Detectors similar to those used to hunt for sterile neutrinos could help guard against the extraction of plutonium-239 for nuclear weapons, writes Patrick Huber.
The first nuclear-weapons test shook the desert in New Mexico 75 years ago. Weeks later, Hiroshima and Nagasaki were obliterated. So far, these two Japanese cities have been the only ones to suffer such a fate. Neutrinos can help to ensure that no other city has to be added to this dreadful list.
At the height of the arms race between the US and the USSR, stockpiles of nuclear weapons exceeded 50,000 warheads, with the majority being thermonuclear designs vastly more destructive than the fission bombs used in World War II. Significant reductions in global nuclear stockpiles followed the end of the Cold War, but the US and Russia still have about 12,500 nuclear weapons in total, and the other seven nuclear-armed nations have about 1500. Today, the politics of non-proliferation is once again tense and unpredictable. New nuclear security challenges have appeared, often from unexpected actors, as a result of leadership changes on both sides of the table. Nuclear arms races and the dissolution of arms-control treaties have yet again become a real possibility. A regional nuclear war involving just 1% of the global arsenal would cause a massive loss of life, trigger climate effects leading to crop failures and jeopardise the food supply of a billion people. Until we achieve global disarmament, nuclear non-proliferation efforts and arms control are still the most effective tools for nuclear security.
Not a bang but a whimper
The story of the neutrino is closely tied to nuclear weapons. The first serious proposal to detect the particle hypothesised by Pauli, put forward by Clyde Cowan and Frederick Reines in the early 1950s, was to use a nuclear explosion as the source (see “Daring experiment” figure). Inverse beta decay, whereby an electron-antineutrino strikes a free proton and transforms it into a neutron and a positron, was to be the detection reaction. The proposal was approved in 1952 as an addition to an already planned atmospheric nuclear-weapons test. However, while preparing for this experiment, Cowan and Reines realised that by capturing the neutron on a cadmium nucleus, and observing the delayed coincidence between the positron and this neutron, they could use the lower, but steady flux of neutrinos from a nuclear reactor instead (see “First detection” figure). This technique is still used today, but with gadolinium or lithium in place of cadmium.
The P reactor at the Savannah River site at Oak Ridge National Laboratory, which had been built and used to make plutonium and tritium for nuclear weapons, eventually hosted the successful experiment to first detect the neutrino in 1956. Neutrino experiments testing the properties of the neutrino including oscillation searches continued there until 1988, when the P reactor was shut down.
Neutrinos are not produced in nuclear fission itself, but by the beta decays of neutron-rich fission fragments – on average about six per fission. In a typical reactor fuelled by natural uranium or low-enriched uranium, the reactor starts out with only uranium-235 as its fuel. During operation a significant number of neutrons are absorbed on uranium-238, which is far more abundant, leading to the formation of uranium-239, which after two beta decays becomes plutonium-239. Plutonium-239 eventually contributes to about 40% of the fissions, and hence energy production, in a commercial reactor. It is also the isotope used in nuclear weapons.
The dual-use nature of reactors is at the crux of nuclear non-proliferation. What distinguishes a plutonium-production reactor from a regular reactor producing electricity is whether it is operated in such a way that the plutonium can be taken out of the reactor core before it deteriorates and becomes difficult to use in weapons applications. A reactor with a low content of plutonium-239 makes more and higher energy neutrinos than one rich in plutonium-239.
Lev Mikaelyan and Alexander Borovoi, from the Kurchatov Institute in Moscow, realised that neutrino emissions can be used to infer the power and plutonium content of a reactor. In a series of trailblazing experiments at the Rovno nuclear power plant in the 1980s and early 1990s, their group demonstrated that a tonne-scale underground neutrino detector situated 10 to 20 metres from a reactor can indeed track its power and plutonium content.
The significant drawback of neutrino detectors in the 1980s was that they needed to be situated underground, beneath a substantial overburden of rock, to shield them from cosmic rays. This greatly limited potential deployment sites. There was a series of application-related experiments – notably the successful SONGS experiment conducted by researchers at Lawrence Livermore National Laboratory, which aimed to reduce cost and improve the robustness and remote operation of neutrino detectors – but all of these detectors still needed shielding.
From cadmium to gadolinium
Synergies with fundamental physics grew in the 1990s, when the evidence for neutrino oscillations was becoming impossible to ignore. With the range of potential oscillations frequencies narrowing, the Palo Verde and Chooz reactor experiments placed multi-tonne detectors about 1 km from nuclear reactors, and sought to measure the relatively small θ13 parameter of the neutrino mixing matrix, which expresses the mixing between electron neutrinos and the third neutrino mass eigenstate. Both experiments used large amounts of liquid organic scintillator doped with gadolinium. The goal was to tag antineutrino events by capturing the neutrons on gadolinium, rather than the cadmium used by Reines and Cowan. Gadolinium produces 8 MeV of gamma rays upon de-excitation after a neutron capture. As it has an enormous neutron-capture cross section, even small amounts greatly enhance an experiment’s ability to identify neutrons.
Eventually, neutrino oscillations became an accepted fact, redoubling the interest in measuring θ13. This resulted in three new experiments: Double Chooz in France, RENO in South Korea, and Daya Bay in China. Learning lessons from Palo Verde and Chooz, the experiments successfully measured θ13 more precisely than any other neutrino mixing parameter. A spin-off from the Double Chooz experiment was the Nucifer detector (see “Purpose driven” figure), which demonstrated the operation of a robust sub-tonne-scale detector designed with missions to monitor reactors in mind, in alignment with requirements formulated at a 2008 workshop held by the International Atomic Energy Agency (IAEA). However, Nucifer still needed a significant overburden.
In 2011, however, shortly before the experiments established that θ13 is not zero, fundamental research once again galvanised the development of detector technology for reactor monitoring. In the run-up to the Double Chooz experiment, a group at Saclay started to re-evaluate the predictions for reactor neutrino fluxes – then and now based on measurements at the Institut Laue-Langevin in the 1980s – and found to their surprise that the reactor flux prediction came out 6% higher than before. Given that all prior experiments were in agreement with the old flux predictions, neutrinos were missing. This “reactor-antineutrino anomaly” persists to this day. A sterile neutrino with a mass of about 1 eV would be a simple explanation. This mass range has been suggested by experiments with accelerator neutrinos, most notably LSND and MiniBooNE, though it conflicts with predictions that muon neutrinos should oscillate into such a sterile neutrino, which experiments such as MINOS+ have failed to confirm.
To directly observe the high-frequency oscillations of an eV-scale sterile neutrino you need to get within about 10 m of the reactor. At this distance, backgrounds from the operation of the reactor are often non-negligible, and no overburden is possible – the same conditions a detector on a safeguards mission would encounter.
From gadolinium to lithium
Around half a dozen experimental groups are chasing sterile neutrinos using small detectors close to reactors. Some of the most advanced designs use fine spatial segmentation to reject backgrounds, and replace gadolinium with lithium-6 as the nucleus to capture and tag neutrons. Lithium has the advantage that upon neutron capture it produces an alpha particle and a triton rather than a handful of photons, resulting in a very well localised tag. In a small detector this improves event containment and thus efficiency, and also helps constrain event topology.
Following the lithium and finely segmented technical paths, the PROSPECT collaboration and the CHANDLER collaboration (see “Rapid deployment” figure), in which I participate, independently reported the detection of a neutrino spectrum with minimal overburden and high detection efficiency in 2018. This is a major milestone in making non-proliferation applications a reality, since it is the first demonstration of the technology needed for tonne-scale detectors capable of monitoring the plutonium content of a nuclear reactor that could be universally deployed without the need for special site preparation.
The story of the neutrino is closely tied to nuclear weapons
The main difference between the two detectors is that PROSPECT, which reported its near-final sterile neutrino limit at the Neutrino 2020 conference, uses a traditional approach with liquid scintillator, whereas CHANDLER, currently an R&D project, uses plastic scintillator. The use of plastic scintillator allows the deployment time-frame to be shortened to less than 24 hours. On the other hand, liquid scintillator allows the exploitation of pulse-shape discrimination to reject cosmic-ray neutron backgrounds, allowing PROSPECT to achieve a much better signal-to-background ratio than any plastic detector to date. Active R&D is seeking to improve topological reconstruction in plastic detectors and imbue them with pulse-shape discrimination. In addition, a number of safeguard-specific detector R&D experiments have successfully detected reactor neutrinos using plastic scintillator in conjunction with gadolinium. In the UK, the VIDARR collaboration has seen neutrinos from the Wylfa reactor, and in Japan the PANDA collaboration successfully operated a truck-mounted detector.
In parallel to detector development, studies are being undertaken to understand how reactor monitoring with neutrinos would impact nuclear security and support non-proliferation objectives. Two very relevant situations being studied are the 2015 Iran Deal – the Joint Comprehensive Plan of Action (JCPOA) – and verification concepts for a future agreement with North Korea.
One of the sticking points in negotiating the 2015 Iran deal was the future of the IR-40 reactor, which was being constructed at Arak, an industrial city in central Iran. The IR-40 was planned to be a 40 MW reactor fuelled by natural uranium and moderated with heavy water, with a stated purpose of isotope production for medical and scientific use. The choice of fuel and moderator is interesting, as it meshes with Iranian capabilities and would serve the stated purpose well and be cost effective, since no uranium enrichment is needed. Equally, however, if one were to design a plutonium-production reactor for a nascent weapons programme, this combination would be one of the top choices: it does not require uranium enrichment, and with the stated reactor power would result in the annual production of about 10 kg of rather pure plutonium-239. This matches the critical mass of a bare plutonium-239 sphere, and it is known that as little as 4 kg can be used to make an effective nuclear explosive. Within the JCPOA it was eventually agreed that the IR-40 could be redesigned, down-rated in power to 20 MW and the new core fuelled with 3.7% enriched fuel, reducing the annual plutonium production by a factor of six.
A 10 to 20 tonne neutrino detector 20 m from the reactor would be able to measure its plutonium content with a precision of 1 to 2 kg. This would be particularly relevant in the so-called N-th month scenario, which models a potential crisis in Iran based on events in North Korea in June 1994. During the 1994 crisis, which risked precipitating war with the US, the nuclear reactor at Yongbyon was shut down, and enough spent fuel rods removed to make several bombs. IAEA protocols were sternly tested. The organisation’s conventional safeguards for operating reactors consist of containment and surveillance – seals, for example, to prevent the unnoticed opening of the reactor, and cameras to record the movement of fuel, most crucially during reactor shutdowns. In the N-th month scenario, the IR-40 reactor, in its pre-JCPOA configuration (40 MW, rather than the renegotiated power of 20 MW), runs under full safeguards for N–1 months. In month N, a planned reactor shutdown takes place. At this point the reactor would contain 8 kg of weapons-grade plutonium. For unspecified reasons the safeguards are then interrupted. In month N+1, the reactor is restarted and full safeguards are restored. The question is: are the 8 kg of plutonium still in the reactor core, or has the core been replaced with fresh fuel and the 8 kg of plutonium illicitly diverted?
The disruption of safeguards could either be due to equipment failure – a more frequent event than one might assume – or due to events in the political realm ranging from a minor unpleasantness to a full-throttle dash for a nuclear weapon. Distinguishing the two scenarios would be a matter of utmost urgency. According to an analysis including realistic backgrounds extrapolated from the PROSPECT results, this could be done in 8 to 12 weeks with a neutrino detector.
Neutrino detectors could be effective in addressing the safeguard challenges presented by advanced reactors
No conventional non-neutrino technologies can match this performance without shutting the reactor down and sampling a significant fraction of the highly radioactive fuel. The conventional approach would be extremely disruptive to reactor operations and would put inspectors and plant operators at risk of radiation exposure. Even if the host country were to agree in principle, developing a safe plan and having all sides agree on its feasibility would take months at the very least, creating dangerous ambiguity in the interim and giving hardliners on both sides time to push for an escalation of the crisis. The conventional approach would also be significantly more expensive than a neutrino detector.
New negotiating gambit
The June 1994 crisis at Yongbyon still overshadows negotiations with North Korea, since, as far as North Korea is concerned, it discredited the IAEA. Both during the crisis, and subsequently, international attempts at non-proliferation failed to prevent North Korea from acquiring nuclear weapons – its first nuclear-weapons test took place in 2006 – or even to constrain its progress towards a small-scale operational nuclear force. New approaches are therefore needed, and recent attempts by the US to achieve progress on this issue prompted an international group of about 20 neutrino experts from Europe, the US, Russia, South Korea, China and Japan to develop specific deployment scenarios for neutrino detectors at the Yongbyon nuclear complex.
The main concern is the 5 MWe reactor, which, though named for its electrical power, has a thermal power of 20 MW. This gas-cooled graphite-moderated reactor, fuelled with natural uranium, has been the source of all of North Korea’s plutonium. The specifics of this reactor, and in particular its fuel cladding, which makes prolonged wet-storage of irradiated fuel impossible, represent such a proliferation risk that anything but a monitored shutdown prior to a complete dismantling appears inappropriate. To safeguard against the regime reneging on such a deal, were it to be agreed, a relatively modest tonne-scale neutrino detector right outside the reactor building could detect a powering up of this reactor within a day.
North Korea is also constructing the Experimental Light Water Reactor at Yongbyon. A 150 MW water-moderated reactor running with low-enriched fuel, this reactor would not be particularly well suited to plutonium production. Its design is not dissimilar to much larger reactors used throughout the world to produce electricity, and it could help address the perennial lack of electricity that has limited the development and growth of the country’s economy. North Korea may wish to operate it indefinitely. A larger, 10 tonne neutrino detector could detect any irregularities during its refuelling – a tell-tale sign of a non-civilian use of the reactor – on a timescale of three months, which is within the goals set by the IAEA.
In a different scenario, wherein the goal would be to monitor a total shutdown of all reactors at Yongbyon, it would be feasible to bury a Daya-Bay-style 50 tonne single volume detector under the Yak-san, a mountain about 2 km outside of the perimeter of the nuclear installations (see “A different scenario” figure). The cost and deployment timescale would be more onerous than in the other scenarios.
In the case of longer distances between reactor and detector, detector masses must increase to compensate an inverse-square reduction in the reactor-neutrino flux. As cosmic-ray backgrounds remain constant, the detectors must be deployed deep underground, beneath an overburden of several 100 m of rock. To this end, the UK’s Science and Technology Facilities Council, the UK Atomic Weapons Establishment and the US Department of Energy, are funding the WATCHMAN collaboration to pursue the construction of a multi-kilo-tonne water-Cherenkov detector at the Boulby mine, 20 km from two reactors in Hartlepool, in the UK. The goal is to demonstrate the ability to monitor the operational status of the reactors, which have a combined power of 3000 MW. In a use-case context this would translate to excluding the operation of an undeclared 10 to 20 MW reactor within a radius of a few kilometres , but no safeguards scenario has emerged where this would give a unique advantage.
Inverse-square scaling eventually breaks down around 100 km, as at that distance the backgrounds caused by civilian reactors far outshine any undeclared small reactor almost anywhere in the northern hemisphere. Small signals also prevent the use of neutrino detectors for nuclear-explosion monitoring, or to confirm the origin of a suspicious seismic event as being nuclear, as conventional technologies are more feasible than the very large detectors that would be needed. A more promising future application of neutrino-detector technology is to meet the new challenges posed by advanced nuclear-reactor designs.
The current safeguards regime relies on two key assumptions: that fuel comes in large, indivisible and individually identifiable units called “fuel assemblies”, and that power reactors need to be refuelled frequently. Most advanced reactor designs violate at least one of these design characteristics. Fuel may come in thousands of small pebbles or be molten, and its coolant may not be transparent, in contrast to current designs, where water is used as moderator, coolant and storage medium in the first years after discharge. Either way, counting and identification of the fuel by serial number may be impossible. And unlike current power reactors, which are refuelled on a 12-to-18-month cycle, allowing in-core fuel to be verified as well, advanced reactors may be refuelled only once in their lifetime.
Neutrino detectors would not be hampered by any of these novel features. Detailed simulations indicate that they could be effective in addressing the safeguard challenges presented by advanced reactors. Crucially, they would work in a very similar fashion for any of the new reactor designs.
In 2019 the US Department of Energy chartered and funded a study (which I co-chair) with the goal of determining the utility of the unique capabilities offered by neutrino detectors for nuclear security and energy applications. This study includes investigators from US national laboratories and academia more broadly, and will engage and interview nuclear security and policy experts within the Department of Energy, the State Department, NGOs, academia, and international agencies such as the IAEA. The results are expected early in 2021. They should provide a good understanding of where neutrinos can play a role in current and future monitoring and verification agreements, and may help to guide neutrino detectors towards their first real-world applications.
The idea of using neutrinos to monitor reactors has been around for about 40 years. Only very recently, however, as a result of a surge of interest in sterile neutrinos, has detector technology become available that would be practical in real-world scenarios such as the JCPOA or a new North Korean nuclear agreement. The most likely initial application will be near-field reactor monitoring with detectors inside the fence of the monitored facility as part of a regional nuclear deal. Such detectors will not be a panacea to all verification and monitoring needs, and can only be effective if there is a sincere political will on both sides, but they do offer more room for creative diplomacy, and a technology that is robust against the kinds of political failures which have derailed past agreements.
J Ashenfelter et al. 2018 Phys. Rev. Lett. 121 251802.
A Bernstein et al. 2020 Rev. Mod. Phys. 92 011003.
A A Borovoi et al. 1978 Soviet Atomic Energy 44 589–592.
R Carr et al. 2019 Science & Global Security 27 15–28.
A Haghighat et al. 2020 Phys. Rev. Appl. 13 034028.
C Stewart et al. 2019 Nat. Comm. 10 3527.