Smaller reactors bring nuclear power closer to fulfilling its promise

Nuclear power plants have not exactly sprouted like daisies across the United States. The country’s first commercial nuclear reactor of the twenty-first century didn’t come online until 2016. That was Unit 2 of the Watts Bar Nuclear Plant in Tennessee, which arrived two decades after Unit 1.

But that drought, and a similar one in Europe, might be ending. Two reactors began operating near Baxley in Georgia in 2023 and 2024, with governments and private parties around the world keen to invest in nuclear-plant developers. Some are eyeing up conventional light-water reactors, which use normal water as a coolant — the only type currently used commercially in the United States. But a lot of effort and funding is now going towards developing a new breed of smaller reactors that rely on different fuels and coolants, and which will require the development of new structural materials.

Nature Outlook: Nuclear power

Known as small modular reactors (SMRs), they can produce up to 300 megawatts of electrical power (MWe) — enough to run about 300,000 homes. This is much less than the 1,000 MWe typically produced by conventional light-water reactors, but SMRs should be cheaper and easier to construct. Existing commercial reactors are “very efficient, very good for the grid, but also very expensive to build”, says Jacopo Buongiorno, a nuclear engineer who directs the Center for Advanced Nuclear Energy Systems at Massachusetts Institute of Technology in Cambridge.

Part of the advantage of SMRs lies in their modularity. Instead of constructing an entire power plant from the ground up, large portions of the system will be built in a factory and shipped to the site for assembly. Think Lego, says Buongiorno: “I have my prefabricated bricks and I connect them to make my reactor.”

In March 2025, the US Department of Energy (DoE) announced US$900 million in grants to support the deployment of SMRs. One year later, the European Commission said it would invest up to €200 million (US$228 million) in the construction of SMRs.

Critics argue that SMRs are not necessarily more economical than are larger nuclear power plants. Edward Lyman, a physicist and director of nuclear power safety for the Union of Concerned Scientists, a non-profit organization in Cambridge, Massachusetts, has said that the smaller power output of SMRs means that factories would have to produce dozens of modular reactors for them to become more cost-effective than conventional reactors. He has argued that some SMR designs might be more dangerous than are existing reactors, owing to their different fuels and coolants.

Despite these concerns, the lure of faster and cheaper construction — and growing demand for clean energy, not least from data centres that run artificial-intelligence systems — has made developing modular reactors commercially attractive. Kairos Power in Alameda, California, is building a test reactor called Hermes 1 in Oak Ridge, Tennessee, and began construction on a 50-MWe demonstration plant, Hermes 2, in April. The company expects to begin commercial operations in 2030, and has a deal to sell its power to Google.

Fixing the fuel

Hermes 1 is the first non-light-water reactor approved by the US Nuclear Regulatory Commission (NRC) in more than 50 years. Light-water reactors, with their familiar cooling towers, are cooled by water and usually powered by ceramic uranium dioxide pellets, which are packed into zirconium alloy tubes called fuel rods. Hermes 1, by contrast, is fuelled by tristructural isotopic (TRISO) pebbles — poppy-seed-sized particles of uranium, encased in a carbon-ceramic shell. The reactor is cooled not by water but by a molten salt: a mixture of lithium fluoride and beryllium fluoride known as FLiBe.

TRISO fuel is the oldest and best understood example of what researchers call accident-tolerant fuels, says Nicholas Brown, a nuclear engineer at the University of Tennessee, Knoxville. A kernel of uranium dioxide, which is 350–600 micrometres in diameter, is surrounded by a shell of porous carbon that captures fission products that escape the kernel. Around that is a denser layer of carbon, which is itself encased by a layer of silicon carbide that acts as a miniature containment vessel.

Thousands of these particles are packed into each golf-ball-sized graphite pebble. When the pebbles are brought together inside the reactor, fission occurs, and the resultant heat is transferred into the coolant. Used pebbles typically flow out of the reactor like the contents of a gumball machine, and fresh ones are added from above. This enables refuelling without having to stop the reactor.

Another advantage stems from the FLiBe coolant that Hermes 1 uses. Water boils at 100 °C, so to remain liquid at the 250–325 °C temperatures of light-water reactors requires pressures of about 150 atmospheres. Hermes 1’s molten-salt coolant, by contrast, boils at about 1,430 °C — much higher than the reactor’s 650 °C operating temperature. That means the reactor can run at normal atmospheric pressure, removing the need for expensive containment vessels. And if there is a leak, there is no pressure to spew the coolant and the radioisotopes into the air. Furthermore, reactors using molten salt as a coolant can run at higher temperatures, leading to more efficient energy production.

Walk-away safe?

The fuel won’t melt at temperatures that the reactors can reach. In one test, TRISO survived for more than 12 days at 1,800 °C — much hotter than what even a worst-case accident could produce. That makes the fuel what the nuclear industry calls walk-away safe. “You can throw the keys away, walk away and not worry about a progression towards a severe accident like a Fukushima or Chernobyl,” Brown says.

The trouble with TRISO is that it requires high-assay low-enriched uranium (HALEU), in which the fuel material is enriched to contain about 15–20% uranium-235. In January, the DoE awarded a total of $2.7 billion to three companies to create domestic enrichment capacity for HALEU. Until that capacity is built up, Brown says, making TRISO will be slow and expensive.

Scientists are working on other fuels, both for conventional reactors and for SMRs. Brown’s group is investigating a 3D-printed insert made of molybdenum that could go in the middle of a uranium dioxide pellet. Because molybdenum conducts heat well, the insert could improve heat flow out of the fuel pellet by an order of magnitude — reducing the likelihood of it melting and causing an accident.

Another fuel and coolant are the basis of the Natrium reactor that TerraPower is building in Kemmerer, Wyoming. Natrium’s 345-MWe power level puts it just over the limit of what is usually considered an SMR, says Eric Williams, a mechanical engineer who is chief operating officer at the company in Bellevue, Washington. Its fuel is an alloy of uranium and zirconium. For a coolant, it uses liquid elemental sodium, which has a boiling point of nearly 900 °C — well above the 500 °C at which the reactor operates. Fission heats the liquid sodium, which in turn heats up molten sodium chloride. This molten salt stores the heat when electricity demand is low, then boils water to turn electricitygenerating turbines when demand rises.

Because the fuel, the cladding that surrounds it, the liquid sodium and the vessel itself are all metal, the system is excellent at conducting heat, says Williams. Even when a reactor shuts down, the radionuclides that have been produced by fission continue to decay, creating extra heat that is initially equivalent to about 7% of what the reactor produces at full power. The all-metal design simplifies handling of that decay heat. “We can actually cool the outside of the vessel by just having air flow by natural circulation,” Williams says. Of course, liquid sodium can burn if it comes into contact with air or water, but the company says that it has extra barriers and sensors in place to prevent that happening.

When the Fukushima Daiichi nuclear plant in Japan failed in 2011 after a tsunami, operators tried to pump in seawater to cool the reactors. But the earthquake that had caused the tsunami had also damaged the electrical grid and most of the plant’s backup power. As a result, it was impossible to cool the reactors enough, and radioactive material was released into the environment. Because the TerraPower reactor is air-cooled, cooling can continue indefinitely without anything needing to be pumped in.

There’s another built-in safety feature to using liquid sodium as a coolant, Williams says. Among the most dangerous radioactive by-products of fission are caesium-137 and iodine-131. Both are highly radioactive, and could get into the water and food supply in the event of a leak. But sodium bonds with both elements; therefore, a sodium-based coolant keeps them in the tank. “Even if the fuel were to all fail and all of the caesium and iodine generated gets into the sodium, it stays in the sodium,” Williams says.

Conventional light-water reactors are not necessarily more dangerous than an SMR such as TerraPower’s, says Raluca Scarlat, a nuclear engineer at the University of California, Berkeley. Nuclear plants can have all manner of safety features to ensure that they’re not dangerous: active safety, such as electrically powered pumps to carry coolant to the reactor, and passive backups that rely on gravity to drop water from a rooftop tank or drain fuel from the reactors. But some SMR designs have a greater number of inherent safety features that rely on physics rather than engineering, such as being able to operate at atmospheric pressure, and also provide more opportunities for designing passive safety systems.

The different safety characteristics of some SMRs mean that they might require smaller emergency buffer zones around them. That’s both because smaller plants necessarily produce less radioactive material, and because their design means that radioactive material is less likely to escape in the first place. In the United States, light-water reactors have long had an emergency planning zone radius of 16 kilometres, in which people would be required to evacuate or take shelter immediately after an accident, and a 80-kilometre-radius ingestion planning zone, in which radiation might get into soil and groundwater. In 2023, the NRC issued a rule saying that planning zones for SMRs could be set on a case-by-case basis, depending on their technology. Smaller zones could enable nuclear power plants to be sited closer to where power is needed, Williams says.

Salt power

Molten salt has more uses than as a coolant or energy storage medium. Mixing a radioactive element, such as uranium or thorium, into the salt creates a fluid that can act simultaneously as both fuel and coolant. In fact, the term molten-salt reactor often refers to designs using such a liquid-fuel mixture. Although TerraPower is focused on commercializing their Natrium technology, which is not a molten-salt reactor, the company has also worked on the Molten Chloride Reactor Experiment at the DoE’s Idaho National Laboratory in Idaho Falls, which runs on a mixture of molten sodium and uranium chloride.

Both chloride and fluoride salts are promising for fuel mixtures, says radiochemist Patricia Paviet, who directs the DoE’s research on molten-salt reactors. “We love them because it’s a higher temperature, more efficiency,” Paviet says. “We love them because it’s low pressure, so accident scenarios with pressure will not happen.”

Similarly to Kairos Power’s Hermes 1 pebble-bed reactor, a reactor powered by molten salt can be refuelled without being switched off — spent fuel can be drained out and new fuel pumped in while it’s operating. Another advantage over light-water reactors is that actinides — radioactive elements created during a nuclear reaction — can fuel subsequent reactions. In molten-chloride reactors, the nuclear-fission chain reaction is sustained by higher-energy neutrons. By contrast, the water in light-water reactors acts as a moderator, slowing the neutrons down. As a result, molten-chloride reactors can burn the actinides as extra fuel, which light-water reactors cannot do.

Molten-salt reactors, both chloride and fluoride, also present an opportunity to minimize waste, Paviet says. Spent fuel from a light-water reactor contains about 95% uranium and 1% plutonium. France and Russia have extensive recycling programs, but the United States doesn’t recycle any nuclear waste, in part because it’s cheaper to dispose of it. Processing liquid fuel is easier than manufacturing or taking apart solid fuel, Paviet says.

The presence of salts in machines operating at extremely high temperatures raises the spectre of corrosion. Scarlat argues that, if the systems are designed to keep water and oxygen out, oxidation should not be a problem. But even if that’s accomplished, there are other mechanisms that can weaken structural materials in a reactor. For example, molten salts tend to dissolve chromium, the element added to steel to make it stainless. This means that reactor parts should not contain more than about 7% chromium, says Steven Zinkle, a nuclear engineer and materials scientist at the University of Tennessee, Knoxville.

Structural materials in these newer, high-temperature reactors also have to withstand the effects of radiation, which can create defects by knocking atoms out of place. The high temperatures make it easy for those defects to migrate and form clusters, which make the material more brittle. Zinkle says that out of the thousands of known alloys, just six are qualified for use in high-temperature reactors by the American Society of Mechanical Engineers. And none of them, he says, are qualified for the highest-temperature reactor that designers would like to be able to use.

Zinkle is leading a DoE-funded project to develop more alloys for use in reactors. One is a magnetic steel and the other is a vanadium-based alloy. Both are precipitate strengthened, a process that uses repeated cycles of heating and cooling to form small bits of material inside the larger structure that block defects and have been shown to provide radiation resistance. Other materials scientists are trying to engineer the boundaries of grains in a material’s crystalline structure to act as a trap for defects.

Developing new materials, fuels and reactor designs is a slow process that requires patience, Brown says. But he thinks that the results will be worth the effort. Nuclear power is “really important for the future of the country and also for the future of the world. It’s very safe. It is really the next-generation energy source that is going to enable us to move beyond fossil fuels.”