A fast-neutron reactor (FNR) or simply a fast reactor is a category of nuclear reactor in which the fission chain reaction is sustained by fast neutrons (carrying energies of 5 MeV or greater), as opposed to thermal neutrons used in thermal-neutron reactors. Such a reactor needs no neutron moderator, but requires fuel that is relatively rich in fissile material when compared to that required for a thermal-neutron reactor.
Natural uranium consists mostly of three isotopes: 238 U, 235 U, and trace quantities of 234 U, a decay product of 238U. 238U accounts for roughly 99.3% of natural uranium and undergoes fission only by fast neutrons. About 0.7% of natural uranium is 235U, which undergoes fission by neutrons of any energy, but particularly by lower-energy neutrons. When either of these isotopes undergoes fission, it releases neutrons with an energy distribution peaking around 1 to 2 MeV. The flux of higher-energy fission neutrons (> 2 MeV) is too low to create sufficient fission in 238U, and the flux of lower-energy fission neutrons (< 2 MeV) is too low to do so easily in 235U.
The common solution to this problem is to slow the neutrons using a neutron moderator, which interacts with the neutrons to slow them. The most common moderator is water, which acts by elastic scattering until the neutrons reach thermal equilibrium with the water. The key to reactor design is to carefully lay out the fuel and water so the neutrons have time to slow enough to become highly reactive with the 235U, but not so far as to allow them to escape the reactor core.
Although 238U does not undergo fission by the neutrons released in fission, thermal neutrons can be captured by the nucleus to transmute the uranium into 239 Pu. 239Pu has a neutron cross section similar to that of 235U, and most of the atoms created this way will undergo fission from the thermal neutrons. In most reactors this accounts for as much as ⅓ of generated energy. Some 239Pu remains, and the leftover, along with leftover 238U, can be recycled during nuclear reprocessing.
Water has disadvantages as a moderator. It can absorb a neutron and remove it from the reaction. It does this just enough that the concentration of 235U in natural uranium is too low to sustain the chain reaction; the neutrons lost through absorption in the water and 238U, along with those lost to the environment, results in too few left in the fuel. The most common solution to this problem is to slightly concentrate the amount of 235U in the fuel to produce enriched uranium, with the leftover 238U known as depleted uranium. Other designs use different moderators, like heavy water, that are much less likely to absorb neutrons, allowing them to run on unenriched fuel. In either case, the reactor's neutron economy is based on thermal neutrons.
Although 235U and 239Pu are less sensitive to higher-energy neutrons, they still remain somewhat reactive well into the MeV range. If the fuel is enriched, eventually a threshold will be reached where there are enough fissile atoms in the fuel to maintain a chain reaction even with fast neutrons.
The primary advantage is that by removing the moderator, the size of the reactor can be greatly reduced, and to some extent the complexity. This was commonly used for many early submarine reactor systems, where size and weight are major concerns. The downside to the fast reaction is that fuel enrichment is an expensive process, so this is generally not suitable for electrical generation or other roles where cost is more important than size.
Another advantage to the fast reaction has led to considerable development for civilian use. Fast reactors lack a moderator, and thus lack one of the systems that remove neutrons from the system. Those running on 239Pu further increase the number of neutrons, because its most common fission cycle gives off three neutrons rather than the mix of two and three neutrons released from 235U. By surrounding the reactor core with a moderator and then a layer (blanket) of 238U, those neutrons can be captured and used to breed more 239Pu. This is the same reaction that occurs internally in conventional designs, but in this case the blanket does not have to sustain a reaction and thus can be made of natural uranium or depleted uranium.
Due to the surplus of neutrons from 239Pu fission, the reactor produces more 239Pu than it consumes. The blanket material can then be processed to extract the 239Pu to replace losses in the reactor, and the surplus is then mixed with uranium to produce MOX fuel that can be fed into conventional slow-neutron reactors. A single fast reactor can thereby feed several slow ones, greatly increasing the amount of energy extracted from the natural uranium, from less than 1% in a normal once-through cycle, to as much as 60% in the best fast reactor cycles.
Given the limited reserves of uranium ore, and the rate that nuclear power was expected to take over baseload generation, through the 1960s and 1970s fast breeder reactors were considered to be the solution to the world's energy needs. Using twice-through processing, a fast breeder increases the energy capacity of known ore deposits by as much as 100 times, meaning that existing ore sources would last hundreds of years. The disadvantage to this approach is that the breeder reactor has to be fed expensive, highly-enriched fuel. It was widely expected that this would still be below the price of enriched uranium as demand increased and known resources dwindled.
Through the 1970s, experimental breeder designs were examined, especially in the US, France and the USSR. However, this coincided with a crash in uranium prices. The expected increased demand led mining companies to expand supply channels, which came online just as the rate of reactor construction stalled in the mid-1970s. The resulting oversupply caused fuel prices to decline from about US$40 per pound in 1980 to less than $20 by 1984. Breeders produced fuel that was much more expensive, on the order of $100 to $160, and the few units that reached commercial operation proved to be economically disastrous. Interest in breeder reactors were further muted by Jimmy Carter's April 1977 decision to defer construction of breeders in the US due to proliferation concerns, and the terrible operating record of France's Superphénix reactor.
Fast-neutron reactors can reduce the total radiotoxicity of nuclear waste  using all or almost all of the waste as fuel. With fast neutrons, the ratio between splitting and the capture of neutrons by plutonium and the minor actinides is often larger than when the neutrons are slower, at thermal or near-thermal "epithermal" speeds. The transmuted even-numbered actinides (e.g. 240Pu, 242Pu) split nearly as easily as odd-numbered actinides in fast reactors. After they split, the actinides become a pair of "fission products". These elements have less total radiotoxicity. Since disposal of the fission products is dominated by the most radiotoxic fission product, caesium-137, which has a half life of 30.1 years, the result is to reduce nuclear waste lifetimes from tens of millennia (from transuranic isotopes) to a few centuries. The processes are not perfect, but the remaining transuranics are reduced from a significant problem to a tiny percentage of the total waste, because most transuranics can be used as fuel.
Fast reactors technically solve the "fuel shortage" argument against uranium-fueled reactors without assuming undiscovered reserves, or extraction from dilute sources such as granite or seawater. They permit nuclear fuels to be bred from almost all the actinides, including known, abundant sources of depleted uranium and thorium, and light-water reactor wastes. On average, more neutrons per fission are produced by fast neutrons than from thermal neutrons. This results in a larger surplus of neutrons beyond those required to sustain the chain reaction. These neutrons can be used to produce extra fuel, or to transmute long half-life waste to less troublesome isotopes, as was done at the Phénix reactor in Marcoule, France, or some can be used for each purpose. Though conventional thermal reactors also produce excess neutrons, fast reactors can produce enough of them to breed more fuel than they consume. Such designs are known as fast breeder reactors.
The main disadvantage of fast-neutron reactors is that they are costly to build and operate, and are not likely to be cost-competitive with thermal-neutron reactors unless the price of uranium increases dramatically.
Some other disadvantages are specific to some designs. Sodium is often used as a coolant in fast reactors, because it does not moderate neutron speeds much and has a high heat capacity. However, it burns and foams in air. It has caused difficulties in reactors (e.g. USS Seawolf (SSN-575), Monju), although some sodium-cooled fast reactors have operated safely for long periods (notably the Superphénix and EBR-II for 30 years).
Another problem is related to neutron activation. Since liquid metals other than lithium and beryllium have low moderating ability, the primary interaction of neutrons with fast reactor coolant is the (n,gamma) reaction, which induces radioactivity in the coolant. Neutron irradiation activates a significant fraction of coolant in high-power fast reactors, up to around a terabecquerel of beta decays per kilogram of coolant in steady operation.
Some fast reactors also have positive void coefficient: boiling of the coolant in an accident would reduce coolant density and thus the absorption rate. This is dangerous and undesirable from a safety and accident standpoint. This can be avoided with a gas-cooled reactor, since voids do not form in such a reactor during an accident; however, activation in the coolant remains a problem. A helium-cooled reactor would avoid this, since the elastic scattering and total cross sections are approximately equal, i.e. few (n,gamma) reactions are present in the coolant and the low density of helium at typical operating conditions means that neutrons have few interactions with coolant.
Due to the low cross sections of most materials at high neutron energies, critical mass in a fast reactor is much higher than in a thermal reactor. In practice, this means significantly higher enrichment: >20% enrichment in a fast reactor compared to <5% enrichment in typical thermal reactors. This raises nuclear proliferation and nuclear security issues.
Water, the most common coolant in thermal reactors, is generally not feasible for a fast reactor, because it acts as a neutron moderator. However the Generation IV reactor known as the supercritical water reactor with decreased coolant density may reach a hard enough neutron spectrum to be considered a fast reactor. Breeding, which is the primary advantage of fast over thermal reactors, may be accomplished with a thermal, light-water cooled and moderated system using uranium enriched to ~90%.
All operating fast reactors are liquid metal cooled reactors. The early Clementine reactor used mercury coolant and plutonium metal fuel. In addition to its toxicity to humans, mercury has a high cross section for the (n,gamma) reaction, causing activation in the coolant and losing neutrons that could otherwise be absorbed in the fuel, which is why it is no longer considered as a coolant. Molten lead has been used in naval propulsion units as well as some prototype reactors. Sodium-potassium alloy (NaK) is popular in test reactors due to its low melting point. All large-scale fast reactors have used molten sodium coolant.
Another proposed fast reactor is a molten salt reactor, in which the salt's moderating properties are insignificant. This is typically achieved by replacing the light metal fluorides (e.g. lithium fluoride - LiF, beryllium fluoride - BeF2) in the salt carrier with heavier metal chlorides (e.g., potassium chloride - KCI, rubidium chloride - RbCl, zirconium chloride - ZrCl4). Moltex Energy proposes to build a fast-neutron reactor called the Stable Salt Reactor. In this reactor design the nuclear fuel is dissolved in a molten salt. The salt is contained in stainless steel tubes similar to those use in solid fuel reactors. The reactor is cooled using the natural convection of another molten salt coolant. Moltex claims that their design is less expensive to build than a coal-fired power plant and can consume nuclear waste from conventional solid fuel reactors.
Gas-cooled fast reactors have been the subject of research commonly using helium, which has small absorption and scattering cross sections, thus preserving the fast neutron spectrum without significant neutron absorption in the coolant.
In practice, sustaining a fission chain reaction with fast neutrons means using relatively enriched uranium or plutonium. The reason for this is that fissile reactions are favored at thermal energies, since the ratio between the 239Pu fission cross section and 238U absorption cross section is ~100 in a thermal spectrum and 8 in a fast spectrum. Fission and absorption cross sections are low for both 239Pu and 238U at high (fast) energies, which means that fast neutrons are likelier to pass through fuel without interacting than thermal neutrons; thus, more fissile material is needed. Therefore a fast reactor cannot run on natural uranium fuel. However, it is possible to build a fast reactor that breeds fuel by producing more than it consumes. After the initial fuel charge such a reactor can be refueled by reprocessing. Fission products can be replaced by adding natural or even depleted uranium without further enrichment. This is the concept of the fast breeder reactor or FBR.
So far, most fast-neutron reactors have used either MOX (mixed oxide) or metal alloy fuel. Soviet fast-neutron reactors use (high 235U enriched) uranium fuel. The Indian prototype reactor uses uranium-carbide fuel.
While criticality at fast energies may be achieved with uranium enriched to 5.5 (weight) percent uranium-235, fast reactor designs have been proposed with enrichments in the range of 20 percent for reasons including core lifetime: if a fast reactor were loaded with the minimal critical mass, then the reactor would become subcritical after the first fission. Rather, an excess of fuel is inserted with reactivity control mechanisms, such that the reactivity control is inserted fully at the beginning of life to bring the reactor from supercritical to critical; as the fuel is depleted, the reactivity control is withdrawn to support continuing fission. In a fast breeder reactor, the above applies, though the reactivity from fuel depletion is also compensated by breeding either 233U or 239Pu and 241Pu from thorium-232 or 238U, respectively.
Like thermal reactors, fast-neutron reactors are controlled by keeping the criticality of the reactor reliant on delayed neutrons, with gross control from neutron-absorbing control rods or blades.
They cannot, however, rely on changes to their moderators because there is no moderator. So Doppler broadening in the moderator, which affects thermal neutrons, does not work, nor does a negative void coefficient of the moderator. Both techniques are common in ordinary light-water reactors.
Doppler broadening from the molecular motion of the fuel, from its heat, can provide rapid negative feedback. The molecular movement of the fissionables themselves can tune the fuel's relative speed away from the optimal neutron speed. Thermal expansion of the fuel can provide negative feedback. Small reactors as in submarines may use Doppler broadening or thermal expansion of neutron reflectors.
A 2008 IAEA proposal for a Fast Reactor Knowledge Preservation System noted that:
List of fast reactors
- Clementine was the first fast reactor, built in 1946 at Los Alamos National Laboratory. It used plutonium metal fuel, mercury coolant, achieved 25 kW thermal and used for research, especially as a fast neutron source.
- Experimental Breeder Reactor I (EBR-I) at Idaho Falls, in 1951 became the first reactor to generate significant amounts of power. Decommissioned in 1964.
- Fermi 1 near Detroit was a prototype fast breeder reactor that powered up in 1957 and shut down in 1972.
- Experimental Breeder Reactor II (EBR-II) was a prototype for the Integral Fast Reactor, 1965–1994.
- SEFOR in Arkansas, was a 20 MWt research reactor that operated from 1969 to 1972.
- Fast Flux Test Facility (FFTF), 400 MWt, operated flawlessly from 1982 to 1992, at Hanford Washington. It used liquid sodium drained with argon backfill under care and maintenance.
- SRE in California, was a 20 MWt, 6.5 MWe commercial reactor operated from 1957 to 1964.
- LAMPRE-1 was a molten plutonium fueled 1 MWth reactor. It operated as a research reactor from 1961-1963 at Los Alamos national Lab.
- Dounreay Loop type Fast Reactor (DFR), 1959–1977, was a 14 MWe and Prototype Fast Reactor (PFR), 1974–1994, 250 MWe, in Caithness, in the Highland area of Scotland.
- Dounreay Pool type Fast Reactor (PFR), 1975–1994, was a 600 MWt, 234 MWe which used mixed oxide (MOX) fuel.
- Rapsodie in Cadarache, France, (20 then 40 MW) operated between 1967 and 1982.
- Superphénix, in France, 1200 MWe, closed in 1997 due to a political decision and high costs.
- Phénix, 1973, France, 233 MWe, restarted 2003 at 140 MWe for experiments on transmutation of nuclear waste for six years, ceased power generation in March 2009, though it will continue in test operation and to continue research programs by CEA until the end of 2009. Stopped in 2010.
- KNK-II, in Germany a 21 MWe experimental compact sodium-cooled fast reactor operated from Oct 1977-Aug 1991. The objective of the experiment was to eliminate nuclear waste while producing energy. There were minor sodium problems combined with public protests which resulted in the closure of the facility.
- Small lead-cooled fast reactors were used for naval propulsion, particularly by the Soviet Navy.
- BR-5 - was a research-focused fast-neutron reactor at the Institute of Physics and Energy in Obninsk from 1959-2002.
- BN-350 was constructed by the Soviet Union in Shevchenko (today's Aqtau) on the Caspian Sea, It produced 130 MWe plus 80,000 tons of fresh water per day.
- IBR-2 - was a research focused fast-neutron reactor at the Joint Institute of Nuclear Research in Dubna (near Moscow).
- RORSATs - 33 space fast reactors were launched by the Soviet Union from 1989-1990 as part of a program known as the Radar Ocean Reconnaissance Satellite (RORSAT) in the US. Typically, the reactors produced approximately 3 kWe.
- BES-5 - was a sodium cooled space reactor launched as part of the RORSAT program which produced 5 kWe.
- BR-5 - was a 5 MWt sodium fast reactor operated by the USSR in 1961 primarily for materials testing.
- Russian Alpha 8 PbBi - was a series of lead bismuth cooled fast reactors used aboard submarines. The submarines functioned as killer submarines, staying in harbor then attacking due to the high speeds achievable by the sub.
- Monju reactor, 300 MWe, in Japan, was closed in 1995 following a serious sodium leak and fire. It was restarted on May 6, 2010 but in August 2010 another accident, involving dropped machinery, shut down the reactor again. As of June 2011, the reactor had generated electricity for only one hour since its first test two decades prior.
- Aktau Reactor, 150 MWe, in Kazakhstan, was used for plutonium production, desalination, and electricity. It closed 4 years after the plant's operating license expired.
- Clinch River Breeder Reactor, United States
- Integral Fast Reactor, United States. Design emphasized fuel cycle based on on-site electrolytic reprocessing. Cancelled in 1994 without construction.
- SNR-300, Germany
- BN-600 - a pool type sodium-cooled fast breeder reactor at the Beloyarsk Nuclear Power Station. It provides 560 MWe to the Middle Urals power grid. In operation since 1980.
- BN-800 - a sodium-cooled fast breeder reactor at the Beloyarsk Nuclear Power Station. It generates 880 MW of electrical power and started producing electricity in October, 2014. It reached full power in August, 2016.
- BOR-60 - a sodium-cooled reactor at the Research Institute of Atomic Reactors in Dmitrovgrad. In operation since 1968. It produces 60MW for experimental purposes.
- FBTR - a 10.5 MW experimental reactor in India which focused on reaching significant burnup levels.
- China Experimental Fast Reactor, a 60 MWth, 20 MWe, experimental reactor which went critical in 2011 and is currently operational. It is used for materials and component research for future Chinese fast reactors.
- KiloPower/KRUSTY is a 1-10 kWe research sodium fast reactor built at Los Alamos National Laboratory. It first reach criticality in 2015 and demonstrates an application of a Stirling power cycle.
- Jōyō (常陽), 1977–1997 and 2004–2007, Japan, 140 MWt is an experimental reactor, operated as an irradiation test facility. After an incident in 2007, the reactor was suspended for repairing, recoworks were planned to be completed in 2014.
- PFBR, Kalpakkam, India, 500 MWe reactor with criticality planned for 2019. It is a sodium fast breeder reactor.
- CFR-600, China, 600 MWe.
- BN-1200, Russia, built starting after 2014, with operation planned for 2018–2020.
- Toshiba 4S was planned to be shipped to Galena, Alaska (USA) but progress stalled (see Galena Nuclear Power Plant)
- KALIME is a 600 MWe project in South Korea, projected for 2030. KALIMER is a continuation of the sodium-cooled, metal-fueled, fast-neutron reactor in a pool represented by the Advanced Burner Reactor (2006), S-PRISM (1998-present), Integral Fast Reactor (1984-1994), and EBR-II (1965-1995).
- Generation IV reactor (helium·sodium·lead cooled) US-proposed international effort, after 2030.
- JSFR, Japan, a project for a 1500 MWe reactor began in 1998, but without success.
- ASTRID, France, project for a 600 MWe sodium-cooled reactor. Experimental operation is planned for 2020.
- Mars Atmospherically Cooled Reactor (MACR) is a 1 MWe project, planned to complete in 2033. MACR is a gas-cooled (carbon dioxide coolant) fast-neutron reactor intended to provide power to proposed Mars colonies.
- TerraPower is designing a molten salt reactor in partnership with Southern Company, Oak Ridge National Laboratory, Idaho National Laboratory, Vanderbilt University and the Electric Power Research Institute. They expect to begin testing a loop facility in 2019 and is scaling up their salt manufacturing process. Data will be used to assess thermal hydraulics and safety analysis codes.
- Elysium Industries is designing a fast spectrum molten salt reactor.
- ALFRED (Advanced Lead Fast Reactor European Demonstrator) is a lead cooled fast reactor demonstrator designed by Ansaldo Energia from Italy, it represents the last stage of the ELSY and LEADER projects.
- Aristos Power is designing a molten salt reactor cooled by molten lead.
- Future FBR, India, 600 MWe, after 2025