Fast-neutron reactor

• 1. Introduction

  ›

  The fast-neutron reactor began in 1946 at Los Alamos National Laboratory, with a machine called Clementine. It ran on plutonium metal fuel and produced only 25 kilowatts of heat. That was enough for research, especially as a source of fast neutrons.

  Fast neutrons are the heart of the matter. In a fast reactor, the fission chain reaction is sustained by neutrons carrying energies above 1 MeV on average. Nearly every commercial reactor does the opposite and deliberately slows its neutrons down.

  Why would anyone build a reactor that refuses to slow them? The answer touches three of nuclear power's hardest problems: a possible shortage of fuel, a stubborn legacy of waste, and the risk that plutonium could be diverted to weapons.

  It also raises a puzzle. Since Clementine, only 13 fast breeder reactors have been built for commercial power, alongside 65 fast-spectrum research reactors. If the idea is so promising, why has it stayed so rare?

  381016

• 2. The Ping Pong Problem

  ›

  Natural uranium is 99.3% uranium-238 and only 0.7% uranium-235. That scarce 0.7% is the isotope that fissions with both fast and slow neutrons.

  The trouble is that fission releases fast neutrons, and fast neutrons are poor at triggering the next fission. Slower neutrons are about 585 times more likely to do the job. So most reactor designers add a moderator, a material that slows the neutrons down. Ordinary water is the most common choice. Heavy water, beryllium and graphite are others.

  The source offers a neat comparison with ping pong balls. When a fast ball hits a stationary or slow one, both end up with about half of the fast ball's original energy. When a fast ball hits a bowling ball, it keeps virtually all of its energy.

  Water brings its own costs. Light water absorbs enough neutrons that natural uranium cannot sustain a chain reaction in it. The fuel must be enriched, and the leftover material is known as depleted uranium. Heavy water and graphite absorb far fewer neutrons, which lets reactors such as the CANDU run on natural uranium.

  Water also boils at a low temperature, and steam turbines work better at higher pressure and temperature. A water-cooled reactor therefore runs under high pressure, inside steel vessels as much as 30 centimetres, or 12 inches, thick. A modern pressurized water reactor runs near 350 Celsius at around 85 bar. Modern coal-fired steam circuits exceed 500 Celsius. As a result, only about 30-33% of a pressurized water reactor's heat becomes electricity.

  The third weakness appears after shutdown. Spent fuel still holds radioactive elements that generate substantial heat. If the water that cools it is lost, temperatures can pass the melting point of the zircaloy cladding. That is what happened in the multiple meltdowns of the Fukushima disaster.

  Fast reactor advocates say their designs answer every one of these complaints. They begin by throwing away the moderator entirely.

  45920

• 3. Breeding More Than Burning

  ›

  A ratio of 12 to 10 sums up the best real-world performance of fast breeders. They finish a fuel cycle holding 20% more fissile material than they started with. The theoretical best is 14 to 10.

  The trick lies in uranium-238, the abundant isotope that thermal reactors mostly ignore. In a fast spectrum, it can capture a neutron and become plutonium-239, which is fissile. When it captures a fast neutron, it also fissions about 11% of the time.

  Surround the core with a blanket of fertile material, and the surplus neutrons breed new fuel there. Remove the blanket and choose the core materials carefully, and the reactor only conversion-breeds, making fissile fuel one-for-one. The blanket material can be processed and mixed into MOX or REMIX fuel for conventional reactors. A single fast reactor can supply its own fuel indefinitely and feed several thermal ones.

  The payoff is in the numbers. In a thermal once-through cycle, less than 1% of the mined uranium is consumed. In the best existing fast reactor cycles, up to 60% of it is fissioned.

  There is a catch. Fast neutrons are likelier to pass straight through fuel without interacting. The ratio of fission to absorption is about 100 in a thermal spectrum and only 8 in a fast one. A fast reactor therefore needs far more fissile material, with enrichment above 20% compared with under 5% in typical thermal reactors. It cannot run on natural uranium alone.

  Most fast reactors have used either MOX or metal alloy fuel. Soviet reactors started on highly enriched uranium and, according to the source, switched to MOX in 2022. India's prototype uses uranium carbide.

  6781014

• 4. Shrinking the Waste

  ›

  Strontium-90 has a half-life of 28.8 years, and caesium-137 lasts 30.1 years. These two fission products dominate the problem of disposing of the waste that fast reactors do produce.

  The bigger burden in ordinary spent fuel is the transuranic elements, such as americium and curium. In a thermal reactor they accumulate. In a fast reactor they undergo fast fission, and even the even-numbered actinides split almost as easily as the odd-numbered ones. After splitting, they become pairs of fission products with less total radiotoxicity.

  The effect on timescales is dramatic. Waste lifetimes shrink from tens of millennia to a few centuries. A commercial-scale fast reactor would produce a little more than a ton of fission products per year, and the remainder should be stored for about 500 years. The process is not perfect, but most transuranics can be used as fuel instead of discarded.

  Reprocessing makes this possible. A series of chemical extraction steps, mostly based on the PUREX process, splits spent fuel into three streams. The first is unchanged uranium-238, which is the vast bulk and has very low radioactivity. The second is fission products. The third is the transuranic elements.

  The surplus of neutrons in a fast reactor can also be spent on transmutation, turning long-lived waste into less troublesome isotopes. Fast reactors can even draw on depleted uranium, thorium and light-water reactor wastes. That answers the fuel shortage argument without assuming undiscovered reserves or extraction from granite or seawater.

  The Phénix reactor at Marcoule in France tested the waste idea in practice. It was restarted in 2003 at 140 MWe for six years of experiments on transmuting nuclear waste.

  78102223

• 5. Metal in the Veins

  ›

  Mercury once flowed through Clementine's core, and it is the reason that metal is no longer considered useful. Mercury is toxic to humans. It also readily absorbs neutrons, which activates the coolant and wastes neutrons that could have gone into the fuel.

  Water is generally ruled out for fast reactors because it works as an effective moderator. Every operating fast reactor instead uses a liquid metal: sodium, lead, or a lead-bismuth eutectic. Russia developed lead and lead-bismuth reactors, which saw larger-scale use in naval propulsion, notably in the Soviet Alfa-class submarine. A sodium-potassium alloy is popular in test reactors because of its low melting point.

  Other coolants have been proposed. Helium and purified nitrogen-15 have very low neutron absorption. Molten salts can work too, but the salt formula is crucial, because some salts moderate neutrons and others do not.

  The case for molten metal rests on margins. Sodium boils at 883 Celsius, and lead boils at 1749 Celsius. Reactors typically run around 500 to 550 Celsius, so the coolant stays liquid with no rise in pressure. An inert gas blanket at about 0.5 atmospheres means any leak pushes material outward, with no high pressure jets. After shutdown, passive air cooling is possible without pumps. Hotter coolant also means hotter steam, and around 40% thermal efficiency.

  The drawbacks are real. Sodium burns and foams in air, and sodium leaks have caused difficulties at Monju and aboard the USS Seawolf (SSN-575). Neutron bombardment also creates sodium-24, which has a 15-hour half-life, so almost none remains after two weeks. A cold trap removes its decay product from the sodium.

  The sodium leaks do not release radioactive elements, because sodium fast reactors always use a two loop system. Some have run safely for long periods despite minor leaks and fires. Phénix and EBR-II ran for 30 years. A pure lead design could use a single loop, since pure lead has virtually no activation.

  Refuelling is awkward. A core full of opaque molten metal defeats cameras, so operators need costly, carefully calibrated robotic tools. The metal also melts well above room temperature, so all the tubing and the reactor volume must be heated electrically before nuclear operation can begin.

  Russia now has around eighty reactor years of experience with lead-cooled fast reactors, a design the source says is rapidly gaining interest.

  810212328

• 6. Cores That Switch Themselves Off

  ›

  In 1986, a demonstration at EBR-II tested what happens when a fast reactor gets too hot. The extra heat was readily absorbed by the large volume of liquid sodium. The reactor shut itself down without any operator intervention.

  Two effects make this possible. A rise in temperature causes Doppler broadening of the neutron spectrum, and it very slightly enlarges the physical core. Both let more neutrons escape the core, which lowers reactivity.

  Control works differently here than in moderated reactors. Both kinds rely on delayed neutrons, together with control rods or blades for gross control. But fast reactors have no moderator, so the techniques common in light-water reactors, such as a negative void coefficient of the moderator, do not apply.

  In a moderated reactor, delayed neutrons give operators or computers time to adjust reactivity. In a fast reactor they play virtually no role. Very short-term control, within one second, therefore falls to thermal expansion and Doppler broadening. Longer-term control comes from control rods filled with a neutron-absorbing material.

  There is one design pitfall. A defective fast reactor could have a positive void coefficient, where boiling coolant reduces density and thus the absorption rate. No such design has been proposed for commercial service. Gas cooling avoids voids, but then reactivity control becomes difficult.

  Small reactors, such as those in submarines, may use thermal expansion of neutron reflectors for this feedback.

  182181016

• 7. Boom and Bust

  ›

  Uranium cost about US$40 per pound in 1980, and by 1984 it had fallen below $20. That collapse helped sink the breeder dream.

  The dream had been strong. In the 1960s, estimates of uranium ore reserves were low, and nuclear power was expected to take over baseload generation. Through the 1960s and 1970s, fast breeders were seen as the solution to the world's energy needs. Twice-through processing would stretch known ore deposits over hundreds of years. Experimental breeder designs were examined, especially in the US, France and the USSR.

  Then the expected demand failed to arrive. Mining companies had expanded supply, and the supply came online just as reactor construction stalled in the mid-1970s. The oversupply drove prices down. Breeder fuel cost on the order of $100 to $160 per pound, and the few units that reached commercial operation proved economically unfeasible.

  Politics added to the damage. In April 1977, Jimmy Carter decided to defer construction of breeders in the US over proliferation concerns. Fast reactors are judged a higher proliferation risk because breeders require reprocessing, which can be redirected to produce weapons-grade plutonium. In France, the suboptimal operating record of Superphénix and serious opposition from environmentalist groups muted interest further.

  A 2008 IAEA proposal for a Fast Reactor Knowledge Preservation System described the aftermath. It noted stagnation over the previous 15 years. All studies had stopped in Germany, Italy, the United Kingdom and the United States, leaving only decommissioning work. Many specialists had retired or were close to retirement. Even in France, Japan and Russia, young scientists and engineers were not moving into the field.

  Even so, around 25 reactors have been built since the 1970s, adding up to over 400 reactor years of experience.

  222324810

• 8. A Global Roll Call

  ›

  Monju, a 300 MWe reactor in Japan, closed in 1995 after a serious sodium leak and fire. It restarted in May 2010, but another accident involving dropped machinery shut it down again that August. By June 2011, it had generated electricity for only one hour since its first test two decades earlier.

  The record elsewhere is mixed. Experimental Breeder Reactor I, near Arco, Idaho, became the first reactor to generate significant amounts of power in 1951. Fermi 1, near Detroit, powered up in 1957 and shut down in 1972. The Fast Flux Test Facility at Hanford, Washington, operated flawlessly from 1982 to 1992.

  Superphénix was the largest of all, designed to deliver 1,242 MWe. France closed it in 1997 because of a political decision and high costs. In Germany, the KNK-II reactor closed after minor sodium problems combined with public protests.

  Some fast reactors did more than make electricity. BN-350, built by the Soviet Union in Shevchenko, today's Aqtau, on the Caspian Sea, produced 130 MWe plus 80,000 tons of fresh water per day. Between 1989 and 1990, the Soviet Union launched 33 space fast reactors in the program the US called RORSAT. Each typically produced approximately 3 kWe.

  Russia keeps two big breeders at the Beloyarsk Nuclear Power Station. BN-600 has provided 560 MWe since 1980. BN-800 started producing electricity in October 2014 and reached full power in August 2016. At Dubna, the IBR-2 pulsed reactor, commissioned on the 10th of February 1984, is the only nuclear reactor in the world with a movable reflector.

  Asia is now where much of the new activity sits. China's experimental reactor went critical in 2011, and its 600 MWe CFR-600 is under construction. India's 500 MWe Prototype Fast Breeder Reactor achieved criticality in April 2026, after its prime minister attended the initial core loading in March 2024. Japan expects to restart its Joyo experimental reactor at the end of 2026.

  Some designs never ran at all. The Integral Fast Reactor in the United States was built around on-site electrolytic reprocessing and was cancelled in 1994 without construction. The GEN IV initiative has proposed six new reactor types, three of which would operate with a fast spectrum.

  The most distant proposal is the Mars Atmospherically Cooled Reactor. It would be a 1 MWe fast reactor cooled by carbon dioxide, planned to complete in 2033 to power proposed Mars colonies.

  1423272931