In the design of liquid fluoride thorium reactors, there is a Gordian Knot frustrating the realization of very low costs. At one extreme, there is a large, low-power density core, with large inventories of salt, but a long-lived graphite moderator. At the other extreme, there is a small core, with a short-lived graphite moderator.
In the current Flibe Energy design concept, the salts are alternated between two small, high-power-density reactors, and graphite moderator prisms are reshuffled, much like the solid fuel elements of commercial light water reactors. Must there be two reactors, and in turn, a doubling of certain capital costs? Another possibility is a single reactor with a “random-packed bed of graphite spheres,” which are continuously recirculated while the reactor is online. However, the salt volume fraction in a monodisperse pebble bed is suboptimal, and the pressure drop across a polydisperse pebble bed with the optimal salt volume fraction would be excessive.
In the Seaborg Technologies design concept, the fluoride fuel salt is contained by an array of (probably metallic) tubes, and interlaced with a liquid hydroxide (or deuteroxide) “moderator salt.” The structureless moderator has an indefinitely long life, and the tubes might live three times as long as graphite, all else equal.
“Corrosion in molten hydroxides can be controlled by maintaining the redox potential and the oxoacidity of the melt at a particular range of values where limited dissolution of the container material occurs. For many metals and alloys, including nickel-based alloys, this is generally an acidic melt with a reducing potential (i.e. a reduction potential of the molten hydroxide which is lower than the reduction potential of the material in contact with the molten hydroxide), but not so reducing as to form hydrogen or hydride… the potential in a molten salt can be controlled by the relative amounts of multivalent soluble compounds, and the oxoacidity can be controlled by bubbling gas (e.g. H₂O in the case of hydroxides) through the salt or by adding fixed amounts of strong oxide donors (e.g. Na₂O). Furthermore… the oxoacidity can be controlled by controlling the composition of the cover gas, e.g. the partial pressure of H₂O can be controlled.”
However, the somewhat neutron-poor thorium fuel cycle places a premium on neutron economy, and militates against the parasitic neutron absorption of metallic tubes in the active region.
Interestingly, there is some precedent for the use of hydroxide salts in homogeneous reactors, dating back to the earliest days of molten salt reactor research (the Aircraft Nuclear Propulsion program).
“…since graphite has a much larger slowing-down distance than low-temperature hydrogen-based moderators such as H₂O or D₂O, reactors moderated with graphite need a much greater volume of graphite mixed with the fuel or surrounding the fuel for the reactor to achieve criticality. Therefore, graphite-moderated reactors will have a much lower volumetric power density, and therefore, may also have higher capital costs per installed unit of power… in the 1950s, research scientists in the United States and the United Kingdom contemplated the use of alkaline hydroxides (e.g. NaOH) as a low-pressure, high-temperature reactor coolant or as a carrier for a fluid-fueled reactor… Hydroxide coolants were also seriously considered for use in the Aircraft Nuclear Propulsion program under development in the United States during the 1950s. It was recognized that hydroxides would have the advantage over molten salts (such as mixtures of LiF, BeF₂, ZrF₄, and UF₄) in that the hydroxides would contain hydrogen (or could contain deuterium), which would enhance moderation and reduce the reliance on graphite as a moderator and allow more compact reactor cores… These hydroxides have melting points that are comparable to that of FLiBe (~460°C)… Due to their hydrogen content, hydroxides also have slightly larger heat capacities than [fluoride] salts… reduced migration area… would significantly reduce reactivity losses due to neutron leakage.”
“The high effectiveness of the hydroxides in slowing down neutrons makes for relatively small reactor sizes and hence small shield weights, which is particularly essential in aircraft application. The use of hydroxides… leads to less complicated reactor core structures… No pressurization is required to keep the hydroxides in the liquid state at the operating temperatures required for the aircraft application; in addition, the hydroxides maintain reasonably high densities at these temperatures.”
“Liquid fuels possessing the capacity for self-moderation would permit the simplest reactor design… The logical, and indeed almost the only, choice of materials for self-moderating fuels are solutions of uranium compounds in the alkali hydroxides… Uranium trioxide seems to be soluble in a mixture of sodium and lithium hydroxides to an extent that would permit construction of a homogeneous high-temperature reactor…”
Hydroxide salts were abandoned for three reasons. First, although a mixture of equal parts ⁷LiOH and NaOH can dissolve enough UO₃ at 750°C to achieve criticality, fluoride salts accommodate significantly higher concentrations of uranium, and other materials. Second, it was assumed that, at high temperatures, the hydroxide salt would be desiccated, making it basic, and so corrosive as to be utterly irrepressible. Third, highly depleted lithium was not available at the time.
If the corrosion problem has indeed been solved, and highly depleted lithium becomes readily available for molten salt reactors of all kinds, then it might make sense to revisit the simplest molten salt reactors ever conceived, in which hydroxide salts are coolants, moderators, and fuels, all in one.
The most obvious technical hurdle is the solubility of uranium in the hydrous, acidic hydroxide salt. This paper describes how “the solubility of anhydrous Na₂WO₄” in molten NaOH “drastically increased with the partial pressure of water vapor,” and how “commonly, the solubility of a metal oxide in an NaOH melt, the so-called acidic dissolution, increases with the activity of H₂O.” This is strikingly similar to how Seaborg Technologies intends to tame molten hydroxide salts, and might just work for Na₂UO₄ (or Na₂U₂O₇).
In an LFTR, leakage of the fuel salt (into the drain tank or otherwise) will tend to separate the fuel and moderator, and reduce reactivity — outside of a bizarre accident scenario in which the moderator is pulverized and slurried in the fuel salt, perhaps. This is not possible with a “self-moderating” fuel, so the liquid hydroxide thorium reactor will have to prevent recriticality in much the same way as a fast reactor.
An LHTR will produce short-lived ²⁴Na, and a liquid deuteroxide thorium reactor will produce somewhat more tritium than a heavy water-moderated CANDU reactor.
Interestingly, zirconium is among “those metals which show a measure of corrosion resistance toward fused [NaOH] under suitable conditions at 500°C.” Zirconium is “essentially transparent to neutrons” and “prized for internal reactor parts,” and the co-design of a zirconium alloy with the various corrosion protection systems of the reactor is probably the most realistic route to a long-lived barrier with high resistance to both neutron embrittlement and the “attack of fused hydroxides” (exacerbated by fission products and radiation, of course). Beryllium is a particularly attractive alloying agent, as it increases strength and corrosion resistance, without causing embrittlement, or reducing neutron economy. The high power density and soft neutron spectrum of an LHTR imply infrequent and relatively easy replacement of a small and long-lived barrier, with minimal generation of activated waste. The activation products of zirconium and beryllium are not short-lived, however.
The LHTR could be ‘more LFTR than LFTR,’ with an even smaller footprint, and a moderator that (like the fuel) is not damaged by radiation, and does not need to be fabricated or handled.
