Ask most people how you shut down a nuclear reactor and you will hear about control rods, which is right as far as it goes. Rods drop into the core, fission stops, everybody goes home. The awkward part shows up a day or so later, once the core has cooled off and the chemistry starts pushing back the other way.<\/p>\n
On September 1, NuScale Power and a Tennessee ceramics manufacturer called MillenniTEK said they had fabricated the first boron-oxide pellets<\/a> for the system that handles exactly that problem. The pellets go into baskets inside the reactor’s containment vessel. In an emergency they dissolve themselves into the cooling water.<\/p>\n None of them are installed in anything. This is a manufacturing milestone, which sounds like thin news until you look at what these things are being asked to do.<\/p>\n Boron eats neutrons, and it is unusually good at it because of one isotope. Natural boron runs about 19.9 percent boron-10 and 80.1 percent boron-11. Boron-10 absorbs slow neutrons with a cross section of roughly 3,840 barns, while boron-11 sits at about 0.005 barns. Almost all of the work is done by a fifth of the material.<\/p>\n Fewer loose neutrons means fewer uranium atoms splitting. Every pressurized water reactor in the country already keeps boric acid dissolved in its coolant for this, turning the concentration up and down over a fuel cycle. Control rods are the emergency brake. Boron is the parking brake you leave on afterward.<\/p>\n The plumbing is simpler than you would expect, and there is no pump anywhere in it.<\/p>\n When NuScale’s emergency core cooling system fires, two vent valves open at the top of the reactor vessel and let steam escape into the containment vessel wrapped around it. The steam hits the containment wall, condenses, and runs down. Two recirculation valves lower down then open and let that water back into the reactor. A blocking valve holds those recirculation valves shut until the pressure difference between the two vessels drops far enough, so the sequence happens in order rather than all at once.<\/p>\n Some of the condensate running down the wall gets caught and routed into baskets bolted to that wall, which are packed with boron oxide. The water dissolves the pellets and carries the boron down to the bottom of the containment, where the recirculation valves feed it back into the core. The rest of the condensate goes down through mixing tubes so the whole inventory stays evenly mixed. The NRC calls the whole arrangement the ECCS supplemental boron system.<\/p>\n So why not just carry more boron in the coolant from the start and skip the hardware?<\/p>\n Because boiling separates the two. When the coolant boils, the boron stays behind in the liquid and the steam that vents into containment is essentially clean water. Condense that steam, run it back into the reactor, and you are diluting the core at the worst possible moment. The baskets exist to load that water up on the way past.<\/p>\n Here is the bit I had to read twice. A reactor that is getting colder is a reactor that wants to start back up.<\/p>\n Over the 72 hours the system is designed to cover, several things pull in opposite directions at once. Boiling concentrates boron in the core region, which pushes the reactor further down. The baskets add more on top of that. Every control rod is counted as inserted except the single strongest one, which the analysis assumes is stuck out. Samarium builds up in the fuel and holds the core down too.<\/p>\n Pulling the other way, the core is cooling, and colder water is a better neutron moderator, which adds reactivity. Xenon does the same thing on a delay. It is a powerful absorber that spikes right after a shutdown and then decays, and by hour 72 there is almost none of it left.<\/p>\n The NRC’s Advisory Committee on Reactor Safeguards, which reviews this material before the staff signs anything off, put it about as plainly as a federal committee can<\/a>: the conditions that are good for cooling the core, meaning low temperatures and low decay heat, are the ones that make it harder to keep the core shut down.<\/p>\n So you want the core cold, and being cold is exactly what makes it harder to hold down.<\/p>\n The same committee document contains the number that made me sit up. Across the cases NuScale analyzed, the smallest calculated margin to criticality was 28 parts per million of boron, and the committee noted that boron concentration uncertainty in a normal pressurized water reactor typically runs 50 to 100 ppm. So the calculated cushion is smaller than the usual error bar on the measurement.<\/p>\n The committee did not leave it there, and neither should anyone quoting it. NuScale builds conservatisms into the methodology, including deliberately pessimistic temperatures. NRC staff also ran their own computational fluid dynamics calculations and found the company’s boron tracking model was leaving margin on the table, worth roughly another 180 ppm of shutdown margin in the example they published. With those included, the committee agreed the core stays subcritical and said it had no objection to the safety evaluation being issued.<\/p>\n Both things are true, and I think the second one matters more than the first. Still, a 28 ppm headline figure is a useful reminder that “passive” does not mean “effortless.” Somebody spent years arguing about this in public documents.<\/p>\n NuScale has already run dissolution tests, feeding heated water through baskets at various flow rates and temperatures. The NRC’s read on those tests was that the rig was not prototypical and the results were not correlated to the actual module conditions, which is a polite way of saying it proved the chemistry rather than the product.<\/p>\n So the staff attached a condition. Whoever first builds a plant using this methodology has to run a dissolution test on the first module during startup testing, and the acceptance criteria have to confirm the as-built hardware works: basket dissolution rates, how much condensate the collection rails actually catch, mixing tube flow, and the boron concentration that ends up in the lower containment.<\/p>\n That test cannot happen until a module exists in a plant, which is the honest size of this week’s news. A ceramics shop in Knoxville proved it can press and fire the pellets to nuclear-grade specification, and that is a real step, because the supply chain for parts like this is thin enough<\/a> that individual suppliers become news.<\/p>\n The approvals deserve precise language, because coverage tends to blur them. NuScale’s 50-megawatt design was certified by the NRC in January 2023. The 77-megawatt version, the one these pellets are built for, received a standard design approval on May 29, 2025. An approval means the design can be referenced in a license application. It is not permission to build anything, and a construction permit or combined license is a separate process that the NRC says commonly takes 30 months or more.<\/p>\n The module is a pressurized water reactor, so it burns ordinary fuel enriched below 5 percent rather than the high-assay stuff most microreactor startups are waiting on<\/a>. Each one is 76 feet tall and 15 feet across, weighs around 700 tons, and ships from the factory in three pieces by truck, rail, or barge.<\/p>\n NuScale markets the module as able to “shut down and self-cool indefinitely”<\/a> without operator action or outside power. The methodology the NRC actually reviewed runs to 72 hours and explicitly stops there, leaving post-72-hour recovery outside its scope. That gap is not a contradiction so much as a difference between what a machine may do and what a regulator has signed off on it doing, and it is the sort of distinction that gets flattened in a press release.<\/p>\n Commercially, the company is still waiting. NuScale reported $1.9 billion in cash and investments on August 5, with quarterly revenue down $8.0 million from a year earlier, after the Fluor engineering work for the Romanian project finished in late 2025. Its R&D spending rose $6.6 million over the same quarter, driven by exactly this kind of component work.<\/p>\n Its partner ENTRA1 is in discussions with the Tennessee Valley Authority toward a definitive power purchase agreement rather than holding one. The six modules planned for a former coal plant site at Doice\u0219ti, in Romania, are the furthest along of any NuScale project anywhere.<\/p>\nSo what is the boron actually for?<\/h2>\n
The pellets sit in baskets on the containment wall<\/h2>\n
Cooling the core is what causes the trouble<\/h2>\n
The margin is tighter than the brochure suggests<\/h2>\n
Nobody has dissolved these in a real module yet<\/h2>\n
Where the reactor itself stands<\/h2>\n