the best possible semiconductor, best possible semiconductor, great. But you could also get the worst of both worlds, right? You can get all the bad properties of the supergraduate, all the bad properties of semiconductor and then you're not helpful. And the more different materials, the more layers you have, the more interfaces.
Chetan Nayak:And the more interfaces is the more places where the interface can be rough, for instance, or there can be interdiffusion between different species of the interface. So interfaces do let you get the best of both worlds, but they also bring challenges. In any case, was a pivot And we pivoted towards this class of materials and class of device designs around it. And in that context, thought about, okay, this, based on this nanowire idea, what is architecture actually going look like for a large scale quantum computer? And in 2016, I guess it was published in 2017, a large group of us wrote a paper kind of laying out, okay, this is we had a couple of choices.
Chetan Nayak:This is a set of like two or three different qubit designs and the architecture we build around that, the types of operations, a little bit about the kinds of error correcting codes we put on top of it. And that was sort of in 2005, when we launched Station Q, it was based on this first kind of practical topological qubit idea, which was taking a lot of abstract ideas and making a concrete device, which people have attempted to make and there's some interesting partial results on that from among others, Bob Willett, who's at Dell Labs. But the device in the 2017 paper was sort of taking a lot of these nanowire ideas and putting them together into a qubit and qubit architecture. Right?
Sebastian Hassinger:Right. And and actual like like Craig Mundie's point, it's it's at least it's a systems architecture rather than just a physics experiment.
Chetan Nayak:Exactly. That's exactly exactly the right way to think about it. And and and so because we then have to also start thinking about like, how are we gonna measure and manipulate these qubits and make sure that we do it fast because we wanna be able to run, deep circuits. So a lot of thought went into that. And where we've been is really on a journey in some sense of making that more and more real.
Chetan Nayak:Was still as concrete as it was. There were still a lot of assumptions and idealizations there and we were We making that more and more grew substantially right after that starting around 2017. That's when we started building up labs. And instead of working with various academic groups that we're funding, we brought a lot of work in house so that we could have tight collaboration and coordination between the design of the devices, the conceptual ideas, the design of devices, the growth, etcetera. Mean, that
Sebastian Hassinger:that sponsored academic research, it's an enormously powerful tool, but it it does tend to I mean, you know, once you have done the high risk, high reward sort of phase and you've gotten your results, the it's a natural sort of next stage to pull in the most promising, you know, the the the capabilities you need to plumb the most promising results out of that phase. So that makes total sense.
Chetan Nayak:Exactly. Exactly. I mean, the way that I describe it is I say, what academic research is extremely good at, good for and good at is exploring a lot of face space because in many cases, you have the freedom to explore a lot of ideas, some wild, some safe, some not so safe, right? And really kind of like do a random walk through a huge parameter space. And in an academic lab, might have five or six different projects and they all are usually have relative they don't overlap that much, right?
Chetan Nayak:By design because different people are trying to like the students are trying to graduate, get a postdoc, trying to go off with a professor and they need to actually differentiate themselves, right? And have distinct projects so that they can, of course, are somewhat constrained by their own same lab and the same capabilities, but they're trying to differentiate themselves and they Totally. And you're trying to encourage them to branch out in different directions. And so you explore a lot of business. But if you have a roadmap, once you're at the point at which you can formulate a roadmap with a set of milestones on the roadmap and things that you know you have to do in a certain order in a certain way, Well, that exploration is actually not the optimal way to go down the road map.
Chetan Nayak:We want to be much more much more disciplined and much more tightly coordinated.
Sebastian Hassinger:Right. And You need to bring in some engineering discipline. Exactly.
Chetan Nayak:Exactly. And so once we were at the point at which we could start to trace out and then make more concrete what the roadmap is, and of course, of course, it made sense. We started bringing a lot more things in house. Now the reality is even once you have a roadmap, of course, you learn things along the way. You adjust the roadmap as you learn more things.
Chetan Nayak:You make pivots as necessary. And in the very early days, we were thinking about these VLS and nanowires. And we realized was it was never going to carry us very far down our roadmap anyway.
Sebastian Hassinger:Very challenging fabrication task. Yeah.
Chetan Nayak:Exactly. And so we know that it wasn't that there was not a long term roadmap based on that. And so there actually wasn't really that much benefit in the short term. So we made a pivot and said, look, we're going to go into these gate defined nanowires based And on two different electron we see a roadmap heading out upon that. We made that roadmap more and more concrete.
Chetan Nayak:And we organize ourselves very much more into, like you say, more of an engineering organization. Obviously, we still learn new things and we're at the cutting edge of science. So we're still making scientific discoveries and reacting to what we learn. But at the same time, we are trying to make deterministic progress along a specific path towards a quantum computer. And so I think what's unique about quantum computing is that it is very much at cutting edge of science.
Chetan Nayak:Discoveries are being made of really kind of fundamental things. And at the same time, we're building a technology. You know, the reality is, as people were first making steam engines, they hadn't discovered entropy yet, right? So the often dynamics was formulated as steam engines were being used and built. So as engines were as internal combustion engine was being developed.
Chetan Nayak:So there it's not the unique time in science that that you're developing the technology That's
Sebastian Hassinger:good point.
Chetan Nayak:About the science.
Sebastian Hassinger:I mean, the the the, whatever, Rutherford's electron beam, to, you know, proving that the that the electron exists and measuring its its its mass is also the basis of the CRT and television. Right? I mean, like, similar kind of fundamental. So so that's super interesting, Chetan. I I mean, there's so much I wanna follow-up on, but but couple things sort of stuck in my mind.
Sebastian Hassinger:One is so those Majorana zero modes at the ends of the wires, those are quasi particles. Those are are, you know, sort of emergent behaviors out of that system that you've engineered, much like a Josephson junction creates sort of that that artificial atom they talk about in in, Transmon qubits. Does that going back to to the origins of your interest in the non abelian statistics and quantum Hall effect, do those quasi particles sort of bolster the case that Majoranas are a real particle? Or is that sort of, orthogonal to the the whether Majoranas are real or or not?
Chetan Nayak:Yeah. So interesting question. In some sense, the word quasi in front of particle is
Sebastian Hassinger:Does a lot of work.
Chetan Nayak:It's a more it's a matter of doing a lot of work. It's also a matter of taste because the particles that we think of as particles like the electron, they like the electron has a mass and a charge in this particular vacuum state of the universe. It has electric weak symmetry breaking and electrons would be massless in another phase of the universe in early universe before electroweak symmetry breaking. So the particles we have are emergent properties of the vacuum state that we live in, which is not the unique vacuum state. It may not may not be the same vacuum everywhere in the universe and and Right.
Chetan Nayak:Certainly not in the maybe the very earliest stage of the universe. So so I think okay. That's a little bit of a philosophical point.
Sebastian Hassinger:It's a good one.
Chetan Nayak:But, if I just restrict myself to solid state, physics, then in solids, there are states of matter in which you have excitations that have a particle like nature that we call quasi particles. And an example is if you make a solid, it has the vibrations in a solid. They're called phonons, right? They are particle like, in some sense, they're a little bit like photons, which are the particles associated with light. But they are called phonons because they're like the particles associated with sound waves, right?
Chetan Nayak:The vibrational waves in a solid. If you melt the solid, you go to liquid, they go away, right? They property are of the fact that the solid has this lattice structure and you have these lattice vibrations and the liquid doesn't have those. And so as a particle like excitation at zero temperature, right? So it doesn't have those.
Chetan Nayak:And so that's just a property of states of matter is that the particular excitations that you have, the ways in which you can excite above that ground state depends on what the state is, right? And the quantum Hall effect or topological superconductor has excitations that are very different from a metal or an insulator, right? In an insulator, like a semiconductor is a type of insulator, right? Just has a small band gap. And the excitations in semiconductor are basically like an electron.
Chetan Nayak:They look very much like an electron or a hole, right? Because if you take the absence of an electron in a semiconductor when you're sitting in the band gap and the valence band is filled and the conduction band is empty, you create a hole by plucking an electron out of the valence band. And that looks like in some ways, it looks a little bit like a positron, right? Because there's positive charge. But we call it a hole.
Chetan Nayak:It's an excitation of semiconductor, but it has positive charge. And the interesting thing is in a semiconductor like the seminars that we deal with, those holes can actually be can have very different properties from what an electron looks like in vacuum, right? And in fact, the electrons in the conduction band can look very different electron to vacuum by virtue of the fact that they're in a solid that their magnetic moment or their G factor, which tells you how strongly they interact, the spins feel and magnetic field can be extremely different. Their spin orbit coupling can be extremely different. Their effective mass can be extremely different.
Chetan Nayak:So the properties of a quasi particle of an electron like excitation in a semiconductor are very different from what they look like in a vacuum. And so that's kind of the simplest example where even though they have the charge of electron, right? And they're fermions. So it's a fermion, it has spin half, it has charge e, but its G factor is very different, its spin orbit is very different, its effective mass is very different.
Sebastian Hassinger:That's interesting.
Chetan Nayak:Many of these properties are quite different. Do I call People tend to call these electrons and holes, but they are quasi particles of some kind. Right? They're actually in that system.
Sebastian Hassinger:It's really interesting. It's almost like you were saying, I mean, that you're you're you're bringing engineering into the the pursuit of the fundamentals of science, and you're engineering these devices to to bring out the attributes, these these statistical nature of these different behaviors that that are exploitable from a computation or a technological perspective, but are also exhibiting sort of these fundamental behaviors of the the universe that we've been discovering. So it's
Chetan Nayak:That's right.
Sebastian Hassinger:That's interesting. I hadn't really thought about how fuzzy the line is between, like, you know, fundamental particle and and quasi particle behaviors. But but
Chetan Nayak:the And a lot of and a lot of our tech sorry. Just to finish the thought, like, a lot of our technologies that we have depend on
Sebastian Hassinger:Of course. Yeah.
Chetan Nayak:And so on. Right? The the these properties. Now in semiconductor, those properties, like I said, the charge is still E and the spin is still a half and it's still a fermion, but lots of other parameters are different.
Sebastian Hassinger:Right.
Chetan Nayak:Only part is we're talking about have a more dramatic difference, right? Which is like a Myron or Fermion is Myron is zero mode. Now you're starting to talk about particles that don't have a charge. And if you talk about a defect that supports a Myron is zero mode, you're not talking about something that's neither a fermion nor a boson. So now we're talking about more radical departures,
Sebastian Hassinger:are Totally weird.
Chetan Nayak:Which are even weirder and which are possible precisely because of restricted dimension that you've got quasi two dimensional where they live in two dimensional plane. Or like I said, at the ends of a 1D wire that you can have these particle like excitations called them quasi particles if you like. And those objects really look very different from an electron or a photon or any of the excitations that we have in our vacuum, in the vacuum of empty free space. And so just connecting it back to the point you made, and like what this does for us from a computational perspective is, this particular one is, if we have a nanowire, okay, with myron zero modes at the end, what that means is that now it's superconducting. First of all, there's a superconducting nanowire and superconducting is very famous to have this property that the electrons form pairs called Cooper pair.
Chetan Nayak:And those pairs condense into a superfluid correlated state where they all kind of condense into the ground state. And as a result, when you take mesoscopic superheroving devices, there's actually a big difference in even an odd number of electrons because even number of electrons, they've all formed pairs and they can all condense to the ground state and everything is cool, right? And if you have an odd number, there's an energy penalty because there's some electron that didn't pair up and they all want to pair. So if something if one electron is left frustrated, then there's energy penalty associated with it, it's a measurable energy difference. This is something that's observable and has been observed.
Chetan Nayak:Now, the cool thing about topological supernonductors is actually they can actually accommodate an electron without any concern, right? Because of the Meijer on zero mode, those are places where so everywhere else in the middle of the wire, a pair can either condense into the ground state, but an unpaired electron can't because it's it's unpaired. And condensate is made up of pairs, but the ends of the wires here, myron zero modes, you can actually absorb that unpaired electron and then it just gets mixed into the condensate. And so or emit an unpaired electron and send it off. So as a result of these Myron zero modes at the ends of the wire, the wire, a topological superheater can accommodate an even or odd number of electrons perfectly well.
Chetan Nayak:It's just as happy. And so as a result, and you can't tell, like this condensate looks the same. No local measurement can tell you the difference. So you look at the condensate and it just looks the same. And so therefore now a topological nanowire can be either even or odd number of electrons.
Chetan Nayak:Right. So that's a zero or one. And it's a zero or one that's very well protected because nothing locally can tell you whether it's a zero or one.
Sebastian Hassinger:Right. I was gonna say that's you just described the the the fault tolerance right there with
Sebastian Hassinger:the the ability to accommodate. Okay. So Interesting.
Chetan Nayak:So that's kind of the the the basic idea and and what we did in in that twenty seventeen papers, we figured, okay, well, how how are we gonna make qubits out of it? Turns out for for technical reasons, it's actually really nice to make a qubit out of two of these wires rather And than I can explain the technical reason, which is two wires, both wires could be odd number of electrons or both could be an even number of electrons. That would be our two states we wouldn't deal with like one being odd or one being even. And the nice thing about that is if your two basis states are odd, odd or even even, then the entire thing can have a fixed number of electrons, right? You don't have to worry about like exchanging electrons with a reservoir or anything like that.
Chetan Nayak:Right. You just have you got a fixed number of electrons. It could be like 10,000,000, let's say. And, you know, you have your two wires connected together so that you can't isolate either one from the other. And the total number of electrons is, you know, whatever it is.
Chetan Nayak:But one wire can be on the other odd or one can be even and the other even. And that is that basic h like structure with
Sebastian Hassinger:Right. Tetron.
Chetan Nayak:In a connection between the tetron is what I presented at the APS meeting and we have a paper on the archive about it from the summer showing that we can do the Z and X measurements on it. And so pushing that that idea to its logical conclusions in terms of making building or designing a qubit architecture, quantum computing architecture around it is what we started the process of project. Started in 2017, a lot of basic ideas were there. And a lot of the challenge has been, okay, now we got to make devices that have sufficient quality to enable us to do that. Because if you remember, I said, once you start making complex devices with many different material types, you always run the risk that you're gonna get the worst of both worlds rather than the best of both worlds, right?
Chetan Nayak:And so, this kind of situation, you have to then make a lot of choices about because a lot of decisions have to be made to make progress. Even once you have the basic architecture, you have to decide, well, okay, what semiconductor we're going to use? Because there's a lot of choices. I mean, indium arsenide is the basis of what we're using now. But there's indium antimonide, there's mercury telluride, there's lead telluride.
Chetan Nayak:There's a lot of different potential of semiconductor materials, each of which has like nice properties. And if you point to one or the other, can sort of, if you focus on one metric, different one is going to win according to any metric. There's There's a lot of different superconductors you could choose. So there was a complex decision making process there where I said, hey, look, we need to take all these factors together, really like have a rigorous decision making process around taking the different material properties, the technology readiness level, the availability of process techniques, the compatibility of materials. So there are many different criteria and many different material combinations.
Chetan Nayak:And we landed where we landed because of the combination of the highest score in essence, right, along these many different metrics. And but we also understood that that was part of that score was like, can we do today as opposed to what can we do a year from now, two years from now, five years from now. So the roadmap includes changes in material type because we knew, okay, look, we want to make a device now, we should do this combination. But, in two years, we want this other combination to be ready for us to be able to make devices with that because we know that we need these extra properties as we get to more complex devices. And so our roadmap actually has in sense, it's moving on two axes.
Chetan Nayak:One axis is just device complexity, more qubits and more features and more functionality. And the other is underlying properties of the qubit, which is things like how large is the topological dApp, which remember the electrons form pairs, it's the energy cost of breaking a pair. What is the signal to noise ratio of our ability to measure out that qubit, which depends on materials properties and system properties. So a number of different sort of quality type metrics. So on one axis, complexity metrics, you know, or device functionality metrics, the other, you know, and a road map takes us in a certain way through this two dimensional space.
Chetan Nayak:Mhmm. Ultimately, we're trying to get to, you know, utility scale, but the exact route depends on, you know, choices that you make. So so that
Sebastian Hassinger:And and the the the intrinsic hardware design, the qubit design has a degree of error protection in it. Right. But you're intending to still implement some sort of of topological code in the in the way that you actually program and interact with the measurement based sort of computation. Do do you have a sense at this stage, you know, how many of these tetron qubits you'll need to have in a device to be able to implement error correction at the logical level and and also like what the kind of ratio is gonna be for your physical qubits to logical qubit kind of implementation?
Chetan Nayak:Yes. Excellent question. So the the answer is yes. So we have published papers on exactly this topic, which is looking at different codes that we can use. And one of the things that's really great about this particular architecture is if you think about error correcting codes like the surface code or there's class of codes called Floquet Hot Codes.
Chetan Nayak:Floquet Codes developed by Matt Eastings and John Wan at Microsoft, part of our team, which are also based on topological states. What either of these codes do is at the end, they can be formulated as what you're really doing in a quantum computer is you're doing lots and lots of two qubit measurements. Right. You're measuring XX, ZZ, YY, you're doing lots and lots of two qubit measurements. You do logical gates through like lattice surgery or something like that.
Chetan Nayak:It's set effectively just by you skip some of those measurements. Every now and then you skip some of those measurements and that's kind of how you do a logical gate. And then you do also you prepare offline magic states and you inject them in, right? So it's a lot of two qubit measurements, a lot of magic state preparation, distillation and then injection, magic states. So really aside and I'll come back to the magic state preparation, but by and large what you're doing is two qubit measurements.
Chetan Nayak:And what's cool about the Tetron architecture is that two qubits are a native operation. And this is something this is a work that's very much in progress right now. But the way we measure our qubits is we come back, if you recall, our qubit is it an even or odd number of electrons in a topological nanowire? And the way we measure that is you can't locally measure it, but we can't if you can connect the two ends, which we do through a quantum dot, then actually an electron that tunnels around the loop, every time an electron goes past another electron, it picks up a minus sign, right? That's Fermi statistics.
Chetan Nayak:And so when electron goes around the loop, it counts whether you have an even or odd number of minus signs. And so we can tell you, do I have an even or odd number of electrons by going around that loop? That's a Z measurement on our qubit. Right. And the X measurement is, it's a loop that bridges the two wires.
Chetan Nayak:So it goes around half of one wire along a trivial backbone and then the other half the other wire. And it's something that doesn't commute with the So the order matters when you do Z measurement first or X measurement first. The two qubit measurement is actually a very similar kind of measurement. And the way you measure this is So this even around is getting printed on the quantum dot. And the quantum dot what you do is you measure effectively its capacitance.
Chetan Nayak:The capacitance shifts, we measure the capacitance between a gate and a quantum dot, which we do by embedding it in an LC circuit and we measure the change in the RF response to that circuit. So you make a small shift in the capacitance that changes, for instance, the resonant frequency of the circuit. And we can measure that in a dispersive measurement. So sit on like kind of the shoulder of that resonance. To keep it measurement, actually like if you have two of these Tetrons coming in like this, there are quantum dots that you can activate that let you do a loop that connects the two qubits.
Chetan Nayak:It kind of goes around half one qubit and then half the other qubit either XX on neighboring or ZZ on neighboring this way. And which are the native measurements you need for these kinds of codes. And again, that turns out to leave an imprint on a quantum dot, you measure it dispersively from a capacitance shift. So two qubit measurement is a native measurement for us. It doesn't involve in many cases, it's not usually a native measurement.
Chetan Nayak:Usually what you have to do is you have the measurement you have as you measure in the computational basis. If you want to measure in another basis, like you want to measure the X basis, you do a Hadamard gate. Right. And if you want to measure two qubit, you want to measure XX, well, you run some circuit so that when you measure Z on an ancilla, it's the same as measuring XX on your two qubits. But we don't have to do any of that.
Chetan Nayak:It's a And native so the native measurements for us, so that means that there's actually a lot of efficiency in how we implement these codes because the native measurements are the measurements you need for our correction. So having said that, long term, we've done resource estimation. And again, it depends on exactly what error rates, physical error rates were hit. It depends on the exact precision with which we're able to do these even on this measurements, right? Right now, what we demonstrated was something like on the Z measurement was like 95% assignment fidelity or 1% assignment error rate.
Chetan Nayak:Actually a little bit better than 99.5 in the archive paper. So we ultimately would want like three, four nines of fidelity measurements, which depends on a lot of things, amplifiers and resonator Q factors and very important engineering of the system. And so a lot depends on the exact numbers, but at a high level, we're targeting millions of qubits.
Sebastian Hassinger:Yeah.
Chetan Nayak:We're targeting millions of qubits and a ratio of physical to logical qubits that's in the 100 to hundreds. So that's what we're talking about in the hundreds. And that's that we want to have thousands of logical qubits. So depending on the application, we've worked out, we've estimated the resources that are needed for that application, including the thing that I didn't discuss in much detail, which is the magic states, that you need magic state factories and distillation and injection. So long story short is, we have a pretty concrete target and a roadmap of what we're doing.
Chetan Nayak:It definitely involves quantum error correcting codes such as the surface code or Hastings hot code. And I think that the, you know, the things that I would say that are different, like because lots of people you know, lots of approaches do involve things like the service. Like, so what are the differences? So I think the differences are that the measurements you need to do are native for us. We just do those natively.
Chetan Nayak:Secondly, those measurements, remember I said that this comes about from quantum dots that are coupled to our qubits. So the basic operation you do is you just turn on and off the coupling of the quantum dots.
Sebastian Hassinger:Yeah.
Chetan Nayak:So it's very digital. You know, if you remember
Sebastian Hassinger:That's nice.
Chetan Nayak:So we're back to the, like, the original Peter Shor, there was this feeling like, hey, this almost sounds like computing.
Sebastian Hassinger:Well and and with microwave control on on most superconducting devices, it still looks pretty analog.
Chetan Nayak:It looks pretty analog. Right?
Sebastian Hassinger:So the Digital control would be great.
Chetan Nayak:Yeah. So this is digital control because what we're doing is we're doing a set of on off, instructions, turn on or off this coupling of the quantum dot to the qubit. One of the things that enables, it enables to put a lot of that control down in the cold. So we've developed cryogen, so CMOS chips that operate at cryogenic temperatures so that instead of trying to send a lot of instructions from room temperature down into the to control a million cubits. Because one of the things that isn't always appreciated about a quantum computer is unlike a classical computer where you could have like billions of transistors with thousands of control lines.
Chetan Nayak:If have a million qubits, it's a million or order of a million control line every qubit I you can know. Right? So that's one of the real difficulties here. And because our control is on off switches, we can generate that control down in the cold, right? Relatively is to say we can design chips, classical chips that are sort of stripped down that don't generate a lot of heat.
Chetan Nayak:The problem ordinary of trying to put a chip in the cold is for a, a, it might not run well in the cold. Right. It needs to run over temperature. But b, if you try to run-in the cold, it's gonna just generate a lot of heat and you're gonna cook your your cubits. Right?
Chetan Nayak:That's right. But for the type of control that we need, which is all these on off switches, that is something that you can actually generate down the coal. And so so I think that's, you know, the the other key difference I I would say.
Sebastian Hassinger:That's interesting. Yeah. Yeah. I hadn't thought of that. I mean, that really does give you a potential advantage for physical scalability that the not having to stuff a wire down there for every single cube is a huge advantage.
Sebastian Hassinger:So so, okay. So
Chetan Nayak:And sorry, I know we're out of time. So if I could hit on two other things, which is
Sebastian Hassinger:Please.
Chetan Nayak:Which is the scale of the qubits for the materials we've chosen and the parameter range. The scale of qubits is like our wires are currently we're around 3.5 microns, but we're not going to go much longer than five to 10 micron long wires. So it's on the order of a few to 10 microns long. Right now in this direction, if you include all the quantum dots and the ancillary structures, it's again, it's under 10 microns. So these qubits are, they're pretty small, but they're not extremely small.
Chetan Nayak:It's actually Goldilocks zone because at that scale of 10 micron by 10 micron, you can put millions of these onto a small chip, onto a few centimeter by few centimeter chip. And so that means that at least as far as the footprint is concerned, you don't need multiple wafers, you don't need multiple chips, they don't even need to be multiple frigates, right? So because we don't need a massive number of control lines coming into the fridge from room temperature and because we don't need a massive amount of real estate to put millions of cubits, we a very can large quantum computer into a single fridge with a classical controller near it. So I think the size, there is a Goldilocks regime where we are in size. Some qubits are extremely small.
Chetan Nayak:And so it's very hard to get all the control lines in. Some qubits are very big, so it will have to be a multimodal system and we're neither of those two cases. And the other thing is the operation time, you know, we're targeting is, you know, basic operations measurement is operation time on the microsecond scale. Right. Microsecond is neither too slow nor too fast because if you're much slower than microsecond then things like Shor's algorithm or quantum chemistry is going to take years to run.
Chetan Nayak:And much faster is nice from some perspectives, but it's actually also a little difficult because in error correction, have to do a lot of real time feedback. You have to process all those data and then make some decisions. And if you had to go much faster, that also is a little tricky. So it's actually in kind of a sweet spot on size, speed, achievable what we think are the achievable error rates and then of course the digital control. So I think there are lot of aspects of what we're doing.
Chetan Nayak:Since you asked by error correction, which every kind of quantum computer has some kind of error correction. I think topological quantum computers, Microsoft's quantum computer, it's not going be something where everything is done in hardware and there's no error correction. There will definitely be convention where we might call it conventional error correction, but the implementation is unique and the other things that are come out of the topological character actually impact the practical engineering, the number of control lines, the footprint, the speed and so on.
Sebastian Hassinger:Fantastic. Well, this has been super interesting, Chad. And I the last question I wanna ask you is, do you have and this may be an unfair question. Do you have any sense of when people are gonna be able to get their hands on a device like this to start, actually, you know, experiencing the uniqueness of this platform for themselves?
Chetan Nayak:Yeah, that's a great question. The answer is gonna be a little complicated. As I said, we've been very much focusing on doing things in house because you have the control. We are looking at some collaborations now with external groups, academic groups where they could either try to run some small algorithm or experiments on our devices or maybe we even send some devices to people and they have a chance to test them. We're working with DARPA.
Chetan Nayak:So we're in the final stage of the DARPA US 2QC program. There's like 30 something companies in this program, only two Us and one of the companies have gotten to the final stage, phase C of this program. So we're actually working very closely with the US government who is getting their hands on our devices and systems. I'd say, I think we're at that stage of working closely with the US government, starting towards some academic groups. In terms of when it would be sort of larger group and where you could just pull out a credit card and go on to Azure and start using it.
Chetan Nayak:That's going to be a little bit further down the road. But some of these kinds of interactions are already happening, starting to happen.
Sebastian Hassinger:That's fantastic. Well, that's great. I really appreciate your time. It's been super, super interesting. Obviously, totally unique modality with lots of interesting engineering attributes and challenges and also fundamental science, which is really fascinating.
Sebastian Hassinger:So I really appreciate your time, Chetan.
Chetan Nayak:Awesome. Thanks a lot, Sebastian. Thank you. It's a pleasure talking to Thank
Sebastian Hassinger:you for listening to another episode of the podcast, a production of the New Quantum Era hosted by me, Sebastian Hassinger, with theme music by OCH. You can find past episodes on www.newquantumera.com or on blue sky at newquantumera.com. If you enjoy the podcast, please subscribe and tell your quantum curious friends to give it a listen.
Creators and Guests
