From fission to function

In 2021, Paul Blumberg, who began his career on a nuclear-powered submarine before working for the better part of three decades in facility management at Los Alamos National Laboratory, was asked to take a job at the Nevada National Security Sites in Nevada. Blumberg had expected to finish his career in northern New Mexico, but out of a sense of duty, he agreed to temporarily help lead a Los Alamos group that conducts experimental work at the National Criticality Experiments Research Center (NCERC), which the Laboratory operates in Nevada.

Blumberg’s first visit to NCERC changed his mind about ending his career in New Mexico. “I flew out to Nevada on a Sunday, got a tour on Monday, and then on Tuesday, I saw a criticality experiment,” he says. Blumberg watched from a control room as NCERC staff used a remotely operated machine to bring pieces of nuclear material millimeter by millimeter closer together, monitoring the experiment via a video feed. Although one couldn’t tell just by watching that the material had gone critical—reached the point at which the splitting, or fissioning, of atomic nuclei became self-sustaining—the control room’s instruments indicated that criticality had been achieved. 

During his decades in the Navy and in management, Blumberg had never witnessed a critical reaction firsthand, which made the NCERC experiment a singular experience. “I couldn’t believe it,” Blumberg says. “I was like, ‘I’m never leaving.’”

NCERC is the only place in the United States where researchers carry out general-purpose criticality experiments. Nuclear materials react in many configurations—in reactors and weapons, for example—which makes understanding these reactions key to a breadth of scientific and engineering endeavors. Los Alamos researchers have conducted criticality research since the Manhattan Project—the World War II–era effort to build the world’s first nuclear weapons. Today, criticality experiments are completed with specialized machines that are called critical assemblies, which are, in effect, very small nuclear reactors.

Unlike nuclear reactors that are used to generate power, the energy released through nuclear fission during NCERC experiments is very small. Moreover, by carefully controlling the conditions under which criticality is achieved, NCERC staff ensure that experiments never come close to the kinds of reactions that take place inside nuclear weapons. Criticality experiments also differ from the subcritical experiments conducted at Nevada’s Principal Underground Laboratory for Subcritical Experimentation in that during subcritical experiments, the nuclear material never achieves criticality (although at PULSE, unlike NCERC, experiments can use high explosives to simulate nuclear weapon assemblies).

Beyond criticality research, NCERC plays a unique role in training criticality safety personnel—that is, personnel who analyze processes involving nuclear material to prevent the material from going critical—and others who study or work with nuclear material. As Blumberg notes, many nuclear engineers go their entire careers without ever seeing nuclear material achieve criticality, which is one reason that witnessing a criticality experiment at NCERC can be so impactful. The unique expertise of NCERC’s staff is why the facility is also one of a few places in the country that produces radiation test objects—subcritical assemblies of nuclear material used for testing radiation detectors, which are used to test detectors and train personnel to locate nuclear material or determine its composition.

Today, at a time when companies are investing heavily in new kinds of nuclear power reactors to support technologies such as artificial intelligence (AI) and when Los Alamos is ramping up its production of plutonium pits (the cores of nuclear weapons), NCERC is carrying out more training and more research than ever. NCERC staff are meeting this need by using innovative techniques to design and execute criticality experiments, furnishing data to support national security now and into the future.

A legacy of criticality research

In the aftermath of World War II, two fatal accidents at Los Alamos set the stage for NCERC’s approach to criticality research.

The first accident, in August 1945, occurred when physicist Harry Daghlian dropped a tungsten carbide brick onto an experiment he was building around a plutonium sphere, creating a supercritical reaction that emitted a burst of radiation. Daghlian died 25 days later. Although the Laboratory implemented new safety protocols and relocated its criticality research to an isolated canyon site, a second accident occurred less than a year afterward when, during an experimental demonstration involving a plutonium sphere, a screwdriver used by physicist Louis Slotin to lift a beryllium tamper slipped, creating a supercritical reaction that led to Slotin’s death nine days afterward.

After the Slotin accident, the Laboratory implemented even stricter safety measures for criticality experiments. Chief among these measures was the cessation of criticality experiments conducted by hand. Instead, future experiments would rely on remotely controlled machines stationed inside concrete-reinforced buildings. The facility that was built for these experiments, the Los Alamos Critical Experiments Facility (LACEF), conducted its first experiment in 1947.

The earliest experiments at LACEF were designed to answer questions about nuclear weapon design—to ensure, for instance, that weapon designs based on the Fat Man bomb, which was detonated over Nagasaki near the end of World War II, wouldn’t accidentally achieve criticality while in storage or transit. In the half-century that followed, experiments at LACEF addressed other weapons-related questions as well as questions about nuclear power reactors, naval nuclear propulsion, and the fundamental properties of nuclear materials such as uranium and plutonium. By the early 2000s, when the National Nuclear Security Administration (NNSA) decided to relocate criticality experiments from Los Alamos to Nevada, LACEF had become an important resource for programs like NNSA’s Nuclear Criticality Safety Program and Nuclear Emergency Response Program.

The move to Nevada

Between 2004 and 2011, Los Alamos’ criticality research capability was relocated to the Nevada National Security Sites. The decision to move this research was made for several reasons. LACEF was located in a canyon, which was beneficial from a safety perspective but made the site expensive to defend from evolving security threats. Moreover, criticality experiments, which previously had been conducted at laboratories and facilities across the country, had dwindled such that only one facility, which would become NCERC, was needed for them.

“Decades ago, a lot of people could do criticality work,” says Jesson Hutchinson, NCERC’s chief scientist. “But over time, the ability to conduct simulations improved and the cost of security went up, so the facilities that did these kinds of experiments started to go away.” Centralizing criticality experiments in a single location was a way to reduce costs and increase the efficiency of research designed to answer ongoing questions related to nuclear weapons and power, materials science, and more.

In Nevada, NCERC was established inside an existing building called the Device Assembly Facility (DAF) that was originally constructed to host the assembly and disassembly of underground nuclear test devices. When a moratorium on underground testing went into effect in 1992, the facility was no longer needed to support nuclear testing. However, the location of the DAF—far away from the public, in the interior of the Nevada National Security Sites—and the DAF’s robust security made the facility an excellent place to store, and conduct research with, a large inventory of special nuclear materials—that is, nuclear materials such as uranium and plutonium that could be used in weapons. Altogether, NCERC stores hundreds of kilograms of special nuclear material.

The DAF comprises a series of individual steel-reinforced concrete buildings designed with features, such as blast doors and high-efficiency particulate air-filter ventilation systems, that are intended to mitigate the risk of explosions and radiological contamination. NCERC contains two “high bays” for subcritical measurements, storage vaults, and a counting laboratory. NCERC also uses two “cells” inside the facility that contain a total of four experimental machines (also known as critical assemblies) that are separated from operators by blast doors and thick concrete walls. These machines allow researchers to investigate special nuclear materials in different ways. “NCERC is the only place in the country, and possibly the world, where you can do these kinds of experiments with highly enriched uranium and plutonium,” Hutchinson says. 

Understanding nuclear materials

In the past few years, researchers at NCERC have conducted innovative experiments to answer vital questions about plutonium. Plutonium is arguably the most complex element in the world, which is why, more than eight decades after the element’s discovery, researchers still don’t fully understand how plutonium behaves under certain conditions. Experiments at NCERC aim to produce data that could fill in gaps in this understanding and support applications such as nuclear weapons production and nuclear reactor design.

One important NCERC project related to plutonium, the Chlorine Worth Study, was conducted in 2021. The goal of this research was to understand how chlorine absorbs neutrons—a subject relevant to the aqueous chloride process that allows for the extraction of plutonium and americium from waste produced during plutonium pit production. 

During this process, plutonium-bearing salt mixtures are dissolved in a chlorine-bearing solution to extract and purify the plutonium. However, without understanding how chlorine absorbs neutrons, it is difficult to determine exactly how much nuclear material can be safely contained in an aqueous chloride solution at a given time. To prevent a criticality accident, operators had to err on the side of caution, incorporating less plutonium into the solution than would be possible if chlorine’s neutron-absorbing characteristics were better validated.

Plutonium-bearing solutions are difficult to handle and study in part because the concentrations of nuclear material can be challenging to measure. Rather than use solutions for the Chlorine Worth Study, researchers developed a combination of plutonium plates and a plastic (polyvinyl chloride) to simulate plutonium solutions. These materials were arranged in three configurations designed to mimic different plutonium solution concentrations. Using NCERC’s Planet critical assembly machine, which has a platform that can be raised and lowered to bring different parts of an experimental configuration together and achieve criticality, researchers were able to simulate conditions in the aqueous processing line and acquire valuable data to support pit production at the Laboratory.

“Solution limits in Los Alamos’ Plutonium Facility have been a bottleneck in the production line,” says Theresa Cutler, a researcher at NCERC who co-led the study. “The Chlorine Worth Study was important as a way to potentially help increase those limits.”

A complementary project called TEX-Cl (Thermal/Epithermal eXperiments—Chlorine) was conducted in 2024. Where the Chlorine Worth Study used plutonium, TEX-Cl used high-enriched uranium (uranium consisting of more than 20 percent U-235) as the fuel. Conducted in collaboration with Lawrence Livermore National Laboratory, TEX-Cl supported uranium processing at the Y-12 National Security Complex in Tennessee, demonstrating how research at NCERC supports laboratories and production facilities throughout the nuclear security enterprise.

Optimizing with AI and ML

To design the configurations of nuclear and nonnuclear materials used in criticality experiments at NCERC, researchers have historically used tools such as Los Alamos’ Monte Carlo N-Particle simulation code to study proposed configurations. Recent projects at NCERC have augmented this approach by using AI and machine-learning (ML) techniques to advance experimental design.

For the EUCLID (Experiments Underpinned by Computational Learning for Improvements in Nuclear Data) project, researchers at Los Alamos first used ML techniques to seek out areas where errors could be hiding in nuclear data libraries—large collections of data about all known isotopes that include attributes such as neutron and proton cross sections, fission properties, and more that are the product of decades of research around the world. Next, the researchers used AI to help design criticality experiments that could help determine what combination of data was more likely to be correct. By designing two different experimental configurations, researchers were able to develop experiments that probed specific types of nuclear data (such as neutron scattering or nuclear absorption).

A second project, PARADIGM (PARallel Approach of Differential and InteGral Measurements), built on EUCLID’s approach. For PARADIGM, researchers used ML to locate gaps in data libraries related to plutonium’s intermediate energy range. Cutler explains that when a special nuclear material such as plutonium-239 reaches criticality, it emits high-energy neutrons that then slow down into low-energy neutrons. “Neutrons are born fast, but they die slow,” Cutler says. “Between those points, you have the intermediate range.”

Criticality research has tended to focus on understanding plutonium’s high- or low-energy ranges, leaving data gaps in the intermediate range. One reason for this is that it is challenging to consistently slow neutrons down just the right amount and to measure them when they are in the intermediate energy range. “It’s difficult to build experiments that are sensitive to the intermediate energy range, and it’s hard to develop the nuclear theory that’s sensitive to it,” Cutler says. “We had to ask, ‘How do you design an experiment around those neutrons if there’s no way to generate them at that energy level?’”

Using AI and ML, the PARADIGM team determined that copper was an optimal material to use in experimental assemblies that would probe plutonium’s intermediate energy range. By surrounding the plutonium with copper and other materials, and by bringing the plutonium to criticality, it was possible to maximize the chance of causing fission in just the right energy range. Researchers then used AI to evaluate many possible experimental configurations, considering qualities that included the materials’ composition, thickness, shape, and more. Ultimately, the researchers selected two configurations for experiments at NCERC, which were conducted in early 2025. 

In addition to experiments at NCERC, the PARADIGM team is conducting complementary research at the Los Alamos Neutron Science Center (LANSCE)—a kilometer-long particle accelerator in New Mexico—to gather information about copper that will help reduce the uncertainty of the experimental data. The work at LANSCE, which takes advantage of a recent upgrade to the facility that allows researchers to target materials in the intermediate energy range, will help characterize copper’s cross sections—that is, copper’s propensity to scatter or capture neutrons. 

“Criticality experiments are very expensive, and we want to get the most value for the cost,” Hutchinson says. “The old-fashioned way to do this research might involve doing an experiment, only to realize 20 years later that you need to redo the experiment in order to answer certain questions. Now, we’re trying to get the best value by optimizing these experiments.”

Cutler says that the use of AI/ML at NCERC points toward promising future applications. “These days, there’s a lot of research that’s using AI and ML,” she says. “At NCERC, we’re not just using ML for the sake of using ML. We’re using it to meet a need.”

The Deimos test bed

In the past half decade, interest in nuclear power has grown among industry, policymakers, and the public. Part of that interest is because of the proposed development of advanced nuclear reactors such as microreactors, which could be smaller and safer than traditional light-water reactors. 

Many proposed advanced reactors would use fuels such as high-assay low-enriched uranium (HALEU), which has a higher concentration of the uranium-235 isotope than is found in the low-enriched uranium used in traditional power reactors (HALEU contains up to 20 percent U-235 versus up to 5 percent in low-enriched uranium). In the United States, HALEU has never been used in a commercial power reactor, and the fuel must be better understood before advanced reactors can be licensed and deployed. 

A newly developed NCERC capability called Deimos, which came online in 2024, could play a key role in addressing some of the questions that must be answered to help bring a new generation of reactors into service. Deimos comprises a specially machined graphite core that can contain nuclear fuel in diverse configurations. Using NCERC’s Comet critical assembly machine, this core can be raised into a second, larger portion composed primarily of graphite and beryllium as well as additional nuclear fuel, causing the fuel to achieve criticality. 

The Deimos test bed allows researchers to investigate a breadth of phenomena relevant to nuclear reactor design. Developing Deimos involved overcoming many engineering challenges from how to machine the graphite (which is very brittle) into the needed shape to how to accommodate the DAF’s ceiling height, which constrained the test bed’s size.

The first Deimos experiments, which were completed in fall 2024, involved tri-structural isotropic fuel (TRISO)—HALEU encased within layers of ceramic and other materials that prevent fission products from dispersing throughout a reactor system. In February 2025, a series of experiments on Deimos were the first U.S. experiments involving HALEU in more than 20 years. These experiments provided important data relevant to criticality safety that supported Kairos Power, which is developing an advanced reactor that would use TRISO fuel. For example, one set of Deimos experiments involved incorporating steel into the test bed to simulate accidents that could conceivably occur while TRISO fuel was being transported, and another used polyethylene to simulate the accidental incursion of water into a reactor.

Throughout this research, the team demonstrated Deimos’ flexibility and its potential to support future projects. “The goal was to prove that Deimos wasn’t just a one-time experiment, but something that could be taken apart and reassembled for follow-up experiments,” says Cutler, who co-led Deimos. Other experiments are already planned for Deimos, including ones that will support the ZiaCore project—a microreactor design that is being designed at Los Alamos—and Westinghouse, which is seeking data for its proposed eVinci microreactor.

Training and RTOs

The expertise of NCERC’s staff and the facility’s extensive inventory of nuclear material are why, in addition to conducting criticality experiments, NCERC routinely hosts classes for the Department of Energy’s (DOE’s) Nuclear Criticality Safety Program, which conducts research and training to support the criticality safety community. (The Nuclear Criticality Safety Program is NCERC’s primary sponsor.)

Another training program at NCERC is for members of DOE’s Nuclear Emergency Support Team (NEST). NEST is composed of technical experts from across the DOE complex who support a range of programs related to nuclear emergency response—everything from countering improvised nuclear devices and “dirty bombs” to providing preventative nuclear and radiological detection (such as monitoring the Super Bowl for radiological threats) and emergency response (such as the 2011 Fukushima nuclear accident).

To support such diverse missions, NEST responders must understand the specialized detectors that they use to scan for or measure radiation. At NCERC, NEST responders have a unique opportunity to see these tools in action. One way in which trainees gain familiarity with detectors involves the use of radiation test objects (RTOs). RTOs are subcritical configurations of special nuclear material that are hand built at NCERC, and NEST responders’ tools can detect the material in RTOs and other such objects. “Almost all of the detectors that NEST responders use are present at NCERC, and they’re able to train with those,” says NCERC researcher George McKenzie. “There’s just no substitute for hands-on training.”

RTOs have uses beyond training, including benchmark experiments (which provide data to support nuclear data libraries and thereby validate computational models, among other things) and detector testing. At NCERC, RTOs have also supported experiments in areas ranging from nonproliferation and treaty verification to safeguards, emergency response, and more.

Looking to the future

Recent upgrades to equipment at NCERC are supporting the facility’s mission. In the past few years, the control rooms that allow NCERC staff to operate criticality research remotely were completely upgraded. These new control rooms will enable reliable operation for a decade or more.

To support future research at NCERC, with Laboratory-Directed Research and Development funding, McKenzie, Hutchinson, and other NCERC researchers are developing a design for a proposed new critical assembly machine—the first such design to be completed in more than three decades. “Over the years, there have been many thousands, or possibly millions, of simulations run on data from Jezebel,” Hutchinson says, referring to a predecessor critical assembly machine that generated important data about plutonium. “But Jezebel no longer exists, so we can’t get new data from it.” 

The proposed new assembly, called Lilith, would provide another method to validate plutonium data at a time when demand for this data continues to grow. Lilith is being designed to endure for the next hundred years, providing important information about plutonium aging.

NCERC’s staffing ensures that the Center is well positioned for the future, says Blumberg, who today remains the deputy leader of the NCERC Facility Operations (FO) group. Blumberg notes that when he came to the Center in 2021, NCERC-FO had less than half the staff it needed (partly because of the COVID-19 pandemic, which had led some staff to seek jobs elsewhere or that could be performed remotely). Hiring NCERC staff can be challenging: Because of the extensive radiological training and security checks that NCERC staff must undergo, it can take up to three years to onboard each new staff member. 

Today, however, NCERC-FO has a full complement of 14 staff members to support operations. Meanwhile, the team that designs and conducts NCERC’s critical experiments, which comprises researchers from the Laboratory’s Advanced Nuclear Technology group (part of the Nuclear Engineering and Nonproliferation division), is growing to deliver important experimental data at an increasing pace. Joetta Goda, who leads NCERC’s Critical Experiments Team, says that the team—which began with a staff of four researchers who came to NCERC from LACEF—has grown threefold over the years. 

“We still have a ways to go to meet the increasing requests for experiments, measurements, and training,” Goda says, noting that the Critical Experiments Team brings students to NCERC to help cultivate the next generation of researchers. “The student pipeline is very important. Almost all of our crew members, me included, began as students or interns of some sort.”

Blumberg says that improved communication between researchers, facility staff, and the broader nuclear security enterprise have helped bring a sense of purpose and mission to a facility that is being called upon to support an ever-expanding portfolio of experimental work.

“When people talk about capabilities, they tend to talk about equipment, but people are really the capabilities,” Hutchinson says. “One way to sustain and develop that capability is to keep pushing the envelope. The goal is to always keep moving forward.” ★