Foreword: it took me over two years but now I think I’ve finally written something that’s at least 50% correct.
No commercial nuclear reactor has ever had the right combination of coolant and fuel cycle. Oak Ridge National Laboratory was on the right track with the Molten Salt Breeder Reactor, a liquid fluoride thorium converter, but bean counters in the Nixon administration killed it.
ORNL did manage to build a liquid fluoride uranium burner called the Molten Salt Reactor Experiment. From 1965 to 1969, MSRE ran for 13,000 hours at full power, demonstrated the outstanding high-temperature corrosion resistance of the nickel-base alloy Hastelloy-N, and demonstrated online refueling.
A liquid fluoride thorium converter could have lower capital and operating costs than those of a modern coal-fired power plant. In a world of ubiquitous illiteracy and political failure — case in point, fucking Rick Perry is the United States Secretary of Energy, as of this writing—only unaided market forces can put serious downward pressure on carbon dioxide emissions. Yes, wind and solar energy are fine, but they’re intermittent, and as they provide a larger and larger fraction of electric power, Rube Goldberg schemes like grid energy storage, on-site energy storage and superconducting transmission lines will become necessary to guarantee the availability of electric power. Only nuclear and geothermal energy are both carbon-neutral and suitable for absorbing baseload, but geothermal energy will always require more cooling water, all else equal.
It all comes down to how the fuel is dissolved in a liquid fluoride salt. This confers four highly desirable characteristics to the reactor: low gauge pressures, a negative temperature coefficient of reactivity, no excess reactivity, and compatibility with a passive safety system of freeze valves and drain tanks.
Because fluoride salts have very low vapor pressures at the temperatures typical of the core of an MSBR- or MSRE-like reactor, the reactor operates at very low gauge pressures, and stores very little pressure energy with which to disperse radioactive materials in the event of a containment failure. At any rate, “the fluoride salts are chemically stable, bind radioactive fission products to the salt, and have extremely high boiling points.” The salts have large margins to boiling even at atmospheric pressure, and will not boil in the event of a containment failure, like pressurized water. The low gauge pressures obviate expensive pressure vessels, and the large margins to boiling obviate expensive containment buildings.
The negative temperature coefficient of reactivity is illustrated by the negative slope of the MSRE’s overall multiplication factor versus temperature curve in Figure 6.10. If you will, picture the core salt circulating from the core to a heat exchanger and back. The core has a fixed volume, but the core salt does not. The reaction can only proceed in the vicinity of the graphite moderator, which exists only in the core. The reactivity is positively correlated with the mass of fuel inside the core. If some perturbation increased the temperature of the core salt, then the density of the core salt would decrease, the mass of fuel inside the core would decrease, the reactivity would decrease, and the temperature of the core salt would decrease. This is a negative feedback loop, illustrated by the damping of a power transient in Figure 6.11. In brief, “expansion of the fuel salt, which removes fuel from the active core, is… the principal inherent mechanism for compensating any reactivity additions.” The heating power of the reactor tracks the cooling power of the heat exchanger to which it is coupled, and as such, the core salt is a sort of natural thermal-hydraulic closed-loop controller. “Failure of control systems or reactivity insertion events only leads to reactor stabilization at a slightly higher temperature. Inherent heat sinks are available initially to absorb transient and decay heat, with heat losses providing long-term cooling for both core and containment.”
A liquid-fuel reactor can be a continuous reactor, because the fuel can be pumped out of the active region, chemically processed, and pumped back into the active region. In contrast, a solid-fuel reactor is necessarily a batch reactor. It must sustain criticality despite the buildup of neutron-absorbing fission and decay products inside the solid fuel elements, so it has a much larger fissile inventory than would be necessary in their absence. This excess reactivity must be compensated for, to a decreasing extent over the refueling interval, by complex and expensive control systems. In a liquid-fuel reactor, the short residence time of fission and decay products in the active region reduces the size of the fissile inventory. In a catastrophic accident, a liquid-fuel reactor would have less radioactive material to lose than an equivalent solid-fuel reactor. In addition, online refueling reduces downtime.
As if that weren’t enough, a liquid-fuel reactor can be equipped with a passive safety system of freeze valves and drain tanks. If, for whatever reason, cooling power to a plug of frozen salt at the bottom of the core salt loop were cut off, or the core salt simply became hot enough to melt that plug of frozen salt, the core salt would drain into a drain tank. This way, the reactor can be designed to retain heat and neutrons, and the drain tank can be designed to dissipate them. In contrast, a solid-fuel reactor must retain or dissipate heat and neutrons, depending on the circumstances, and so it is more complex and expensive.
The mass of transuranic wastes per unit energy produced by uranium-233 burners such as the MSBR is about one tenth that of prevailing uranium-235 burners with uranium fuel cycles. This is because the ‘waste’ of a uranium-233 burner is fertile uranium-234, or rather, a uranium-233 burner also happens to be a uranium-234 converter and a uranium-235 burner. In order to become a problematic transuranic nucleus, a uranium-233 nucleus must react with two more neutrons than a uranium-235 nucleus, and one of these additional reactions is an additional chance, about nine-in-ten, for the nucleus to fission rather than capture a neutron. The fission product wastes only have to be stored for about four human lifetimes, although the fluoride waste form may not be a good idea. And, of course, long-lived fission products like technetium-99 and iodine-129 can be transmuted by thermal neutrons, transuranic wastes can be burned by fast reactors, and at any rate, nuclear waste can be safely disposed of in ductile salt structures.
Two-fluid liquid fluoride thorium converters will require a barrier to separate the core and blanket salts, and withstand the large thermal neutron flux through their boundary for long enough. There are many candidate barrier materials, and, for what it’s worth, I doubt that any such barrier, or the reactor vessel as a whole, would have to be replaced more frequently than solid-fuel reactors are refueled. It’s possible that the difficulty of the barrier material selection problem is exaggerated, because the graphite moderator could also separate the core and blanket salts for long enough. If no suitable barrier materials exist at all, then there are alternative one-fluid designs with no barriers, but more complex chemical processing. To be fair, reactor-grade enriched uranium is not prohibitively expensive, and one-fluid liquid fluoride uranium burners may be a better proposition for the time being.
It is often said that side reactions in the thorium fuel cycle inevitably breed uranium-232, and that uranium-233 contaminated by this uranium-232 is almost impossible to use in a nuclear weapon, because there are prolific gamma emitters in the uranium-232 decay chain. There is a liquid-liquid extraction process by which protactinium can be selectively reduced from the blanket salt by lithium-7 in a column of bismuth, and online protactinium extraction may suppress the breeding of uranium-232 in the blanket salt…
²³²Th (n,γ) ²³³Th (β−) ²³³Pa (n,2n) ²³²Pa (β−) ²³²U
²³²Th (n,2n) ²³¹Th (β−) ²³¹Pa (n,γ) ²³²Pa (β−) ²³²U
²³⁰Th (n,γ) ²³¹Th (β−) ²³¹Pa (n,γ) ²³²Pa (β−) ²³²U
…but the protactinium would nevertheless be sequestered in a part of the plant that is difficult to surreptitiously access, very hot, and highly radioactive. And when the protactinium-233 decays into uranium-233, and is added to the core salt…
²³²Th (n,γ) ²³³Th (β−) ²³³Pa (β−) ²³³U (n,2n) ²³²U
…it inevitably contaminates itself with uranium-232.
A liquid fluoride thorium converter is the only nuclear reactor that operates at low gauge pressures, and has a small footprint, a high outlet temperature, a small fissile inventory, and on-site reprocessing. Even if DEMO succeeds, its capital and operating costs will be much higher than those of a liquid fluoride thorium converter.