Simulating the invisible chemistry that could make or break molten salt reactors

Molten salt reactors have long been one of nuclear energy’s most tantalizing ideas: a reactor whose fuel is not a ceramic pellet encased in metal, but a glowing liquid salt that circulates through the core like blood through a heart. The concept dates back to the 1950s and 1960s, when Alvin Weinberg’s team at Oak Ridge National Laboratory built and successfully operated the Aircraft Reactor Experiment and the Molten Salt Reactor Experiment, proving the chemistry of the idea was more than science fiction. Now, as China’s thorium molten salt reactor came online in 2023 and companies across the United States, Europe and Canada race toward commercial designs, a quiet technical problem threatens to hold the whole enterprise back: nobody has been able to watch, atom by atom, what the fuel salt actually does inside a running reactor.

A new study from researchers at Idaho National Laboratory, published in Frontiers in Nuclear Engineering, tackles exactly that challenge. Samuel A. Walker, Mauricio Tano, Parikshit Bajpai and Abdalla Abou-Jaoude have built a multiphysics species tracking framework inside the Multiphysics Object-Oriented Simulation Environment, or MOOSE, a suite of open-source simulation tools developed under the U.S. Department of Energy’s Nuclear Energy Advanced Modeling and Simulation program. The framework welds together three physics domains that are inseparable in a liquid-fueled reactor: neutron transport and fuel depletion, thermal-hydraulics, and thermochemical equilibrium. With those couplings in place, the model can follow fission products, corrosion products and actinides as they dissolve, migrate, precipitate and plate out across an entire engineering-scale reactor system.

Why does this matter so much? Because in a molten salt reactor, everything is coupled to everything else. Fission splits uranium atoms, which changes the salt’s composition. That change alters the salt’s redox potential, effectively its chemical appetite for electrons. A rising redox potential makes the salt hungrier for metal atoms, encouraging chromium and iron to leach out of the nickel-based structural alloy that holds the reactor together. Meanwhile, hot salt in the core and cooler salt after the heat exchanger create temperature gradients that drive galvanic corrosion, the electrochemical equivalent of a battery wired through the reactor’s own plumbing. The researchers call the overall effect depletion-driven thermochemistry: the fuel’s ongoing consumption slowly pushes the salt into a more oxidizing state, because fission produces noble gases like xenon and noble metals like molybdenum that do not replace the cationic charge lost from uranium.

The consequences ripple outward. Gaseous species such as cesium-iodide and molecular iodine volatilize more readily as the salt oxidizes, which directly affects the radioactive source term, the inventory of material that could escape in an accident, and the design of off-gas systems. Insoluble noble metal fission products deposit on heat exchangers and vessel walls, where their radioactive decay generates heat in precisely the places cooling systems may not expect it. The migration of delayed neutron precursors through the circulating loop alters the effective delayed neutron fraction, a parameter that governs how quickly and safely a reactor can respond to changes in power demand. And from a safeguards standpoint, species tracking could give International Atomic Energy Agency inspectors real-time chemical signatures of any attempt to divert fissile material.

The technical machinery behind the framework is formidable. Fuel depletion is handled by Griffin, a NEAMS neutronics code that solves the Bateman equations, the bookkeeping of nuclide creation and destruction through fission, transmutation and radioactive decay, with added source and sink terms for feed and waste streams such as gas removal. Thermochemistry is handled by Thermochimica, a Gibbs energy minimization library coupled into MOOSE’s Chemical Reactions Module and running against the Molten Salt Thermodynamic Database, MSTDB-TC version 3.1. Given a temperature, pressure and composition, Thermochimica computes the stable phase assemblage and, crucially, the fluorine chemical potential, which converts directly into an electrochemical potential at the salt-metal interface. Species transport is solved with the Nernst-Planck equations in MOOSE’s Navier-Stokes module, capturing diffusion, electromigration and advection of ionic corrosion products, while Butler-Volmer kinetics describe the redox currents that drive leaching and plating at solid surfaces.

The team demonstrated the framework on a generic Molten Salt Fast Reactor model drawn from the National Reactor Innovation Center’s Virtual Test Bed, a 3,000-megawatt-thermal design with 18 cubic meters of LiF-UF4 fuel salt circulating through the core roughly every five seconds. The simulations revealed a striking spatial story. In the hot zone at the top of the core, where temperatures are highest and flow is slowest, the salt aggressively dissolves chromium from the structural alloy; the thermochemical equilibrium predicts roughly 120 moles per cubic meter of chromium ions in the salt boundary layer there. As the salt cools through the heat exchanger, that equilibrium concentration collapses to about 12 moles per cubic meter, so the corrosion products precipitate and plate out in the cold leg at the core inlet. The reactor essentially becomes a galvanic cell with a corrosion source in the hot leg and a deposition sink in the cold leg, establishing a steady circulating flux of stolen structural metal.

The simulations also quantified how much control operators can exert over this process. By adding small amounts of beryllium, a more thermodynamically stable metal that reduces the salt’s fluorine potential, the researchers showed the salt’s capacity to carry ionic corrosion products could be dramatically suppressed: total chromium in the 18-cubic-meter salt volume peaked at 11.6 parts per million without control, versus 2.89 and 0.96 parts per million under progressively stronger redox control. The model further showed that without any intervention, depletion of uranium steadily raises the redox potential over roughly 17 years of operation, driving chromium concentrations upward until precipitation or material degradation intervenes. In other words, redox control is not an optional refinement but a structural necessity for reactor lifetimes measured in decades.

Perhaps the most dramatic demonstration came from a simulated reactor shutdown. Insoluble noble metal fission products, treated as a single condensed pseudo-element, were tracked as they formed in the core and deposited on structures through advection and a mass transfer correlation adapted from the classic Dittus-Boelter heat transfer relation used in the original MSRE. When the control rods were inserted and the pumps shut off, the deposited noble metals became the dominant heat source, creating a hot zone above the heat exchanger at the top of the core that natural circulation, driven by buoyancy, could not fully reach. The results show why spatially resolved decay heat modeling matters: after shutdown, the geometry of where fission products settled during operation determines where the reactor stays dangerously hot.

The study also produced a sobering result for accident analysis. Depletion-driven oxidation caused the vapor pressure of gaseous iodine, I2, to jump by roughly nine orders of magnitude, from about 3.3 times ten to the minus eleventh atmospheres to about 1.3 times ten to the minus second, with cesium-iodide vapor pressure rising by a factor of about eighty. Notably, the calculations showed this volatilization is spatially non-uniform, peaking where the redox potential is highest, which means off-gas system designers cannot assume a homogenized salt chemistry when deciding where to place liquid-gas interfaces or how to size gas-handling equipment.

The authors are candid about the framework’s limits: diffusion of species within the solid alloy itself, including chromium migration along grain boundaries, is not yet modeled, and the thermochemical database currently represents pure metals rather than the full nickel-based alloy, so the findings are illustrative rather than definitive. Future work will add phase-field models of dealloying, validation against experimental salt loop data, and extensions to electrorefining and even molten salt tritium breeding blankets for fusion, where lithium consumption drives the same corrosive chemistry. But the trajectory is clear. For a technology that once lost federal support partly because of the complexity of its chemistry, the arrival of a tool that can simulate, at engineering scale, the leaching, plating, precipitation and volatilization that once had to be discovered the hard way, may prove to be the difference between molten salt reactors remaining a laboratory legend and becoming the reactors that finally deliver on the liquid-fuel promise first demonstrated at Oak Ridge more than half a century ago.

Subject of Research: Multiphysics species tracking and corrosion modeling in molten salt reactors

Article Title: Multiphysics methods for species tracking in molten salt reactors

Article References: Walker, S. A., Tano, M., Bajpai, P., & Abou-Jaoude, A. (2026). Multiphysics methods for species tracking in molten salt reactors. Frontiers in Nuclear Engineering, 5, Article 1779804. https://doi.org/10.3389/fnuen.2026.1779804

Image Credits: AI Generated

DOI: 10.3389/fnuen.2026.1779804

Keywords: molten salt reactors, multiphysics simulation, species tracking, corrosion, redox potential, MOOSE framework, Thermochimica, nuclear engineering, fuel depletion, thermal-hydraulics, Molten Salt Fast Reactor, beryllium redox control