Optical atomic clocks with Judith Olson, Infleqtion


Judith Olson, a specialist in optical atomic clocks from Infleqtion, is interviewed by Yuval Boger. They discuss the advancements in timekeeping technology, highlighting the role of optical atomic clocks in scientific research, GPS, and telecommunications, particularly for 5G and 6G networks. They cover challenges in clock synchronization and the push towards miniaturization for wider application, the size, weight and cost of clocks, and much more.

Full transcript

Yuval Boger: Hello, Judith, and thank you for joining me today.

Judith Olson: Hi, nice to be here.

Yuval: Great to have you. So, who are you, and what do you do?

Judith: My name is Judith Olson. I am a PhD physicist who’s pretty obsessed with atomic clocks, in particular, optical atomic clocks are kind of my bread and butter.

Yuval: Is there a particular joke that comes up when people hear atomic clocks, or they ask you what the time is or anything?

Judith: No, I think the closest is the sad reality joke that many people doing clocks and metrology are so heads down that they tend to be late for a lot of meetings, which is a pretty silly feeling when you’re supposed to be people in charge of time.

Yuval: So, what is an atomic clock? What is an optical atomic clock, and why would someone need one?

Judith: An optical atomic clock is a special kind of atomic clock. Most of the atomic clocks out there these days that you can purchase at least are microwave atomic clocks. So all that a clock really is is something that ticks, and then you count those ticks, and you know each tick lasted a certain amount of time and so you can add up all those intervals of time to tell what time it is now. Atomic clocks just tick faster than any other clock that’s out there. We’re talking thousands of times faster than the best microwave clocks and millions of times faster than the average wristwatch. So, by ticking a lot faster it’s similar to adding more checkmarks on your ruler. You can divide time up into smaller pieces, and then the error in each one of those pieces matters a little bit less. So effectively, you can tell time much, much better with an optical clock than you can with other types of clocks.

Yuval: Give me some numbers, please. What’s the frequency of your typical microwave clock, and what’s the frequency of the typical optical atomic clock?

Judith: So if you’re lucky, your microwave clock maybe ticks at 10 gigahertz. A gigahertz is much more common. Some of them are much lower than that. And then, optical atomic clocks, you’re talking about things that are ticking at hundreds of terahertz. So our clock, for instance I think, is at 385 terahertz. It’s going to depend on what specific frequency you’re looking at there. It’s about 10,000 times higher tick rate than the average microwave clock.

Yuval: But a gigahertz clock sounds like a lot. Why do you need a time measurement that’s more fine than one over a gigahertz?

Judith: Yeah. So you start being able to do some really crazy science in practical applications when you can tell time this well. There’s actually an interesting paradigm shift going on right now where the second is to find in terms of cesium and how well cesium fountain clocks tick. And it’s a little bit problematic because we’ve now built clocks like optical lattice clocks and ion clocks that tick more reliably than the second is defined. So it becomes really difficult to talk about timing and clocks and even conceptualize what’s going on when you start dealing with time that is more precisely defined or that time that is more precise than the second and time itself is actually defined. You start doing weird stuff. This also means, though, you can do things like start telling if fundamental constants are actually changing. So the fine structure constant is a big one where they love to measure different clocks, and then they’ll go back and measure them a year or two later and see if the clock ticking rate has changed at all. And that gives you insight into whether or not some of these values we kind of just take for granted as being constant are actually constant. You can also use them to try to search for dark matter or dark energy. That’s another interesting topic these days. They’re core to a lot of other technology out there. So having really, really stable frequencies is very important for the detection of gravitational waves and black hole imaging. All sorts of really interesting things when you start getting to the nitty gritty, where you find out clocks are used. And that’s in addition to they’re at the core of GPS. Whether or not you realize that you’re using an atomic clock every day, just maybe not an optical atomic clock yet. 

Yuval: But does one need an optical atomic clock for a GPS? I mean, I understand that you could use one for a GPS, but if you did, then does it matter? Is it going to be a substantially more accurate GPS?

Judith: So it really depends on how the constellation is employed. This is not quite my area of expertise, but a lot of the GNSS constellations are in geocentric orbits. So they’re pretty darn far away from the earth. So you really care about their long-term kind of holdover. How well are they telling time, hours, or days from now versus now? So, this ends up being really important for deep space navigation and GNSS. And while microwave clocks can do really well there, obstacle clocks have the potential to be even more accurate. So they can do well there, and they can do well there one second from now or an hour or a day from now. So you start being able to do different types of experiments in space when you have better clocks up there.

Yuval: Is there a large market for these kinds of super accurate clocks, or is it just the four people in the world who are looking for dark matter?

Judith: Well, those people definitely have some great clocks. The market is growing. It’s been interesting to be part of it while it is actively growing right now. The clocks, the optical clocks, at least that are coming to market, are not these revolutionary clocks you see in labs that take a cadre of grad students to maintain. These are the clocks that are kind of simplified. And many of them are actually kind of coming in at performance levels that you can buy a commercial clock at. So mazers, active hydrogen mazers, seem to be the target we’ve all agreed upon for obstacle clocks coming to market. This is in part because that’s the best technology that people out there know how to use now. It’s hard to come and say, “Here’s a solution. Help us find a problem.” We’d rather say, “Hey, there’s a problem here already. There are people who already use this technology. What happens if we make this technology smaller, cheaper, and more robust? What happens to this community of researchers who care about this level of timing, and can they do new things with it?” This is happening at a fortuitous time also with telecommunications and the rollout of 5G and 6G and us wanting to send more data faster with lower latency. And that’s another area where optical signal processing and some of the optical references you can get from an optical atomic clock become very useful.

Yuval: How would two of these clocks or more of these clocks be synchronized? How can I make them start at the same time or sort of figure out that they don’t drift or that I understand what one is showing relative to the other?

Judith: It’s one of the hardest parts that gets overlooked by a lot of people who are building the clocks. It’s pretty useless to build the best clock in the world if you can’t measure it against something else. I mean, time doesn’t really matter at all. It’s time intervals that matter. You don’t care about the fact that, I mean, you do if you’re trying to make a meeting. You care about getting there at noon, but usually, you really care about how much time has passed between two events. To do this perfectly, you kind of need those clocks to be at the same location, which defeats the purpose of having a network of clocks. So, people are developing new time synchronization techniques to address this. There’s quite a few of them out there. The one I hear talked about most often is White Rabbit. This was a protocol developed originally at CERN and has kind of made its way out into industrial markets for people really needing high precision synchronization. There’s also a distribution of timing using entangled photons. It’s very similar to quantum key distribution, which is making a lot of headway and is really interesting. In most of these cases, you’re still talking about synchronization to the picosecond level if you’re doing really well, more likely tens of picoseconds or worse. And our clocks, meanwhile, you can start talking about femtoseconds. You can be talking about orders of magnitude better performance that we still don’t have good ways to transfer over long distances. So, it’s an active problem that’s being looked at and worked on. To my knowledge, there’s no really easy, succinct solution that can do femtoseconds. The closest thing would probably be optical frequency comb free space-time transfer I’ve heard can do close to that.

Yuval: About a year ago, I think I remember a conversation with Scott, your CEO, and he mentioned something about data centers. Am I misremembering, or is there a data center application for this?

Judith: There are. They’re going to be pretty niche. So we started talking with a variety of data center folks, and there are two areas that seem to really be worth the extra effort to try to implement an optical class. And those are going to be areas where you need really, really low latency and already have kind of short fibers or specialized fibers. So this is happening in a lot of the optical intro data center communications that are out there. And in doing this, you can have a better timing reference. So there’s less jitter on the signal on the transmit end and on the receive end. You can better correlate them without needing to do so much post-processing. And the post-processing in some communications, especially really short ones, like for cloud computing, can actually be limited in latency by these post-processing needs. So that’s an area we’re looking at where you really want fast, immediate communications, but you’re not sending a lot of information necessarily. And then the other one is kind of the polar opposite of that. Places where you have existing fiber lines, but you want to send more data through them. So you want to increase your capacity. And in this area, we’re looking at ways you can use the optical frequency cone that is built into these clocks, along with the really, really stable optical frequency reference, and start doing novel ways of signal processing that allow you to reduce the channel spacing. So you can cram more data onto one existing fiber, is what we’re thinking with some specialized signal processing and some specialized objects.

Yuval: How large are these clocks? How large, how heavy, how robust?

Judith: Yeah. They’re getting a lot smaller. So, one we have now is already pretty small for an optical clock. I guess I’ve been working on them for so long, it’s easy to take for granted that. Right now, it’s a three-rack-mount box. It’s about 29 liters. It’s the same size as most cesium beam clocks people might use. It’s replacing or comparable to a piece of technology, though, that is about the size of a mini fridge. So, it’s already reduced a great deal volume-wise for the performance you’re getting. And then we think we can make this a lot smaller. So we have efforts towards more of a 10-liter form factor that we’re looking at that’s extremely ruggedized in this case. So we’re talking ridiculous temperature ranges the average person doesn’t even realize are of concern. And then really, really rigorous shock and vibe testing. And that’s an area where atomic clocks have notoriously been lacking in this field ability. They might tick well on a bench in an isolated room with temperature and humidity control, but we want our clock to tick well if you throw it in a backpack or on a mobile vehicle. And then there’s another version of this that we’re working on that’s even smaller where we think we can get down to sub-liter volumes. And at this point, you’re starting to talk about a board scale clock, a clock with performance orders of magnitude better than is available today that you can start putting closer and closer to a CPU that you can put integrated with your network interface cards and timing systems. And that becomes really interesting. Just the different applications you have when you can better co-locate things that need to be well synchronized is interesting. And then the ubiquitousness of it. You can start having these clocks in a lot more locations. And the more clocks you have, the easier it is to build ensembles and to have resiliency in your networks. So we think having the smaller form factor could have a really big impact and open up new applications.

Yuval: And how expensive are they? Can you tell me? I mean, how do they compare to, well, how expensive are they?

Judith: They are about half the price of a maser right now. We plan on reducing costs substantially further from there. But for right now, the performance people can get out of the clock given your alternatives to purchase is, we think, a really good value for customers. It’s going to take a lot of work to bring that price down further. And so part of our go-to-market strategy is really making sure that we’re working with all of our vendors, we’re doing cost reductions internally, and we’re really focused on not just building a clock, but building a manufacturable clock.

Yuval: Can this clock be sold everywhere, or are you aware of export restrictions for these clocks?

Judith: Right now, it’s EAR99. So at present, the ticker prime unit has no export restrictions on it whatsoever. We can sell it anywhere. We do expect there to be restrictions imposed on this type of technology in the future. So right now, Infleqtion has multiple branches throughout the globe, and we’ve kind of seeded this technology while it’s still in a shareable form at these different locations. So if something happens and ITAR kicks in, or we’re not allowed to sell this clock somewhere, chances are we have an office there as well that has already been seeded with this technology and is ready to ramp up. So I think our UK office is a really good example. They recently won funding, and now they’re going to be manufacturing a very similar chain of products out of the UK office in Oxford.

Yuval: You mentioned Tiqker. Is that the name of the product?

Judith: Yeah. So Tiqker is the product line, and Tiqker Prime is this three-rack size unit that we’re bringing to market now, or officially in 2025, I guess, is the language I should be saying.

Yuval: Are they being used today or right now, this is more in the lab and prototyping stage?

Judith: They’re being used. We are still… There’s always going to be a lab prototyping aspect to technology at this level being developed. We’re always going to be looking for ways to improve it or reduce the swap of it, the size, weight, power, or bring down costs. There’s always going to be people tinkering. But the Tiqker Prime unit that we’re building now is kind of exiting the pre-production phase and is entering production, it’s a fixed design. So that one is not tinkering anymore. Now it is building, it is perfecting the manufacturing process, it is doing all of the testing and qualification that’s needed for it to really be a stable product on the market that brings value to customers.

Yuval: How did you get into clocks? Do you have a Swiss background?

Judith: That’d be cool. I was really obsessed with what light was as a kid. I didn’t understand how light works and how we could see things, and it was photons bouncing off of stuff. And so I started reading more books, and I got really obsessed with how light and matter interact. And at the most fundamental level that came down to how are these things happening in time? If photons are moving at the speed of light, and it’s the fastest anything can go, what do these interactions look like from a photon’s perspective? And questions like that I would spend way too much time alone in my room thinking about. And then in undergrad, I had actually another infleqtion employee come back. He gave a talk on Bose-Einstein condensates, which is a novel fifth form of matter. Thought it was so cool. You make Bose-Einstein condensates by shining light at atoms at the right frequencies and the right times. And just this idea that you can use light to interact with atoms and control them and have these really precise interactions was so cool to me. So that naturally led me to want to study clocks because I cared about the timing of these things and the precision of the interaction. 

Yuval: In terms of roadmap, you mentioned the desire to make the clocks smaller and more robust. Is there some other performance parameter that’s really important to work on?

Judith: Yeah, holdover is an area we’re looking at. So, customers for clocks tend to be in two categories. They tend to either want really good short-term performance or low phase noise. And then there are people who want a clock that’s more akin to what you would use in GNSS where you want this very long holdover. You want it so if there’s no external timing, you still trust your local clock. And so those are the two areas we’re pushing on. Holdover, improving that, making it so we can have better holdover than any other clock in the world for extended periods of time, regardless of the deployment environment. Then, it brings down this phase noise. So eventually hope to have phase-stable optical signals, which would be a real game changer, we think in the world of optical signal processing.

Yuval: I should have asked earlier, but what’s inside an optical clock? Is there a vacuum chamber? Are there lasers? I mean, is there a particular type of atom? What’s inside it?

Judith: So the clocks we’re building are deceptively simple. They have a very small little chamber full of rubidium atoms in a vacuum. And then we shoot a single laser at them. It goes through the atoms. And then based on how strongly the atoms respond, we know whether our laser was at the right frequency or not. Then you close the loop using electronic processing. And then you say, you know, our atoms said move over by two Hertz. You move your laser by two Hertz and then sample again. All clocks work on that basic principle, but the more advanced clocks, the ones that you can start measuring gravitational redshift at the millimeter level and do geodesic leveling for those clocks generally, are much more complicated. Most of them have five-ish lasers. So you need to prepare your atoms. You need to trap them. You need to cool them. You need to isolate them. There are all these steps that go into the state preparation, and that gets really complicated. And then you get larger vacuum systems to get all the lasers in and to isolate your atoms enough. So, the clocks coming to market today are so much simpler than a lot of these laboratory clocks. That being said, we still want to miniaturize the laboratory clocks. So we’re also working on a strontium project. And we think that this next generation after the ticker unit that will unlock really new orders of magnitude kind of performance will likely come by switching to a different element like strontium or ytterbium.

Yuval: And last, a hypothetical. If you could have dinner with one of the quantum greats or the clock greats, dead or alive. Who would that person be?

Judith: Norman Ramsey, I think, hands down. He’s the guy who invented the Ramsey method, the method of separating oscillatory fields. So this is kind of the method that almost every precision clock has used in a long time. So it’s really cool. It’s the basis of everything we do. And I’ve also heard he’s just like an excellent teacher and a really nice, encouraging person. So I think he’d be a fun person to pick the brain of and get to learn from for a little bit if that was a choice.

Yuval: Wonderful, Judith, thank you so much for joining me today.

Judith: Yeah, no problem. I hope you had a good time.