Methods to flag fake warheads without revealing state secrets. An algorithm that can more accurately pick out underground nuclear tests in seismic data. The refinement of detectors that can quickly and accurately measure weapons-grade nuclear material. Teaching nearly twice as many students as initially planned.
These are a few of the accomplishments of a $25 million project to help prevent the spread of nuclear weapons. And they’re only two years into the five-year effort.
A team of researchers and nuclear policy experts from 12 universities and nine national laboratories congregated in Ann Arbor on Oct. 19 and 20 to review their progress. The project, called the Consortium for Verification Technology (CVT), is funded by Department of Energy’s National Nuclear Security Administration. Its aim is to develop tools and strategies to monitor nations that have signed nuclear treaties, verifying that the treaties are being upheld.
“Nuclear technology needs to be controlled. I think that we have found that out the hard way,” said Sara Pozzi, a professor of nuclear engineering and radiological sciences at U-M and leader of the effort. “The difficulty is that there’s this dual use. There’s this peaceful use for generating electricity. We want people to have access to that. But then there are weapons programs.”
The project is developing technologies that can make measurements on nuclear warheads and nuclear waste to make sure that no nuclear material goes missing, and methods to analyze secret, underground nuclear weapon tests such as those by North Korea in January and September of this year.
Protecting classified information
“If you want to make verification work, there are two ways to approach it. The first is technology-driven: let engineers and scientists loose to improve detectors,” said Alexander Glaser of Princeton University, who is leader of the project’s policy thrust. “The second is mission-driven. What we try to do on the policy side is turn the question around and look at existing or possible future treaties and say, ‘OK, what are the gaps?’”
And one big gap was the need for a device that could protect state secrets while making measurements that an inspector could trust. Analyzing a warhead to ensure that enriched uranium and plutonium is not being plundered from old nuclear stockpiles to be sold or to make new, secret weapons is straightforward technologically, but it’s politically impossible. Conventional inspection and measurement techniques would reveal important aspects of that warhead’s classified design.
To solve this problem, two CVT teams, led by Glaser and R. Scott Kemp of the Massachusetts Institute of Technology, have borrowed a concept from cryptography. Their different systems can compare the inner workings of two nuclear warheads without ever electronically recording what those workings look like. This may be the only way that the nation under inspection can be sure that the inspector – or a hacker – has not made any secret copy of the weapon’s design.
Glaser’s team had published their concept in 2014, shortly before the CVT was announced, and CVT funds enabled them to bring it to life, comparing the patterns of neutrons absorbed by arrays of metal cubes. The neutrons that aren’t absorbed go on to make bubbles in the detectors behind the object tested.
In the inspection scenario, Glaser says, “the country under inspection would present a trusted warhead as a reference item, perhaps by removing this warhead from a missile with the inspector observing and taking it straight to the facility where the verification activities would take place.”
For the inspection, the challenge is to prove that a second warhead is identical to the first one, said Glaser. To do this, his team devised an interactive exercise using pairs of detectors that the host preloads prior to the inspection. For their experiment, Glaser’s team preloaded bubbles into the detectors, which were developed by Yale University.
The inspector then randomly assigns the preloaded detectors to the two warheads, which are then exposed to a neutron beam. This way, at the end of the test, both detectors show the same measurement – but only if the two warheads are identical. The test can be repeated until the inspector is satisfied. The experiment using the metal blocks and bubble detectors was published in September in Nature Communications.
In July, Kemp’s team published a concept for a comparing two warheads by sending high-precision X-rays through each warhead in the Proceedings of the National Academy of Sciences. The X-rays are recorded in such a way that the image looks like random noise, even under sophisticated mathematical analysis, but the images for two identical warheads would be the same. Glaser and Kemp’s studies were covered in an article in the New Yorker, titled “The Virtues of Nuclear Ignorance.”
Proving out new detectors
For direct measurement of weapons-grade nuclear materials, Pozzi reported progress with the detectors developed by her group, which sense gamma rays and neutrons – two forms of radiation that can signify that a nuclear reaction has taken place. Detectors like these can be taken into nuclear weapon silos to measure the amount of weapons-grade uranium and plutonium present. They can also be placed in airports and seaports for scanning cargo.
To test their detectors, Pozzi’s group needs access to weapons-grade nuclear materials, available at just a few major laboratories around the world. Idaho National Laboratory, a member of the project, is one of them. It has 100-gram plates of plutonium that can be arrayed and measured with the detection system developed at U-M.
The team measured a container with different numbers of plates inside, looking specifically for pairs of neutrons appearing at the same time. When this happens, it’s likely that they both came from one plutonium nucleus spontaneously splitting into two smaller nuclei – called fissioning.
The team was then able to map how frequently these neutron pairs were detected to the amount of plutonium in the cask, making it possible to measure the amount of plutonium inside a closed container. Initial tests showed a prediction error of a few percent.
These detectors owe some of their success to work under Alfred Hero, the R. Jamison and Betty Williams Professor of Engineering in the department of electrical engineering and computer science at U-M. His team has developed algorithms that are good at differentiating between events that look very similar, but one is much rarer – and it’s the rare event that is most interesting.
Pozzi’s detectors often pick up gamma rays in addition to the neutrons. While gamma rays are also emitted by the plutonium, the scant neutrons give cleaner information about fissions, and the algorithms help tell them apart. They can also help pick apart seismic activity, telling the difference between common earthquakes and rare nuclear tests.
Monitoring secret nuclear tests
“North Korean events have influenced us a lot,” said Pozzi. “We have been able to take a look at some of the data that the detectors around the world, especially the ones close to North Korea, Japan and China have picked up from those events.”
Paul Richards at Columbia University led the studies comparing the suspected sizes of the bombs detonated by North Korea in 2016 and in earlier tests, such as those in 2009 and 2013, through seismic measurements.
“When we compare the magnitudes of the explosions, they have increased from test to test,” said Pozzi.
A team at U-M, led by Professor John Lee, examined measurements of radioactive xenon, a form of the noble gas produced by nuclear explosions. They produced a computer model of how the xenon would be expected to travel through the atmosphere, predicting the order in which different monitoring stations would detect the gas.
For the future
While the project has a strong focus on technology, it is also addressing the looming shortfall in qualified professionals who can serve as inspectors and continue to improve the nuclear nonproliferation methods and technologies. The grant proposal promised the education of 60 students over the course of the project, but Pozzi says that already, about 120 students have been involved.
The national labs are especially likely to feel the talent pinch as many nuclear scientists are reaching retirement age, and Pozzi stresses the importance of getting young researchers in to work with them so that the knowledge isn’t lost. Already, 47 students have interned with the national labs.
Over the next three years, members of the CVT will continue to refine the detection of nuclear materials, methods to compare nuclear warheads while protecting state secrets, and analyses of secret nuclear tests. New directions may include an assessment of emerging satellite imaging techniques based on constellations of small satellites (known as “cubesats” or “nanosats”), which have lower resolution than those currently in use, but could provide new imagery of points of interest every 24 hours rather than once a week or once a month.
“Combined with modern machine-learning techniques processing these massive amounts of data, this could open up qualitatively new opportunities for detecting undeclared nuclear activities,” said Glaser.
Other university partners are North Carolina State University, University of Hawaii, Pennsylvania State University, Duke University, University of Wisconsin, University of Florida, Oregon State University, Yale University, and University of Illinois at Urbana-Champaign.
Other laboratory partners include the Princeton Plasma Physics Laboratory and several National Laboratories, including Los Alamos, Lawrence Livermore, Sandia, Lawrence Berkeley, Oak Ridge and Pacific Northwest.
Ensuring the security of our society is a top priority for the U-M College of Engineering’s transformational campaign currently underway. Find out more about supporting the security of our future in the Victors for Michigan campaign.