Calls to restart nuclear weapons tests stir dismay and debate among scientists

When the countdown hit zero on September 23, 1992, the desert surface puffed up into the air, as if a giant balloon had inflated it from below.

It wasn’t a balloon. Scientists had exploded a nuclear device hundreds of meters below the Nevada desert, equivalent to thousands of tons of TNT. The ensuing fireball reached pressures and temperatures well beyond those in Earth’s core. Within milliseconds of the detonation, shock waves rammed outward. The rock melted, vaporized and fractured, leaving behind a cavity oozing with liquid radioactive rock that puddled on the cavity’s floor.

As the temperature and pressure abated, rocks collapsed into the cavity. The desert surface slumped, forming a subsidence crater about 3 meters deep and wider than the length of a football field. Unknown to the scientists working on this test, named Divider, it would be the end of the line. Soon after, the United States halted nuclear testing.

Beginning with the first explosive test, known as Trinity, in 1945, more than 2,000 atomic blasts have rattled the globe. Today, that nuclear din has been largely silenced, thanks to the norms set by the Comprehensive Nuclear-Test-Ban Treaty, or CTBT, negotiated in the mid-1990s.

Only one nation — North Korea — has conducted a nuclear test this century. But researchers and policy makers are increasingly grappling with the possibility that the fragile quiet will soon be shattered.

Some in the United States have called for resuming testing, including a former national security adviser to President Donald Trump. Officials in the previous Trump administration considered testing, according to a 2020 Washington Post article. And there may be temptation in coming years. The United States is in the midst of a sweeping, decades-long overhaul of its aging nuclear arsenal. Tests could confirm that old weapons still work, check that updated weapons perform as expected or help develop new types of weapons.

Meanwhile, the two major nuclear powers, the United States and Russia, remain ready to obliterate one another at a moment’s notice. If tensions escalate, a test could serve as a signal of willingness to use the weapons.

Testing “has tremendous symbolic importance,” says Frank von Hippel, a physicist at Princeton University. “During the Cold War, when we were shooting these things off all the time, it was like war drums: ‘We have nuclear weapons and they work. Better watch out.’ ” The cessation of testing, he says, was an acknowledgment that “these [weapons] are so unusable that we don’t even test them.”

Many scientists maintain that tests are unnecessary. “What we’ve been saying consistently now for decades is there’s no scientific reason that we need to test,” says Jill Hruby, who was the administrator of the National Nuclear Security Administration, or NNSA, during the Biden administration.

That’s because the Nevada site, where nuclear explosions once thundered regularly, hasn’t been mothballed entirely. There, in an underground lab, scientists are performing nuclear experiments that are subcritical, meaning they don’t kick off the self-sustaining chains of reactions that define a nuclear blast.

Many scientists argue that subcritical experiments, coupled with computer simulations using the most powerful supercomputers on the planet, provide all the information needed to assess and modernize the weapons. Subcritical experiments, some argue, are even superior to traditional testing for investigating some lingering scientific puzzles about the weapons, such as how they age.

Others think that subcritical experiments and simulations, no matter how sophisticated, can’t replace the real thing indefinitely. But so far, the experiments and detailed assessments of the stockpile have backed up the capabilities of the nuclear arsenal. And those experiments avoid the big drawbacks of tests.

“A single United States test could trigger a global chain reaction,” says geologist Sulgiye Park of the Union of Concerned Scientists, a nonprofit advocacy group. Other nuclear powers would likely follow by setting off their own test blasts. Countries without nuclear weapons might be spurred to develop and test them. One test could kick off a free-for-all.

“It’s like striking a match in a roomful of dynamite,” Park says.

The rising nuclear threat

The logic behind nuclear weapons involves mental gymnastics. The weapons can annihilate entire cities with one strike, yet their existence is touted as a force for peace. The thinking is that nuclear weapons act as a deterrent — other countries will resist using a nuclear weapon, or making any major attack, in fear of retaliation. The idea is so embedded in U.S. military circles that a type of intercontinental ballistic missile developed during the Cold War was dubbed Peacekeeper.

Since the end of testing, the world seems to have taken a slow, calming exhale. Global nuclear weapons tallies shrunk from more than 70,000 in the mid-1980s to just over 12,000 today. That pullback was due to a series of treaties between the United States and Russia (previously the Soviet Union). Nuclear weapons largely fell from the forefront of public consciousness.

But now there’s been a sharp inhale. The last remaining arms-control treaty between the United States and Russia, New START, is set to expire in 2026, giving the countries free rein on numbers of deployed weapons. Russia already suspended its participation in New START in 2023 and revoked its ratification of the Comprehensive Nuclear-Test-Ban Treaty to mirror the United States and a handful of other countries that signed but never ratified the treaty. (The holdouts prevented the treaty from officially coming into force, but nations have abided by it anyway.)

Nuclear threats by Russia have been a regular occurrence during the ongoing war in Ukraine. And China, with the third-largest stockpile, is rapidly expanding its cache, highlighting a potential future in which there are three main nuclear powers, not just two.

“There is this increasing perception that this is a uniquely dangerous moment.… We’re in this regime where all the controls are coming off and things are very unstable,” says Daniel Holz, a physicist at the University of Chicago and chair of the Science and Security Board of the Bulletin of the Atomic Scientists, a nonprofit that aims to raise awareness of the peril of nuclear weapons and other threats. In January, the group set its metaphorical Doomsday Clock at 89 seconds to midnight — the closest it has ever been.

Some see the ability to test as a necessity for a world in which nuclear weapons are a rising threat. “We are seeing an environment in which the autocrats are increasingly relying on nuclear weapons to threaten and coerce their adversaries,” says Robert Peters, a research fellow at the Heritage Foundation, a conservative think tank. “If you’re in an acute crisis or conflict in which your adversary is threatening to employ nuclear weapons, you don’t want to limit the options of the president to get you out of that crisis.” Testing, and the signal it sends to an adversary, he argues, should be such an option.

Peters advocates for shortening the time window for test preparations — currently estimated at two or three years — to three to six months. The Heritage Foundation’s Project 2025 calls for “immediate test readiness.”

The United States regularly considers the possibility of testing nuclear weapons. “It’s a question that actually gets asked every year,” says Thom Mason, director of Los Alamos National Laboratory in New Mexico. Los Alamos is one of the three U.S. nuclear weapons labs, alongside Lawrence Livermore National Laboratory in California and Sandia National Laboratories in Albuquerque. Each year, the directors of the three labs coordinate detailed assessments of the stockpile’s status, including whether tests are needed.

“Up until this point, the answer has been ‘no,’ ” Mason says. But if scientific concerns arose that couldn’t be resolved otherwise or if weapons began unexpectedly deteriorating, that assessment could change.

If a test were deemed necessary, exactly how long it would take to prepare would depend on the reasons for it. “If you’re trying to answer a scientific question, then you probably need lots of instrumentation and that could take time,” Mason says. “If you’re just trying to send a signal, then maybe you don’t need as much of that; you’re just trying to make the ground shake.”

Studying nuclear weapons without testing

The area of the Nevada desert encompassing the test site is speckled with otherworldly Joshua trees and the saucer-shaped craters of past tests. In addition to 828 underground tests, 100 atmospheric tests were performed there, part of what’s now known as the Nevada National Security Sites. Carved out of Western Shoshone lands, it sits 120 kilometers from Las Vegas. Radioactive fallout from atmospheric tests, which ceased in 1962, reached nearby Indian reservations and other communities — a matter that is still the subject of litigation.

By moving tests underground, officials aimed to contain the nuclear fallout and limit its impact on human health. Before an underground test, workers outfitted a nuclear device with scientific instruments and lowered it into a hole drilled a few hundred meters into the earth. The hole was then filled with sand, gravel and other materials.

As personnel watched a video feed from the safety of a bunker, the device was detonated. “You see the ground pop, and you see the dust come up and then slowly settle back down. And then eventually you see the subsidence crater form. It just falls in on itself,” says Marvin Adams, a nuclear engineer who was deputy administrator for NNSA’s Defense Programs during the Biden administration. “There was always a betting pool on how long that would take before the crater formed. And it could be seconds, or it could be days.”

Kilometers’ worth of cables fed information from the equipment to trailers where data were recorded. Meanwhile, stations monitored seismic signals and radioactivity. Later, another hole would be drilled down into the cavity and rock samples taken to determine the explosion’s yield.

Today, such scenes have gone the way of the ’90s hairstyles worn in photos of underground test preparation. They’ve been replaced by subcritical experiments, which use chemical explosives to implode or shock plutonium, the fuel at the heart of U.S. weapons, in a facility called the Principal Underground Laboratory for Subcritical Experimentation, PULSE.

The experiments mimic what goes on in a real weapon but with one big difference. Weapons are supercritical: The plutonium is compressed enough to sustain chains of nuclear fission reactions, the splitting of atomic nuclei. The chain reactions occur because fission spits out neutrons that, in a supercritical configuration, can initiate further fissions, which release more neutrons, and so on. A subcritical experiment doesn’t smoosh the plutonium enough to beget those fissions upon fissions that lead to a nuclear explosion.

The PULSE facility consists of 2.3 kilometers of tunnels nearly 300 meters below the surface. There, a machine called Cygnus takes X-ray images of the roiling plutonium when it’s blasted with chemical explosives in subcritical experiments. X-rays pass through the plutonium and are detected on the other side. Just as a dentist uses an X-ray machine to see inside your mouth, the X-rays illuminate what’s happening inside the experiment.

Glimpses of such experiments are rare. A video of a 2012 subcritical experiment shows a dimly lit close-up of the confinement vessel that encloses the experiment over audio of a countdown and a piercing beeping noise, irritating enough that it must be signifying something important is about to happen. When the countdown ends, there’s a bang, and the beeping stops. That’s it. It’s a far cry from the mushroom clouds of yesteryear.

The experiments are a component of the U.S. stockpile stewardship program, which ensures the weapons’ status via a variety of assessments, experiments and computer simulations. PULSE is now being expanded to beef up its capabilities. A new machine called Scorpius is planned to begin operating in 2033. It will feature a 125-meter-long particle accelerator that will blast electrons into a target to generate X-rays that are more intense and energetic than Cygnus’, which will allow scientists to take images later in the implosion. What’s more, Scorpius will produce four snapshots at different times, revealing how the plutonium changes throughout the experiment.

And the upcoming ZEUS, the Z-Pinched Experimental Underground System, will blast subcritical experiments with neutrons and measure the release of gamma rays, a type of high-energy radiation. ZEUS will be the first experiment of its kind to study plutonium.

Subcritical experiments help validate computer simulations of nuclear weapons. Those simulations then inform the maintenance and development of the real thing. The El Capitan computer, installed for this purpose at Lawrence Livermore in 2024, is the fastest supercomputer ever reported.

That synergy between powerful computing and advanced experiments is necessary to grapple with the full complexity of modern nuclear weapons, in which materials are subject to some of the most extreme conditions known on Earth and evolve dramatically over mere instants.

To maximize the energy released, modern weapons don’t stop with fission. They employ a complex interplay between fission and fusion, the merging of atomic nuclei. First, explosives implode the plutonium, which is contained in a hollow sphere called a “pit.” This allows fission reactions to proliferate. The extreme temperatures and pressures generated by fission kick off fusion reactions in hydrogen contained inside the pit, blasting out neutrons that initiate additional fission. X-rays released by that first stage compress a second stage, generating additional fission and fusion reactions that likewise feed off one another. These principles have produced weapons 1,000 times as powerful as the bomb dropped on Hiroshima.

To mesh simulations and experiments, scientists must understand their measurements in detail and carefully quantify the uncertainties involved. This kind of deep understanding wasn’t as necessary, or even possible, in the days of explosive nuclear weapons test, says geophysicist Raymond Jeanloz of the University of California, Berkeley. “It’s actually very hard to use nuclear explosion testing to falsify hypotheses. They’re designed mostly to reassure everyone that, after you put everything together and do it, that it works.”

Laboratory experiments can be done repeatedly, with parameters slightly changed. They can be designed to fail, helping delineate the border between success and failure. Nuclear explosive tests, because they were expensive, laborious one-offs, were designed to succeed.

Stockpile stewardship has allowed scientists to learn the ins and outs of the physics behind the weapons. “We pay attention to every last detail,” Hruby says. “Through the science program, we now better understand nuclear weapons than we ever understood them before.”

For example, Jeanloz says, in the era of testing, a quantity called the energy balance wasn’t fully understood. It describes how much energy gets transferred from the primary to the secondary component in a weapon. In the past, that lack of understanding could be swept aside, because a test could confirm that the weapons worked. But with subcritical experiments and simulations, fudge factors must be eliminated to be certain a weapon will function. Quantifying that energy balance and determining the uncertainty was a victory of stockpile stewardship.

This type of work, Jeanloz says, brought “the heart and soul, the guts of the scientific process into the [nuclear] enterprise.”

Is there a need to test nuclear weapons?

Subcritical experiments are focused in particular on the quandary over how plutonium ages. Since 1989, the United States hasn’t fabricated significant numbers of plutonium pits. That means the pits in the U.S. arsenal are decades old, raising questions about whether weapons will still work.

An aging pit, some scientists worry, might cause the multistep process in a nuclear warhead to fizzle. For example, if the implosion in the first stage doesn’t proceed properly, the second stage might not go off at all.

Plutonium ages not only from the outside in — akin to rusting iron — but also from the inside out, says Siegfried Hecker, who was director of Los Alamos from 1986 to 1997. “It’s constantly bombarding itself by radioactive decay. And that destroys the metallic lattice, the crystal structure of plutonium.”

The decay leaves behind a helium nucleus, which over time may result in tiny bubbles of helium throughout the lattice of plutonium atoms. Each decay also produces a uranium atom that zings through the material and “beats the daylights out of the lattice,” Hecker says. “We don’t quite know how much the damage is … and how that damaged material will behave under the shock and temperature conditions of a nuclear weapon. That’s the tricky part.”

One way to circumvent this issue is to produce new pits. A major effort under way will ramp up production. In 2024, the NNSA “diamond stamped” the first of these pits, meaning that the pit was certified for use in a weapon. The aim is for the United States to make 80 pits per year by 2030. But questions remain about new plutonium pits as well, Hecker says, as they rely on an updated manufacturing process.

Hecker, whose tenure at Los Alamos straddled the testing and post-testing eras, thinks nuclear tests could help answer some of those questions. “Those people who say, ‘There is no scientific or technical reason to test. We can do it all with computers,’ I disagree strongly.”

But, he says, the benefits of performing a test would be outweighed by the big drawback: Other countries would likely return to testing. And those countries would have more to learn than the United States. China, for instance, has performed only 45 tests, while the United States has performed over 1,000. “We have to find other ways that we can reassure ourselves,” Hecker says.

Other experts similarly thread the needle. Nuclear tests of the past produced plenty of surprises, such as yields that were higher or lower than predicted, physicist Michael Frankel, an independent scientific consultant, and colleagues argued in a 2021 report. While the researchers advise against resuming testing in the current situation, they expect that stockpile stewardship will not be sufficient indefinitely. “Too many things have gone too wrong too often to trust Lucy with the football one more time,” Frankel and colleagues wrote, referring to Charles Schulz’s comic strip Peanuts. If we rely too much on computer simulations to conclude an untested nuclear weapon will work, we might find ourselves like Charlie Brown — flat on our backs.

But other scientists have full faith in subcritical experiments and stockpile stewardship. “We have always found that there are better ways to answer these questions than to return to nuclear explosive testing,” Adams says.

What counts as a nuclear weapons test?

For many scientists, subcritical experiments are preferable, especially given the political ramifications of full-fledged tests. But the line between a nuclear test prohibited by the Comprehensive Nuclear-Test-Ban Treaty and an experiment that is allowed is not always clear.

The CTBT is a “zero yield” treaty; experiments can release no energy beyond that produced by the chemical explosives. But, Adams says, “there’s no such thing as zero yield.” Even in an idle, isolated hunk of plutonium, some nuclear fission happens spontaneously. That’s a nonzero but tiny nuclear yield. “It’s a ridiculous term,” he says. “I hate it. I wish no one had ever said it.”

The United States has taken zero yield to mean that self-sustaining chain reactions are prohibited. U.S. government reports claim that Russia has performed nuclear experiments that surpass this definition of the zero yield benchmark and raise concerns about China’s adherence to the standard. The confusion has caused finger-pointing and increased tensions.

But countries might honestly disagree on the definition of a nuclear test, Adams says. For example, a country might allow “hydronuclear” experiments, which are supercritical but the amount of fission energy released is dwarfed by the energy from the chemical explosive. Such experiments would violate U.S. standards, but perhaps not those of Russia or another country.

Even if everyone could agree on a definition, monitoring would be challenging. The CTBT provides for seismic and other monitoring, but detecting very-low-yield tests would demand new inspection techniques, such as measuring the radiation emanating from a confinement vessel used in an experiment.

Underground tests are not risk-free

Tests that clearly break the rules, however, can be swiftly detected. The CTBT monitoring system can spot underground explosions as small as 0.1 kilotons, less than a hundredth that of the bomb dropped on Hiroshima. That includes the most recent nuclear explosive test, performed by North Korea in 2017.

Despite being invisible, underground nuclear explosive tests have an impact. While an underground test is generally much safer than an open-air nuclear test, “it’s not not risky,” Park says.

The containment provided by an underground test isn’t assured. In the 1970 Baneberry test in Nevada, a misunderstanding of the site’s geology led to a radioactive plume escaping in a blowout that exposed workers on the site.

While U.S. scientists learned from that mistake and haven’t had such a major containment failure since, the incident suggests that performing an underground test in a rushed manner could increase the risks for an accident, Park says.

Hecker is not too concerned about that possibility. “For the most part, I have good confidence that we could do underground nuclear testing without a significant insult to the environment,” he says. “It’s not an automatic given.… Obviously there’s radioactive debris that stays down there. But I think enough work has been done to understand the geology that we don’t think there will be a major environmental problem.”

While the United States knows its test sites well and has practice with underground testing, “other countries might not be as knowledgeable,” Hruby says. So if the United States starts testing and others follow, “the chance of a non-containment, a leak of some kind, certainly goes up.” A U.S. test, she says, is “a very bad idea.”

Even if the initial containment is successful, radioactive materials could travel via groundwater. Although tests are designed to avoid groundwater, scientists have detected traces of plutonium in groundwater from the Nevada site. The plutonium traveled a little more than a kilometer in 30 years. “To a lot of people, that’s not very far,” Park says. But “from a geology time scale, that’s really fast.” Although not at a level where it would cause health effects, the plutonium had been expected to stay put.

The craters left in the Nevada desert are a mark of each test’s impact on structures deep below the surface. “There was a time when detonating either above ground or underground in the desert seemed like — well, that’s just wasteland,” Jeanloz says. “Many would view it very differently now, and say, ‘No, these are very fragile ecosystems, so perturbing the water table, putting radioactive debris, has serious consequences.’ ”

The weight of public opinion is another hurdle. In the days of nuclear testing, protests at the site were a regular occurrence. That opposition persisted to the very end. On the day of the Divider test in 1992, four protesters made it to within about six kilometers of ground zero before being arrested.

The disarmament movement continues despite the lack of testing. At a recent meeting of nuclear experts, the Nuclear Deterrence Summit in Arlington, Va., a few protesters gathered outside in the January cold, demanding that the United States and Russia swear off nuclear weapons for good. But that option was not on the meeting’s agenda. During a break between sessions, the song that played — presumably unintentionally — was “Never Gonna Give You Up.”