American Scientist

Andrea “Annie” Kritcher has one of the highest-pressure jobs in physics. At Lawrence Livermore National Laboratory, she is the team lead for Integrated Modeling, and she is principal designer for fusion-energy experiments at the National Ignition Facility, or NIF. Those experiments bombard a small pellet with 192 precisely timed laser beams, causing it to implode and reach pressures of more than 100 billion Earth atmospheres. The process is called inertial confinement fusion (ICF); its goal is to live up to the facility’s name and achieve ignition, a long-sought state in which fusion reactions put out more energy than they take in. In 2021, after more than a decade of trials, NIF briefly came close to ignition. Finally, on December 5, 2022, Kritcher and her team achieved ignition, raising new hopes for fusion as a practical energy source. Kritcher spoke with American Scientist special issue editor Corey S. Powell about the historic achievement. This interview has been edited for length and clarity.

From your perspective, what makes fusion such a difficult problem? Why is it so hard to model, and then why is it so hard to control the experiment?

The thing that makes it so difficult to control and model is that we need extreme conditions. We need extreme temperatures. In ICF [inertial confinement fusion] we also need extreme densities. We’re reaching pressures that are more than two times the center of the Sun, and temperatures that are more than five times the center of the Sun, in our experiments. We’re making the most extreme plasma state that you can make on Earth. As you can imagine, since that’s not been done before, there’s quite a bit we don’t know about the materials science.

In these experiments, laser beams enter and hit the inside of a hohlraum [a hollow cylinder made of heavy atoms such as gold or depleted uranium] and create a very intense radiation bath. We have to be able to model that condition, and then we also have to model the plasma conditions of the implosion as it’s imploding. Inside this intense radiation bath sits a spherical capsule, and in our experiments it’s made out of diamond. And inside of that spherical capsule sits deuterium and tritium fuel [two hydrogen isotopes].

When we make this intense radiation bath, it heats the outside of the capsule that holds the fuel. That heat explodes the outside of the capsule, which, in a rocket-like effect, sends the remaining capsule and the fuel inward and squeezes it. We’re taking something the size of a BB and squeezing it down to a size roughly half the diameter of a human hair. We’re squeezing all of this material down to extremely small volumes, extremely high pressures, and reaching very high temperatures. But as the material implodes, it’s not just the end state that’s hard to model. It’s the entire implosion that’s hard to model as well. There are many states along the equation of the density-temperature-pressure relationship that we don’t understand about the materials we’re trying to implode. There’s no experimental data.

You are dealing with temperatures, pressures, and conditions that go beyond known physics. How do you model something when you’re going into such unfamiliar regions?

We have some low-pressure data and single points of parameter space that we use to benchmark our equation of state models. Then we have our theory, our transport models, radiation transport models, and hydrodynamics models. We benchmark these models as we go to integrated experimental data. We don’t have a lot of data to support the basic physics models, or to test them, but we have generated a lot of integrated measurements, integrated datasets. You do an experiment where all this complicated physics is happening, and then you do an after-the-shot simulation to try to match all the observables from that experiment. It’s like an integrated check on how the modeling is doing.

Integrated modeling is basically trying to model the entire system, from the laser hitting the hohlraum to ignition and what comes out of it. It’s modeling how the lasers interact with the gold and depleted-uranium cylinder, how that implodes the capsule, and how that happens symmetrically. And then modeling the plasma physics conditions in the center of the dense plasma that we create, and the diagnostic signatures that come out of that as well.

What difficulties do you encounter when you’re trying to deal with all of these different parts of the modeling?

There’s quite a bit of simultaneous optimization that has to occur. When NIF first started [in 2009], it went for the highest yielding potential design that we had. The highest potential gain design. And that design had issues. It was more susceptible to instabilities. It just didn’t work. There are a lot of situations where your model says one thing, but when you actually go to do the experiments you find that you’re missing physics, or you can’t calculate that physics because you don’t have the resolution, or there’s something you don’t understand.

Over the last several years we’ve been working to rebalance, optimizing what’s good for the implosion and what’s good for the hohlraum together. They’re usually not the same thing. We want to increase the size of the implosion, but when you do that, it becomes a more massive target. If you don’t have more laser energy to blow off the material and implode that extra mass, then the implosion goes slower, and if you can’t get it going fast enough, you can’t squeeze the material fast enough.

We don’t have any more laser energy to drive the implosion, so we have to make the hohlraum that surrounds it more efficient. For a given amount of laser energy we put in, we have to get to higher temperatures. There’s a lot of work around trying to do that symmetrically. To make the hohlraum more efficient, we had to make it smaller compared to the capsule. That way, the laser beams don’t pass by the capsule; they go where you want them to go to get a nice, uniform radiation bath that surrounds the capsule. There’s quite a bit of modeling that went into defining each laser beam to get a symmetric radiation drive during the entire laser pulse.

How long do you have to keep your lasers focused on the target during your fusion experiments?

It’s about nine nanoseconds. There’s also a really small, 2-micron fill tube—for reference, a human hair is 50 or 60 microns in diameter—that goes into the capsule, which holds the deuterium-tritium fuel. That little fill tube can send a jet of capsule material into the hot plasma and radiate the energy away. It’s asymmetric, but the bigger issue is that the capsule material is high-Z [made of heavy, high-density elements]. If it shoots into the plasma material, it very quickly radiates the energy away from the hot plasma and cools it down. That has been one of the biggest challenges to model.

The goal is to reach just the right conditions in the hot plasma so that the fusion takes over and heats the plasma itself. Without what we call self-heating, we would never reach ignition conditions in these experiments. Out of the fusion reaction comes a neutron and helium. The neutrons escape, but the helium gets reabsorbed. That process carries a lot of energy. It self-heats the plasma, very rapidly increases the temperature, and ignites the fuel. We’re trying to get the capsule to implode, ignite, and stay together long enough that we can burn up as much of the fuel as possible before it explodes.

NIF had a promising fusion result in 2021, but then the next few runs couldn't reproduce it. Why was it so hard to build on that success?

That plasma was designed to ignite under really good experimental field conditions. Unfortunately, all of the targets that we shot had quality issues. Dust particles, even particles the size of bacteria, can get into the capsule in fabrication or fall on the top of the capsule and ruin the experiment. We can see it with our diagnostics. In the x-ray pictures of the hot plasma, we see this little contaminant radiating all the energy away. Unfortunately, in two or three of the experiments, that killed the implosion. In another one, we had a big unintentional asymmetry, which we call an odd mode asymmetry. There are 192 laser beams, half firing from the bottom and half from the top. If the top half lasers fire just a percent different than the bottom half, it can squeeze the implosion and have it shoot out to one side.

“We’re trying to get the capsule to implode, ignite, and stay together long enough that we can burn up as much of the fuel as possible before it explodes.”

• Facebook
• Twitter

We have been making design changes to try to be more robust with regard to these issues. In September 2022, we conducted the first test with extra laser energy and a thicker ablator. Since it was the first test, it was hard to get symmetry right, and the material was squashed like a pancake. But even though it was squashed like a pancake, it still produced nearly as much yield as the 2021 experiment, which had been perfectly spherical. That test was a good indication that even if we had this big perturbation, we could still get high yields. The idea is to be more robust to the presence of flakes on the capsule or to asymmetries. That’s what we did with the last design changes.

You’re pushing the experimental envelope and pushing the modeling envelope simultaneously. Have you learned new things on how to model plasma physics from watching and having more data to work from?

We have various ways that the data feeds back into the modeling. A nice example of that is in the hohlraum plasma. We had kinetic effects in the plasma that were resulting in diffusing part of the density of the plasma in the hohlraum. Based on the focused experiments that were done, we knew we had to include that physics in the modeling to accurately represent what was going on in the hohlraum. Some things we’ve figured out and some things we haven’t. But a lot of the process is trying to update our material response understanding in these extreme conditions, which is a physics model, and figuring out what physics we have to include that’s not in the models already. We’re modeling a big system and there’s microphysics going on. A lot of times we have to use reduced-order models. It’s figuring out how to benchmark those reduced-order models to accurately represent the experiment.

You made headlines when your fusion reaction produced more energy than it took in from the laser, though NIF as a whole still ran at a huge energy loss. What would it take to achieve system-wide breakeven?

NIF wasn’t designed to be an efficient fusion energy–generating laser. It was just designed to make it work. The wall plug energy of NIF is 350 megajoules, and we are generating about 3 megajoules of fusion output. We got more fusion out than laser energy on target, but a working fusion plant would need to significantly increase the efficiency of the wall plug to the laser energy. And we need higher gain designs. A realistic prototype fusion power plant would be a gain of 30 megajoules, and a realistic operating plant with newer laser systems would be a gain of 100 megajoules. Right now, we’re at a gain of 1.5 megajoules. We’re a lot closer than we used to be. What we did was proof of principle, but it’s still quite a long way to go for fusion energy.

What might that path look like? How could the tests at NIF lead to a commercial fusion power plant?

There’s a lot of technology and design improvements that would need to happen. Even if we got really good at making uniform targets—which is one of the reasons why it takes us so long right now to do experiments—the design would have to change to do fusion 10 times a second [necessary to generate continuous power]. After this experiment, we checked the box on the physics requirements. Now it’s really hard engineering stuff. It’s not going to happen tomorrow, but it also took a long time from the first flight to having commercial flights. I’m hopeful.

A podcast interview with the scientist: