WASH-1097: Appendix C, Thorium Fuel Reprocessing

Appendix C from WASH-1097, concerning the reprocessing of thorium-based fuels. Very interesting. You can readily see that the ease of reprocessing is strongly based on the form of the fuel. If the fuel is solid, such as an oxide or carbide, there are a great number of steps needed simply to arrive at the chemical separation step, which itself is disadvantaged.

However, if the fuel is a fluoride, and the reactor a two-fluid fluoride reactor, then the reprocessing steps can take advantage of incredibly simple techniques like fluorination and distillation. In addition, protactinium can be actively isolated, which strongly improves conversion ratio. Another argument for a two-fluid fluoride reactor!

---

APPENDIX C

REPROCESSING OF THORIUM FUELS

1. THE GOALS OF REPROCESSING

A spent U-Th fuel element requires reprocessing in order to recover bred fuel, restore the proper fissile/fertile ratio, remove fission products, and in the case of solid elements, repair radiation damage (in conjunction with refabrication processes).

What may be termed a complete thorium cycle processing scheme calls for the separation of uranium, thorium, and protactinium from fission products and from each other. In many circumstances a simpler scheme is adequate. For example, if spent fuel is allowed to cool for 270 days before reprocessing, decay of Pa-233 to U-233 is virtually complete and Pa removal, per se, need not be provided. Or, it may be decided that a rather low decontamination factor of uranium from fission products is acceptable for fuel in which the U-233 is contaminated with U-232 to a degree which would preclude direct refabrication, even if fission product activity were made negligible. Thus, reprocessing methods may vary with the industrial maturity of the nuclear power industry, and associated technology.

Processing goals and methods will also be a function of raw material prices. When thorium is inexpensive, a reactor product contaminated with highly radioactive Th-228 can be allowed to age until the radioactivity decays; and high fission-product decontamination factors will not be needed for the stored material.

2. GENERAL PRINCIPLES OF THORIUM-URANIUM PROCESS CHEMISTRY

Practical schemes for processing thorium-cycle fuels are based on the separation of Th, U, Pa, and fission products by means of selective partition between an aqueous phase and an organic solvent, or upon differences in volatility among the fluorides of the elements. Which type of process is best in a given case depends on the kind and degree of separation necessary and on the chemical and physical nature of the fuel material. Solvent extraction is in general more versatile, and permits some separations which are not possible in a volatility scheme. For example, the separation of thorium from rare-earth fission products is readily attained with solvent extraction but not with volatility techniques. However, there are situations where a volatility method is clearly the preferred choice, as in the extraction of bred uranium from a molten-salt blanket.

When an aqueous method is to be used, the preparation or head-end steps are used to transform U and Th into nitrates dissolved in aqueous solution, since such solutions lend themselves best to practical solvent extraction processes. When volatility methods are used the U and Th must be converted to fluorides if they are not already present in that form. Some typical head-end operations in both aqueous and volatility flow-sheets are described below.

3. SOLID FUEL HEAD-END PROCESSES

3.1 Decladding

3.1.1 METAL-CLAD ELEMENTS

Among the possible procedures for decladding metal-clad fuel elements are: mechanical opening of fuel elements, followed by leaching of U and Th; dissolution of the clad by a solvent which does not affect U and Th (in whatever form they may be present); dissolution of the entire element; electrolytic decladding and/or dissolution; and gas-reaction decladding in a fluidized bed, as for instance by an HF-O2 mixture, followed by leaching or fluorination of the bed. Development of the first two of these has been carried much further than the others.

In the first procedure mentioned, which is known as "chop-leach" or "shear-leach," the mechanical operation opens the fuel to chemical exposure. In the "leach" portion of the procedure, a reagent is required which will dissolve the core - the alloy or compound of the fissile or fertile material - without dissolving much of the cladding material. Since metal-clad thorium-containing cores generally require fluoride-catalyzed nitric acid for dissolution, shear-leach is suitable only for clads which withstand this reagent. Experiments have shown that Zircaloy is sufficiently resistant (55).

In the second type of process, chemical decladding, the opposite condition is required: the reagent must dissolve the metal covering but not the fuel core. This requirement appears to be satisfied by the "Zirflex" reagent (aqueous NH4F plus NH4NO3) for Zircaloy-clad ThO2-UO2; by the "Sulfex" process (boiling 4 to 6M H2SO4) for stainless-steel-clad oxides; and by an aqueous NaOH-NaNO3 dissolvent for aluminum-clad oxide or metal (52). Much experimental work remains to be done, covering all possible variables of oxide preparation, length of irradiation, etc.; but the available data indicate that there will be no serious difficulty in developing an aqueous head-end step for any likely thorium fuel which will be as convenient and as economical as those now used for uranium fuels.

Certain non-aqueous decladding methods which are being developed for uranium fuels may prove applicable to thorium fuels as well. In the "Zircex" process, zirconium alloy clad is removed by treatment with gaseous HCl at a temperature high enough (500°C) to volatilize the ZrCl4 product; oxide cores remain unreacted under properly controlled conditions. In the HF-O2 process, a mixture of these gases (20 percent-40 percent HF) is used to disintegrate claddings of either Zircaloy or stainless steel. The alloy constituents are converted to a mixture of their fluorides and oxides; uranium oxide cores are also partially converted to fluoride. The reaction is best carried out in a fluidized bed of aluminum oxide. Uranium can be subsequently removed from the bed either by an acid leach or high-temperature fluorination. The Zircex and HF-O2 processes have not been tested on ThO2-containing fuels, but one would expect little reaction between the oxide and the reactant gases.

3.1.2 GRAPHITE-MATRIX ELEMENTS

A fuel which has been developed primarily for the thorium cycle is a dispersion in graphite of spherical particles of sub-millimeter dimensions. Each sphere consists of mixed oxides or carbides of U and Th coated with pyrolytic carbon or silicon carbide, or both, to retain fission products. In an alternative arrangement, which would be advantageous in the early stages of a reactor system when U-235 rather than U-233 was the predominant fissile material, the U and Th would be incorporated in separate particles. The two kinds of particles would be made in different sizes so as to separate them physically during the processing.

Two head-end processes are under development for graphite-matrix fuels: grind-leach and burn-leach. In the first, the fuel elements are crushed very fine to the point where the fuel-containing particles are all individually ruptured; passage between rollers is a possible method. The powder is then leached with fluoride-catalyzed nitric acid to extract uranium and thorium. If fuel were originally present in oxide form, the resulting nitrate solution would be suitable for solvent extraction purification without further treatment. If fuel were originally in the form of carbides, organic compounds might form which would have to be destroyed by a permanganate treatment or the equivalent.

In burn-leach, which seems at present to be the preferred process, the fuel is crushed to a suitable size and oxidized in a fluidized bed of alumina. The oxides of uranium and thorium which are thus produced are then leached out of the alumina with HNO3-HF reagent. Some decontamination from volatile fission products is achieved in the burning process. The principal process problem is decontamination of burner off-gas. Burn-leach methods are not applicable to particles coated with materials which would resist oxidation, such as SiC, Al2O3, or BeO; these have been proposed as coating materials.

Further development is required for both grind-leach and burn-leach processes, but success does not seem to be in doubt. Engineering feasibility studies (55,56) and cost estimates (57) have been made.

3.2 Solvent Extraction

Most proposed solvent extraction processes for uranium-bearing thorium fuels are variations of "Thorex", which is itself a variation of "Purex" which is used for uranium fuels. The organic extractant is a solution of tributyl phosphate (TBP) in a hydrocarbon diluent. The distributions of uranyl nitrate and thorium nitrate between the TBP and aqueous phases are controlled by adjustment of the aqueous concentrations of nitric acid or aluminum nitrate, or both. Initially, conditions are so adjusted that both uranium and thorium go into the organic phase, while most fission products remain behind. Then, advantage is taken of the fact that thorium has, in general, a stronger affinity than uranium for an aqueous phase. The organic solution is treated with an aqueous phase, which is relatively weak in nitric acid, and the thorium transfers into it while the uranium remains in the organic phase. Finally, the uranium itself is scrubbed out of the organic.

Details of the Thorex process have been extensively described (58-62). It can be considered to be technically feasible but requires considerable improvement. One significant defect is that since the maximum capacity of the solvent for thorium is only about half that for uranium, the effective capacity of the equipment is only about half as much for thorium as for uranium. However, a plant specifically designed for the thorium cycle may not require as many extractions, so processing costs for the two fuel cycles may be comparable. A discussion of relative-cost considerations has been made (47).

No provision is usually made for protactinium recovery in solvent extraction processes; it is assumed to have decayed completely to uranium. Protactinium can be extracted by the Thorex solvent, however, along with uranium and thorium, if the aqueous phase has been made sufficiently acid. The high acid level also results in a transfer of zirconium-niobium fission products, and other methods are now favored for removing protactinium from aqueous solutions. It can be adsorbed on manganese dioxide, silica gel, or unfired Vycor glass. From the latter two, the adsorbed protactinium can be eluted with 0.5M oxalic acid.

3.3 Aqueous Alternatives to Solvent Extraction

ORNL has investigated the possible usefulness of peroxide, oxalate, and phosphate precipitation, electrodialysis, and anion exchange processes. None was judged more promising than solvent extraction methods (55,63).

3.4 Fluoride Volatility Processing

The fluorination of oxide, carbide, or metallic fuels is usually done either in a fused-salt medium or a fluidized bed for better control of temperature and reaction rate. The use of fused salts has been demonstrated for zirconium-uranium alloy fuel, which was immersed in NaF-ZrF4 at 600°-700°C and converted to ZrF4 and UF4 by bubbling anhydrous HF through the melt. Subsequent treatment of such a melt would be essentially the same as that described later for the MSBR. Such a process would probably work for a thorium-uranium alloy, provided the composition of the salt were such that the ThF4 formed could dissolve completely and not form a protective film on the alloy. Recovering the thorium from the fused salt would be very difficult. Fluidized-bed fluorination is more versatile. It would be especially suited for following one of the fluidized-bed head-end steps described above. It has been found that an Al2O3 bed of proper specifications will not react appreciably with fluorine under conditions which result in the conversion of uranium to UF6. Consequently, either the material resulting from the fluidized-bed combustion of graphite-matrix fuel or the HF-O2 decladding of metal-clad elements would be an appropriate feed for a volatility process. The UF6 formed in the fluorination would be taken out of the exit gas stream by absorption on an inorganic fluoride.

Recovery of thorium would be difficult and expensive, using fluoride-volatility methods, unless the method could be improved.

4. MOLTEN-SALT FUELS

Two-fluid Reference Design

The volatility process proposed for the MSBR is more highly developed than that for solid fuels (64). The MSBR is a two-region reactor in which the proposed core fluid would consist of 63.6 LiF, 36.2 BeF2 0.23 UF4 (numbers are mole percent) while the blanket, also molten, would be 71.0 LiF, 2.0 BeF2, and 27.0 ThF4. Both core and blanket would be processed continuously via side streams.

The uranium would be separated from the carrier salt and fission products in processing the MSBR fuel stream. The valuable carrier salt would be separated from the rare-earth fission products by the vacuum-distillation process, with about 6.5 percent of the carrier salt either discarded or unrecovered in the distillation process in order to control fission-product buildup and reduce recovery costs. The fuel salt would be reconstituted by absorbing UF6 in uranium-containing carrier salt and reducing it to UF4 by bubbling hydrogen through the melt.

An important factor affecting both the MSBR breeding gain and the fuel cost is the loss of fissile material in processing. There is considerable engineering experience in fluoride volatility processing that indicates an MSBR fissile material loss of 0.1 percent or less per pass through the processing plant.

4.1 Processing the Fuel Stream

The basic processes considered in processing the fuel would involve fluorination, purification of UF6, vacuum distillation, reduction of UF6 and reconstitution of the fuel, off-gas processing, waste storage, flow control of the salt streams, removal of decay heat, provisions for sampling of the salt and off-gas streams, and provisions for shielding, maintenance, and repair of equipment. The major novel pieces of processing equipment include the fluorinator, UF6 reduction equipment, and vacuum-distillation unit. The fluorinator utilizes a frozen wall of salt and a flowing stream of uranium-containing molten-salt is continuously fluorinated. Coolant is used to freeze a layer of salt on the inner surface of the column to protect the structural material from corrosive attack by the molten-salt-fluorine mixture. When reducing UF6 to UF4, barren salt and UF6 enter the bottom of a column, which contains circulating LiF-BeF2-UF4. The UF6 dissolves in the salt, aided by the presence of UF4, and moves up the column, where it is reduced by hydrogen. Reconstituted fuel is taken off the top of the column and sent to the reactor core. In the vacuum-distillation unit, the still is maintained at about 1000°C.and 1 mm Hg pressure. LiF-BeF2 distillate is removed at the same rate that salt enters. Most of the fission products accumulate in the still bottoms and are drained to waste storage when the heat-generation rate reaches a prescribed limit.

4.2 Protactinium Removal

Even though fluoride volatility processing appears to be a satisfactory process for removal of uranium, the ability to remove Pa-233 directly and economically from the blanket region of an MSBR would significantly improve the performance of the reactor. One possible process involves oxide precipitation of protactinium. Several laboratory experiments have demonstrated that protactinium can be readily precipitated from a molten fluoride mixture by addition of thorium oxide, and that the precipitate can be returned to solution by treatment with HF. Experimental results also indicated that treatment of protactinium-containing salt with ZrO2 leads to oxide precipitation of the protactinium and that after beta decay of the protactinium, the resulting UO2 will react with ZrF4 to give UF4.

More recent experimental results have indicated another method for removing protactinium directly from the blanket fluid. This involves treating the molten blanket salt with a stream of bismuth containing dissolved thorium metal. The thorium reduces the protactinium (and also any uranium) to metal, which can then be accumulated on a stainless-steel-wool filter, or recovered directly from the liquid metal. The metal can be hydro-fluorinated and/or fluorinated to return the protactinium (and any uranium) to the fuel-recycle process as the fluoride. Thus there is experimental evidence that direct processes are available for removal of protactinium from the blanket stream of molten-salt reactors.

If protactinium is not removed directly from the blanket stream, then the blanket salt is processed by the fluoride-volatility process alone. Any uranium not removed during the blanket processing would be returned to the blanket and removed by subsequent processing.

5. GENERAL CONSIDERATIONS

Only statements of a qualitative nature can be made relative to the economic and technical impact of processing thorium fuels because of the developmental nature of this fuel cycle. It is believed that reprocessing of thorium fuels is technically feasible in several ways; and that the reprocessing cost, in mills/kWh, should not be significantly different than for the U-Pu cycle, provided that the two cycles are compared on the same terms, i.e., the sizes of plant should be about the same, and if "equilibrium" fuel is used in one case, it should be used in the other also. Isotopic impurities are often mentioned as a liability of U-233, but the plutonium cycle will also accumulate them, and will require at least semi-remote fabrication (70). While Pu-239 emits only a small amount of hard radiation, its emission is the source of very high toxicity if ingested. While the plutonium fuel could be handled in a glovebox to guard against hazards, long irradiations and high burnups, on the other hand, produce substantial amounts of other isotopes with a consequent increase in the radiation level. The gloveboxes would then no longer be adequate and shielding would be required around the equipment and remote control techniques would be used in handling the fuel. The absolute amount of actinide activity present would be a function of the isotopic ratio and cooling time before and after processing.

If thorium fuels are reprocessed in equipment designed for uranium, or with processes such as Thorex which use reagents developed for a different system, complications will occur and good economy cannot be expected. Processing of thorium fuel should be based on plants designed for that specific purpose and an intensive research program on new separation methods would be desirable and economically justified.