How a research reactor works

A research reactor is a neutron source. It fissions uranium like any other, but the heat is a byproduct dumped to the air or the sea: what is used are the neutrons leaving the core, to irradiate targets and produce radioisotopes, to test materials and fuels under an intense flux, to do neutron scattering in physics, chemistry and biology experiments, to dope silicon, analyze samples by activation and train operators and students. That is why the power is small: the IAEA database lists about 228 in operation in 54 countries, and all of them together add up to less than 3,000 MW thermal, what a single power reactor produces. They range from zero-power facilities that barely warm up to the 100 MW of PIK in Saint Petersburg, BR2 in Belgium or the Jules Horowitz reactor under construction in France.

What sets them apart in engineering terms is power density: the goal is the highest neutron flux in the smallest volume, with cores of tens of liters reaching 17 kW per cubic centimeter against 5 in a PWR. That is achieved with thin aluminum plate fuel, plenty of cooling surface and a good reflector. The flip side is that they run at atmospheric pressure and below 50 °C, with no steam and no turbine, and most fit in a university building. The first in Latin America was Argentina's RA-1, critical in 1958; today Argentina operates the RA-3, the RA-6 and others, has exported reactors to Peru, Algeria, Egypt and Australia through INVAP, and is finishing the RA-10 at Ezeiza.

The largest family is the open pool, the layout on this page: the core hangs or sits at the bottom of a 10- to 13-meter pool of light water that serves as coolant, moderator and shielding all at once, and can be seen from above. It is the MTR type, Materials Testing Reactor, with plate fuel of uranium silicide or aluminide enriched below 20%: OPAL in Australia, the RA-3 and RA-10 in Argentina, Maria in Poland or SAFARI-1 in South Africa. When power rises and more flow is needed, the core is enclosed in a tank inside the pool with forced circulation at a slight overpressure: that is the tank-in-pool of HFR Petten and BR2.

General Atomics' TRIGA is a family of its own because of its fuel, uranium zirconium hydride in rods, which gives it a unique inherent safety and lets it pulse: 66 have been installed in 24 countries, many at universities. And at the top end are the high-flux reactors, built to push the thermal flux above 10¹⁵ neutrons per square centimeter per second: the ILL in Grenoble, with a single annular fuel element and heavy water, and HFIR at Oak Ridge, with a beryllium reflector. They serve communities of thousands of scientists who book beam time as on a telescope.

The pool solves three problems with one material. Light water moderates the neutrons, removes the heat and, with ten meters of column, shields the core's gamma radiation enough for an operator to stand on the pool bridge during operation. It is also the working medium: fuel assemblies, targets and samples are moved under water with long tools from the bridge, with no refueling machines or casks. The water is kept demineralized and below 50 °C, and the pool is concrete with a stainless steel liner.

Looking into an operating pool means seeing an intense blue glow around the core: Cherenkov radiation. Fission products emit electrons and gamma rays that in turn knock electrons out of the water, many of them faster than the speed of light in water, which is 25% lower than in vacuum. A charged particle exceeding the speed of light in a medium emits a cone of light, the optical equivalent of a sonic boom, concentrated in the blue and ultraviolet. The brightness follows the power, and in a freshly shut-down reactor it persists, fainter, from the decay of the fission products.

OPAL's core, and the RA-10's, is a 4 × 4 array of fuel assemblies about 8 centimeters across and 1 meter long, each made of some twenty plates of uranium silicide in aluminum with 2-millimeter water channels between them. All the fission happens in a volume that would fit in a refrigerator. Around it, a zirconium tank filled with heavy water acts as reflector: it returns escaping neutrons to the core, flattens the flux and creates a zone of high, accessible thermal flux where the irradiation positions and the ends of the beam tubes go. Heavy water absorbs very few neutrons, so the flux in the reflector can exceed that in the core itself.

Power is governed with control plates of hafnium or a silver-indium-cadmium alloy that slide between the fuel assemblies. In INVAP's designs they enter from below, with the drives in a room under the pool, leaving the top free for operations. A second, independent and diverse shutdown system acts on the reflector: valves partially drain the heavy-water tank in seconds and the reactor shuts down by neutron leakage even if the plates do not move. Each operating cycle lasts 30 to 35 days; at the end a few assemblies are replaced and go to an adjacent storage pool.

There is no steam generator and no turbine: the circuit is a heat exchanger, period. Primary pumps draw water from the pool, push it through the core at high speed, upward in OPAL and the RA-10, and send it to plate heat exchangers outside the pool; the water returns to the pool below the core. A secondary circuit, kept separate so that the primary's activity never leaves the building, carries the heat to mechanical-draft cooling towers or to the sea. At 20 or 30 MW the flow rates are an Olympic pool every few minutes, yet the water temperature rises only a few degrees.

That separation between the circuit and the pool is the key to safety. If the pumps stop, flap valves that the pumps held closed fall under their own weight and connect the core to the pool: the decay heat, which in a 20 MW reactor is 1.3 MW at shutdown and about 300 kW an hour later, is removed by natural circulation in the hundreds of tonnes of pool water, with no electricity and no intervention. The core stays under ten meters of water and the pool takes days to heat up. It is the same principle as the passive systems power reactors adopted after Fukushima, but here it has been inherent to the design for sixty years.

The reflector is where the work gets done. Inside the heavy-water tank there are dozens of irradiation positions: vertical tubes into which capsules with targets are lowered to produce radioisotopes, or samples are irradiated for neutron activation analysis, or silicon ingots up to 20 centimeters in diameter are placed for doping. Some positions are pneumatic, with capsules that travel in seconds through a tube to a laboratory or a hot cell for short-lived isotopes. Others are test loops, in which a length of fuel or material is exposed for months to the flux and to the pressure and temperature conditions of a power reactor, to qualify it before use.

Beam tubes are horizontal, evacuated tubes that point tangentially at the reflector and let neutrons out toward a guide hall. Before leaving, they pass through a cold source, a vessel of liquid deuterium at about 25 kelvin that slows the neutrons to long wavelengths, useful for seeing large structures such as polymers and proteins, or through a hot graphite source at 2,000 K that does the opposite. Coated glass guides carry them tens of meters to diffractometers, spectrometers, reflectometers and small-angle scattering or imaging instruments. Because they carry no charge and see nuclei, neutrons pick out hydrogen, detect magnetism and pass through centimeters of metal: they are the complement to X-rays.

More than 40 million nuclear medicine procedures are performed worldwide each year, and 80% use technetium-99m. Technetium-99m has a six-hour half-life, so it cannot be stockpiled: it is obtained from its parent, molybdenum-99, with a 66-hour half-life, which is shipped to hospitals in generators, alumina columns from which technetium is eluted daily for a week. And nearly all molybdenum-99 is made in half a dozen research reactors: low-enriched uranium targets are irradiated for about six days in the reflector, taken to hot cells, dissolved and the molybdenum purified, leaving the site within hours for a generator plant. The whole chain runs against the clock of the half-lives.

BR2 in Belgium, HFR Petten in the Netherlands, Maria in Poland, LVR-15 in Czechia, SAFARI-1 in South Africa, OPAL in Australia and the RA-3 in Argentina are the main suppliers; OPAL covers a quarter of world demand and the RA-10 aims at 20%. The dependence on old reactors became evident in 2008 and 2010, when simultaneous outages at Petten and Canada's NRU left hospitals across half the world without technetium. Beyond molybdenum, reactors make iodine-131 for the thyroid, lutetium-177 for targeted therapies, iridium-192 for industrial radiography and brachytherapy, and dozens more. Transmutation-doped silicon, in which a silicon-30 atom captures a neutron and decays to phosphorus, is the material of high-power electronics: OPAL irradiated 900 tonnes in twenty years.

Research reactors operate inside cities and campuses because their risk is on a different scale. The fission-product inventory is a hundred times smaller than in a power plant, decay heat is measured in hundreds of kilowatts rather than tens of megawatts, there is no pressure that could break anything and no steam that could expand, and the core sits under ten meters of water that takes days to boil away. Add two independent shutdown systems, negative reactivity coefficients and, in modern designs, a confinement building with filtered ventilation. Regulation demands the same as for a power reactor, accident analysis included, but the extreme scenarios end with a core that is still covered.

The extreme case of inherent safety is the TRIGA. Its fuel mixes uranium with zirconium hydride, so the moderator is inside the fuel element itself. If power runs away, the fuel heats up in milliseconds and the hotter hydrogen in the hydride hands energy back to the neutrons instead of slowing them: fission probability drops, leakage rises, and the reaction shuts itself down. It is a prompt negative temperature coefficient, so large that it is used on purpose: a rod is ejected with compressed air, power jumps from kilowatts to thousands of megawatts and falls back within tens of milliseconds with nobody doing anything. That pulse is an experimental tool and, at the same time, the proof that the reactor cannot run away.