Building qubits from carbon nanotubes instead of silicon

Summary

Pierre Desjardins, co-founder and CEO of C12 Quantum Electronics, a company using carbon nanotubes to create a new quantum architecture, is interviewed by Yuval Boger. Pierre and Yuval talk about C12’s unique technology, what carbon nanotubes are, designing custom application-specific quantum chips, and much more.

Read the full transcript on the Quantum Computing Report

Transcript

Yuval Boger: Hello, Pierre, and thanks for joining me today.

Pierre Desjardins: Hello, Yuval.

Yuval: So who are you and what do you do?

Pierre: I'm Pierre. I'm the CEO and co-founder of C12, and C12 is a Paris-based quantum hardware startup. We are leading the next materials leap in quantum computing.

Let me start with what happened in the early days of classical computing to better explain what we are building. In the early days, digital computers were made with vacuum tubes, and with vacuum tubes you could not make today's smartphone at all. What was missing is the silicon transistor.

At C12, what we have is a quantum transistor, and it's made of carbon nanotubes. We think that it's the way to go to build a large-scale, high-performing quantum computer.

Yuval: And C12, the name is because that's the carbon isotope, C12?

Pierre: That's indeed right. Our approach is really to have the cleanest and purest material for encapsulating our qubits.

This vision comes from the fact that, in the end, the quantum errors the whole industry is fighting against come from defects in the qubit material. So the cleaner and purer your material is, the easier it is to have very good fidelity for your quantum computation, meaning very good quality for your qubits.

If you take spins, you indeed have a lot of errors that come from other spins that are close to your electron. Those other spins come from nuclear spins, and if you take the isotope 12 of carbon, then the number is even, and so you don't have a nuclear spin. What we are working with are isotopes that are purified to 99.99%, so it's extremely high, and in the end, what we calculated is that we have less than one nuclear spin on average per qubit.

So this is what we mean by a very clean material.

Yuval: When you say the number is even, do you mean that there are six protons and six electrons, or is it something else?

Pierre: No, when I say 12, it's the atomic number for this isotope, so it's the same number in the nucleus.

Yuval: Excellent. Carbon nanotubes, they sound small. How small are the nanotubes and how do you make them?

Pierre: They are indeed extremely small, and I think it's a big advantage when you scale a technology like we do. We can make them as long as we want, but usually we use a few microns for the really active sites.

In diameter, they are between one and three nanometers, so it's extremely small. What is important is that it's only a single layer of carbon atoms rolled over itself. So it means that, in the end, it's only a few hundred carbon atoms where we trap our spin qubit.

What is actually very exciting about this technology is not only having this incredibly pure material, but also our ability to integrate the carbon nanotube on a silicon chip, the same kind of chip as the one you use in your smartphone or your laptop. We start with those very standard CMOS chips, and we are able to make them quantum by integrating this incredibly pure quantum element, the carbon nanotube.

Yuval: And the qubit inside the nanotube, what is that atom made of? What is that atom inside the nanotube?

Pierre: We use an electron, not an atom, and actually we use the spin of an electron. It's the most natural system to make a qubit, because it's naturally a two-level system, and it's also naturally well isolated from its environment.

The quantum computer that we are building really relies on the first principles laid out by Loss and DiVincenzo when they described the theory of a quantum computer.

Yuval: Now, quantum computers do the computation by having the qubits interact. So you have individual spin tubes, individual carbon tubes, and inside them you have an electron. How do you get the individual qubits to interact with each other?

Pierre: Yes, indeed, because an isolated qubit by itself does not make a quantum computer.

What we have is what we call a quantum bus. It's like in the classical world: it's a communication bus that links all of our qubits. It relies on something that was demonstrated in the lab we come from, at the École Normale Supérieure in Paris, and it's called spin-photon coupling.

It's the way you can actually couple a spin hosted in a carbon nanotube with the photon in this quantum information bus, or quantum communication bus. And this is where it becomes very interesting. Unlike other solid-state qubit technologies, we are not limited to having each qubit connected only to its closest neighbor.

We have much more flexibility in our design, and we can really design qubit architectures, layouts where you have more connectivity than just connectivity to your nearest neighbor, thanks to this very important quantum communication bus.

Yuval: Can you give me some numbers?

What do you have in the lab, or what do you have working in terms of the status of the product to begin with, and how many qubits have you demonstrated? What kind of gate fidelity, if you can share? Any numbers would help.

Pierre: Sure. What we are focusing on right now is really single-qubit devices.

Compared to other, let's say, commercial companies, we think that quantity is not the only metric that matters. Quality is even more important, and it's more difficult. It's probably hard for everyone, but it's more difficult to focus first on quality and then focus on scaling the technology.

That's why we are focusing on single-qubit devices. We are working on two things: increasing the coupling, so how fast we can drive the spin, and also the coherence time of our spin qubit. These two developments are really at the core of our tech roadmap.

The key result that we will soon publish is this: we computed the fidelity that we can get with our technology, given where we are right now and realistic assumptions about our components. And we can achieve one-qubit gates with a fidelity of 99.99%.

For the two-qubit gate, we have a lower bound of 99.5%. So I think this is very promising, but of course we are still working on these developments.

Yuval: And who would be your customers? Would this go to someone who wants to make a quantum computer, or do you intend to make your own quantum computers?

Pierre: In terms of product, what we will deliver is the chip and the low-level control software. In terms of positioning, it's really close to what Nvidia did in the GPU market, where they were able to offer very high-performance GPU chips, but also the low-level control software, which is CUDA.

We have made the strategic choice of not being full stack. We think that a lot of the value is actually in the hardware, and this is the key bottleneck right now in quantum computing. So we are also partnering with companies developing quantum algorithms and quantum software.

We think that it's really the way to go. Our strength and our expertise is in developing incredibly high-performing quantum computers, quantum chips, and some other companies are really good at developing quantum development kits and quantum algorithms.

Yuval: Does your system still need cryogenic cooling?

Pierre: It does, yes. One thing to think about is that cryogenics is a problem if your qubit is really big, like a superconducting qubit. Because in that case, when you scale, the volume that you have to cool down rapidly becomes gigantic, like the size of a soccer field.

But our qubit is 10,000 times smaller than a superconducting qubit. So in the end, we don't have a bottleneck on cooling power in the near term.

Yuval: Now, there are many companies making quantum computers, and they have different modalities. Of course, superconducting is one of them, but you've got trapped ions and neutral atoms and photonics and so on. How do you plan to stand apart and differentiate from these companies?

Pierre: I think what is interesting is that we have a very unique approach.

If you look at the core strengths of other technologies, take first, for example, atoms and ions. They are super good because their qubits lie in vacuum, so they are extremely well isolated. They can have high fidelity. But we know that it's really hard to scale them above a threshold that is maybe 1,000, maybe a bit more.

We know that it's going to be very hard for them to scale. On the other side, if you take superconducting qubits, they rely on semiconductor devices and fabrication, so they're much easier to scale, but they are really at their limits in terms of coherence time. That's why there's actually this huge effort on error correction.

But error correction would be so much easier if you had better qubits. So really what we offer with C12 is to combine the best of both approaches: compatibility with semiconductor devices, together with this ideal architecture of a spin qubit in vacuum.

Yuval: At what stage is the company? If you could tell me a little bit about the number of employees perhaps, or money raised, or most importantly, how close are you to having a product that customers could use?

Pierre: A lot of questions here. We founded the company two years ago, in 2020. Maybe it's worth mentioning that I founded the company with my twin brother, Mathieu.

Mathieu is really the scientific mind behind the technology, and I'm more in charge of business development. Last year we raised our seed round, a total of $10 million. It was led by 360 Capital, which is a large European deep tech fund.

On board we also have Airbus Ventures; BNP Paribas, which is one of the largest global banks in the world; and Bpifrance, which is a French state fund. Now we are 30 people. In terms of commercial products, the first product that we will offer is the C12 simulator, and this is quite exciting.

It's really a digital twin of the chip that we are building. A lot of people want to understand, when the hardware is ready, what kind of algorithms they will be able to run and what kind of applications they will be able to perform. So this will be the first product, and we will announce the release very, very soon. It will be accessible directly through the cloud.

Then we will focus on near-term applications. We think that we don't have to wait 10 years or more, and we don't have to wait for error correction to have a useful application, especially if you look at chemistry applications. That's why we are already launching projects with some customers on designing those first quantum processors that will help them with some of their critical chemical processes.

We have one project with Air Liquide, which is a global gas and services company. With them, we are first identifying the right application we want to focus on, and then designing the right layout for this application, before actually delivering the chip to them, which will be in a couple of years.

Yuval: So are you saying that the chips are essentially customized to the particular application, whether it's the connectivity or number of qubits or something else?

So one chip per type of application?

Pierre: Yes, you got it right. If you want to get the most performance possible from the early quantum computers, it's very interesting to have this customized approach. We call this application-specific design, and indeed there are a couple of parameters that you can modify in the layout to really optimize the quantum resources that you can offer for the specific application.

Yuval: Can you give me an example? For instance, how would a chemistry application be different in terms of the chip layout or resources from, say, an optimization application?

Pierre: Right now, C12 holds six patents, both on fabrication and on design.

But on this specific topic, we are still writing the patent; it's still in the patenting process. So I will have to be a bit vague, but what is important, of course, is that you can customize the number of qubits. For example, if a qubit is really rarely used in your computation, maybe you can actually skip it.

We can also change the connectivity between the qubits. You can have all of them connected together, but maybe that's something you don't really need, so then you have a more advanced design where you use only the part of the connectivity that you need.

And of course, we are also working on special gates that can be very useful for quantum chemistry applications.

Yuval: So when you say special gates, do you mean beyond two-qubit gates? Do you mean multi-qubit gates or something else?

Pierre: That's something I need to stay vague about.

Yuval: Very good. So as we get close to the end of our conversation, I wanna go back to the beginning and ask you again about carbon.

What is unique about carbon that made you choose it and not some other element?

Pierre: One answer is in our name: the fact that you can get a very pure isotope of carbon. The other thing, of course, is that you can grow it into carbon nanotubes and have them suspended over your chip. This suspension is really interesting because then you have a vacuum around your hosting material, which, as shown with atoms and ions, is extremely good for qubits.

Yuval: Excellent.

So how can people get in touch with you to learn more about your work?

Pierre: We have spent some time designing our website, c12qe.com. There you have a lot of information about quantum computing, but also about our technology and really how it works, how we fabricate our chips, and how we operate them.

And of course, we are hiring. On the website you will also see all the different positions that are available right now at C12. So if you want to join the quantum race and you want to build something very unique, let us know.

Yuval: Excellent.

Thank you so much for joining me today.

Pierre: Thank you so much for the invitation, and it was a pleasure to be here.