Transcript | High-Fidelity Qubits

Getting default-tolerant quantum computing requires improving qubit quality. The dream of engineers is the four nines, or 99.99% fidelity. Quantinuum, well-known for its H1 trapped-ion system, has found a way to achieve this elusive goal by switching from ytterbium to barium-137. Find out more about this state of the art in trapped-ion qubits in this episode of The Post-Quantum World. I’m your host, Konstantinos Karagiannis. I lead Quantum Computing Services at Protiviti, where we’re helping companies prepare for the benefits and threats in this exploding field. I hope you’ll join each episode as we explore the technology and business impacts of this post-quantum era. 

As you may have heard, late in 2021, Honeywell Quantum Solutions, who we’ve had on the show, merged with Cambridge Quantum to create Quantinuum. Our guest today is an advanced physicist at Quantinuum, Tony Ransford. Welcome to The Post-Quantum World.

Hi. How is it going?

Good. I’m really excited to have you on because of some potentially game-changing work you’ve been doing. As we covered last time Honeywell was on, the H-series of machines use ytterbium atoms to create trapped ions. Can you tell us a little bit about the limitations or challenges that approach leads to?

Yes. Ytterbium has been the staple. Many people use in ytterbium in academia and industry for trapped-ion qubits, but they do have downsides. Probably the biggest one is that the lasers used to control and do the gates are in the ultraviolet spectrum. 

These are challenging lasers for a lot of reasons, and they’re not compatible with industry laser technology that’s available and cheap and robust. The ideal thing you would want is some visible or infrared so we can leverage telecom equipment. That’s one of the challenges of ytterbium, along with the fidelity that you can measure the ytterbium qubits at. It turns out that, at least, with the standard method for measuring ytterbium, the fidelities are low, comparable to where we want to be. Maybe not comparable to where we are right now, but the fidelity is just over three nines, 99.9% of fidelity — the best ever done in ytterbium. While it’s been a great qubit and worked for academia and industry, it does have its limitations.

Yes, and we’re trying to get to four nines. That’s the idea. This led you to work with barium ions, which could be cheaper — and, if you use the wrong ones, maybe dangerous, if you want to talk a little bit about that.

Yes. I was at UCLA before I came here, and that was the first place to use radioactive barium ions. 133 is the isotope you want to use for quantum computing. That’s radioactive. There are a lot of downsides to using radioactive barium — less obvious than the radiation safety. What we would call loading, where you try to get the material into the ion trap to create the qubits, can be quite challenging with radioactive qubits. However, the structure is optimal with the radioactive ones. The real challenge we had was taking another look at the atoms that nature gave us that weren’t radioactive and figuring out how to use them in order to make qubits out of them.

Yes. So, that brought you to barium-137.

Yes. It has a nuclear spin of three-halves — the radioactive version has one-half. That extra nuclear spin in barium-137 causes some issues. It creates more levels that you have to deal with. We had to come up with techniques to deal with all these other levels in order to prepare the qubit in a pure quantum state, which is the initial goal. That was one of the challenges we had to overcome.

In the end, you were able to use more standard visible green lasers with 137.

Yes — 137 inherently has visible colors, which is why we chose barium. That’s what makes that atom attractive, but initially, it seemed hard to prepare this pure quantum state, but we’re able to use these visible lasers to prepare a pure quantum state.

Then, of course, the big news that was announced was the four nines — 99.9904% — and that’s a measurement of SPAM fidelity. Did you want to explain that to us?

SPAM stands for “state preparation and measurement.” Historically, SPAM fidelities for any qubits have hovered around a little above the 99.9X — a little more than three nines — whether it be ytterbium or even barium-133 or previous ions. With barium, the measurement comes for free, whereas in ytterbium, the measurement is the challenging thing. Barium, that’s quite easy to measure, but preparing it is hard, especially 137.

For the SPAM, you have to address all of these states with lasers in what we call optical pumping, but prepare them into a single state, and this is what we call our qubit zero. Then you have to be able to measure. Our qubit zero dye doesn’t light up. It glows like a star when we shine the laser on it, or in the zero state, it’s dark, and there’s no star. Then, when we put in the one state, it lights up. That is the star. Differentiating these two things and how well we can do it is what SPAM is — how well we can tell we’ve made a zero or a one, and how well we can actually make the zero and make the one.

Also, it creates quite an image in the mind of the listener. I always look for scientific moments like that, where it’s, like, “That’s something you can visualise.” All this comes from DiVincenzo criteria for making quantum computers. So, there were the five for the machine and then the two for communication. SPAM was two of those — state preparation and the measurement.

Yes. Creating fiducial state is one, and being able to measure that state out, so that’s exactly what we’re doing, where you can think of it as the end of the calculators, the inputs, and the thing that you read at the end.

Of course, somewhere in the middle there, you have to have gates. Has any work been done yet on getting these barium-137 qubits to gate with each other?

We’re actively working on that right now. We have technology-development efforts looking into gating of barium-137 qubits as well as looking to transfer. There are a handful of other things you have to make sure work before you build a quantum computer out of them, but getting the SPAM was maybe the one thing that was the most uncertain in our minds at least, and so getting that out of the way opened the floodgates to move forward. And again, now we’re going to be doing gates on barium, but we’re going to be using commercially available technology and these very nice lasers that should help us push the actual gate fidelity better, not just the SPAM fidelity.

Will this give us any other benefits? Do you anticipate being able to control more of these ions compared to ytterbium? Because that was always the rub — how many you can move and control?

Yes. We have good plans for how to move and sort scalable qubits with trapped ions. One of the things you have to solve the scale is, you have to integrate the photonics, and you have to be able to address these atoms with lasers from inside this vacuum chamber that houses the chip. UV turns out to be very challenging for this. These photons are like bullets once they’re blue, and these green photons are a lot softer, so they don’t damage the integrated photonics. There’s been a lot more development historically with visible wavelength and IR wavelength –integrated photonics — just for telecom reasons. Barium will allow us to take the next step forward in doing integrated photonics in addressing large-scale devices with lasers.

Do you anticipate a different layout, a different way? There’s a bow-tie model for how the ions are present. Do you anticipate some kind of new, more three-dimensional latticed structure or something with this kind of laser?

I don’t think the structure will change. We’re moving toward two-dimensional geometries because you need to have that to scale. What becomes challenging is how you deliver all the lasers to address the ions in lots of different locations. Barium will help you there because you’ll be able to route the lasers on a chip as opposed to through free space, trying to hit the ions as they’re levitated above this trap. You can, in a manufacturable way, route all the light to the different zones that we address the qubits with.

This would be a form of redesign. It wouldn’t just be swapping out ions in the current machine.

That’s right. It’s been on our road map for a long time to move in this direction. It’s not so much the change of direction, but it’s an enabling technology to have this visible light. It will make it much smoother moving forward.

The goal here with the high fidelity, obviously, is going to help along things like error correction, correct?

Yes. Historically, the difficult thing has been the two-qubit, the entangling fidelity, the two-qubit gate fidelity, but as we make improvements in that — and these lasers with barium will allow us to make great strides in two-qubit fidelity — we want to make sure that the SPAM fidelity doesn’t become our limiting thing. What this work has shown is that we can achieve the SPAM at a fidelity that we would be happy with long-term so we can check it off the list and redirect our focus onto these gating operations.

Yes. There have been some things said about how many qubits of a trapped-ion type would be required to create a logical qubit. Do you anticipate this one change being a significant change to that?

There’s going to be a lot of technology that has to be developed. It would be pretty hard to point at one thing. We’ve done some recent work with logical qubits — on the order of 10 physical qubits per logical qubit — but the fidelity affects the size of that logical qubit.

Other recent work in fidelity — it seems like there’s a lot of that going on right now. We’re recording this right now toward the end of March. Just a week ago, Microsoft announced that they have some new advances with topological computing. Did you come across that?

I’m definitely not an expert on topological qubits. I did see that they think they measured — it was the zero mode of a Majorana particle, or something like this, but I’m not an expert in that field at all.

I was just wondering if you had any thoughts about how their claims will stack up to improve fidelity in a system that already is known working like trapped ion.

Yes. I’m not so certain what their fidelities would be like with then.

Also, the papers that were published in Nature, did you come across those where they got better fidelity on silicon qubits? They were gating systems together. I don’t know if your team looked at any of those too.

We monitor all the developments. Other than SPAM, which is my specialty, I don’t keep track of where the records are or who’s winning there.

Yes. I didn’t know if you guys keep track —

We do. I’m just not an expert on two-qubit gate fidelities or anything like this.

You had mentioned earlier that this was already in your timeline, this idea.

The moving toward the 2D geometry and then moving toward integrated photonics where the lasers are delivered through the traps — that’s right.

There’s already, potentially, a system that someone behind the scenes has in mind as maybe being the first one where this would appear.

Yes. We think about integrated photonics a lot. We think about scalability a lot. Everything we do is generally with that in mind.

Yes — to create a road map. Are you working on anything else that would affect the timeline? For example, any work in how you would have systems work together like interconnect? Is there anything like that going on?

Yes. We have plans for making things scalable, but there are a lot of pieces that need to be put together, and a lot of technology development needs to happen before we get to that point, but we have an open mind about all these different ways and keeping scalability in mind.

Would you say that barium-137, for now, is the next qubit, or is there still work being done to see if you’ll get even better results with something else?

We have better qubits in mind. It’s important to keep an open mind, but they seem very promising. Our technology development is working on two-qubit gates and transport, and figuring out how to make a scalable system out of barium. That’s what we’re working on right now.

The potential is that that’s all we’re going be hearing one day. Potentially, it’d just be barium-137. Ytterbium goes away.

I’ve been using ytterbium for a decade or so. It’ll be a little sad to see it completely go away, but this UV issue definitely is something that needs to be overcome. Materials, in general, just don’t seem to like ultraviolet light. So, just from a technology perspective, there are a lot of reasons why you move toward barium.

I realise that you’re very specialised in this area. But because your company is a merger of that pure hardware side with Honeywell and the more software side from Cambridge, how does that impact your day-to-day research? Do the two sides think ahead of the other pieces that might be involved? While you’re working on the new type of qubit, is there already thought being given to how the control systems are going to be handled — the software, or anything like that?

Yes. It’s only recently happened. We haven’t been merged for that long, but we’ve already had nice communication with the Quantinuum team that’s over in the U.K. and other parts of the world. There’s already drawing out “How can you guys help us understand what machines we should build for the future?” They have a lot of insight to long-term algorithms and stuff like that. We’re boots on the ground to understand what we’re building. There’s already starting to be some overlap there where we can learn a lot from them, and, hopefully, they can learn a lot about our capabilities from us.

That’s pretty exciting, because I believe that it is a full stack when it comes to quantum computing. You squeeze every little bit across the range. It’s neat that all that new information is coming your way. Are there any other exciting papers that we’ll be seeing soon? Any other stages of this research?

I certainly hope so. Like I said, we’re moving forward. The big question is, we need to get some high-fidelity two-qubit gates out of barium. We’re actively putting a lot of effort into that. We’ll see it, hopefully, before too long. Things move fast here.

This is interesting to me, but the little pieces of technology that are needed incrementally to get these achievements to happen, how far out do you reach? Do you contact other research groups on things that aren’t even commercial yet? Maybe something on the lines of how a laser is controlled, or anything like that? Is there that other kind of behind-the-scenes synergy we’re not really aware of?

We have partnerships, and we interface with some academia and some industry and find overlaps and think both short-term and long-term in that sense. We stage ourselves where some people are working on the immediate problems, and some are thinking about the next steps. That’s important.

Thanks for doing this bleeding-edge work, because we always talk about these unknowns in this industry: What’s the next unknown advance that’s going to make everything change? And increasing fidelity obviously is super important. I’m excited to see how this does when it’s gated and running. Did you have a remote guess at the time range where we might see these things interacting with each other, these qubits?

Everything takes some time. We’re working in a faster regime now where things are years instead of decades. That’s probably the best guess I could have.

It’s still in the years range.

Yes. It’s very hard to tell.

As fun as it is to visualise scooping up the ytterbium with a little micro vacuum and conserving these, it’s not going to be quite that easy.

That’s right. We’ve demonstrated on a system — we have barium in there — so we’re moving forward as quickly as we can.

Exciting, yes. I’ll definitely be tracking your progress. I don’t know if there’s anything else you wanted to share about what’s going on at Quantinuum right now. Any other work?

It’s very exciting. The H-series systems are performing world-class systems. Hopefully, we’ll have the new systems online as fast as possible.

Yes. It’s been really great being able to access them. When Honeywell first came on, we talked about how important that is just to be able to access these systems and being able to run them per shot — that kind of model. It’s opened up a lot to the world of quantum computing. You’re all doing very important work there, so keep it up.

Thank you. Thank you very much.

The quantum economy 2.0 features the use of superposition and entanglement in technology, unlike quantum 1.0, which was based on semiclassical technology like the transistor. This new economy will feature quantum computing, sure, but also networks and sensing. Sensing is already resulting in better medical scanning and will be applied in other fields like improved detection of coming earthquakes. 

Many companies involved in quantum computing will be spending more than earning until around 2030. Starting too late will make it hard to play catch-up, though. Companies need to consider now what their business model is, and what impact quantum will have on it later. When do you enter the fray? How much staff will you need to participate? 

Consider how quantum 1.0 affected our economy last century. Transistor-based tech is found in trillions of dollars’ worth of industry. Quantum 2.0 could be 30% of our overall economy this century. 

There may already be global supply chain issues with quantum 2.0. China’s been known to not respect intellectual property rights. We are already seeing export regulations similar to what happened in the early days of cryptography. NIST is selecting new ciphers to combat the threat that powerful quantum computers will pose to cryptography, but companies will have to do their part and replace obsolete cryptography. This includes digging into legacy areas they may not have even accounted for. It takes time to replace crypto, and the time is now. 

That does it for this episode. Thanks to Tony Ransford for joining to discuss qubit developments at Quantinuum. Thank you for listening. If you enjoy the show, please subscribe to Protiviti’s The Post-Quantum World, and leave a review to help others find us. Be sure to follow me on Twitter and Instagram @KonstantHacker. You’ll find links there to what we’re doing in quantum computing services at Protiviti. You can always DM me questions or suggestions for what you’d like to hear on the show. For more information on our quantum services, check out Protiviti.com, or follow ProtivitiTech on Twitter and LinkedIn. Until next time, be kind, and stay quantum curious.