1.6 Thorium Valence States

The tetravalent state, exemplified by compounds such as ThOj, is the only valence state of thorium of practical importance.

1.7 Thorium Dioxide

Thorium dioxide Th02 is the form in which thorium is proposed for use as reactor fuel for light-water, heavy-water, and liquid-metal fast-breeder reactors. It is a stable ceramic that can be

Source: J. R. Murray, “The Preparation, Properties and Alloying Behavior of Thorium,” Report AERE-M/TN-12, 1952.

heated almost to its melting point of 3370°C without serious deterioration. Physical properties of Th02, from reference [II], are summarized in Table 6.8.

ThO2 forms solid solutions with U02 or Pu02 over the entire composition range from 0 to 100 percent Th02.

Th02, either as the mineral thorianite or as synthetic thoria produced by heating thorium nitrate, oxalate, or hydroxide, reacts only slowly with mineral acids. It can be dissolved in hot, concentrated sulfuric acid or in hot nitric acid containing 0.05 M HF.

1.10 Methods

Lustman and Kerze [LI], pp. 115-129, list a number of processes that have been used on the laboratory scale for separation of zirconium and hafnium. Of these, three that have been used on an industrial scale are

1. Fractional crystallization of double fluorides

2. Solvent extraction of the thiocyanates by hexone

3. Solvent extraction of the nitrates by tributyl phosphate, TBP

These processes are described in Secs. 7.2 through 7.4. Another recently patented [М2, М3] process,

4. Selective reduction of the molten double fluorides by aluminum dissolved in molten zinc seems promising and is described in Sec. 7.5.

1.11 Fractional Crystallization

Solubilities of corresponding salts of hafnium and zirconium are compared in Table 7.11.

Fractional crystallization of the double potassium fluorides was the method originally used to separate hafnium from zirconium. Because these salts form solid solutions and the ratio of solubilities is close to unity (1.54 at 20°C), multiple recrystallizations are necessary for the necessary completeness of separation. In the United States, Kawecki [Kl] has found that 10 recrystallizations of the double potassium fluorides reduced the hafnium content of zirconium from 2.0 to 0.1 percent.

A similar process has been used in the Soviet Union [SI], where the operating temperatures and solubilities were as follows:

After 16 to 18 recrystallizations, the hafnium content of zirconium was reduced to 0.003 percent. The yield of zirconium was about 80 percent.

Because of the large number of independent steps, this fractional crystallization process has been superseded by the solvent extraction processes next to be described.

Oxidation-reduction potentials for plutonium. Oxidation reduction potentials [Al] for plutonium ions are given in Table 9.5..

Other potentials may be obtained as linear combinations of these. For instance,

Pu4+ + 2H20 ->■ Pu022+ + 4ІГ + 2e’

Ett = — y’— = -1.0433 V (9.29)

Equilibria for the reduction of Pu02* or Pu02 2+ to Pu3+ or Pu4+ depend on the fourth power of the hydrogen ion concentration. Increasing acidity displaces equilibrium toward the reduced states.

If the salts and acids in a given system were completely dissociated into simple ions with no complex formation and if the activity coefficients were always the same, oxidation-reduction potentials would be independent of the anionic or cationic species present. However, as we saw in Sec. 1.1, these conditions are rarely obtained with the higher-valent actinides. Therefore, oxidation-reduction potentials for the actinides are usually measured in perchlorate solutions, where relatively little complexing and a high degree of dissociation occur [Al, SI]. These are referred to as “formal potentials,” which are more accurate for practical calculations. To obtain standard potentials these formal potentials, or equivalent thermodynamic data, must be extrapolated to zero ionic strength. Formal potentials for other solutions, such as 1 M HN03 or 1M NaOH, are also reported in the literature [А1]. As pointed out in Sec. 1.3, in a practical system such as nitric acid solutions the equilibria estimated from oxidation-reduction potentials must be corrected for complexing, as well as incomplete dissociation. For example, the oxidation potential for the couple Pu(III)-Pu(IV) is—0.9819 V in 1 M HC104 but —0.92 V in 1 M HNOj

[SI]-

Oxidation-reduction potentials of the actinides. The formal potentials for transition between the valence states of the actinides are listed in Table 9.6.

The stability of an intermediate oxidation number against disproportionation can be obtained as follows. Consider the disproportionation of U(V) according to the following reaction:

Uv02+ -*■ UV1C>22+ + e" £l,6 =-0.063 V (930)

Ц^ + гНіО-* Uv02* + 4H+ + e’ £4,5 = -0.613 V (931)

2Uv02+ + 4H+-UVI022t + U4+ + 2H20 AE = -0.063 -(-0.613) = 0.550 V (9.32) The equilibrium constant for the overall reaction (9.32) is

£ = e38.93(-0.063 + 0.613) = 1 99 X 109 (!) .33)

Hence, pentavalent uranium is unstable in aqueous solution at [H+] > 1.

Disproportionation of valence n to valences n + 1 and n — 1 will proceed spontaneously at pH = 0 if the potential for the oxidation from n to n + 1 is larger or more nearly positive than the potential for oxidation from n — 1 to n. Applying this criterion to the data of Table 9.6,

Table 9.5 Oxidation-reduction potentials for plutonium

we see that in the tetravalent state uranium, neptunium, and plutonium are stable. In the pentavalent state protactinium, neptunium, and americium are stable (cf. Table 9.4).

The positive potential for U(III)-U(IV) indicates that the unstable U(III) would be rapidly oxidized by water in aqueous solution. The relatively low negative potentials for the oxidation of U(III) through intermediate states to U(VI) indicate that the latter should be quite stable in aqueous solutions. The elements of higher atomic number become progressively more difficult to oxidize to the hexavalent state.

Solutions containing U(V1) and Pu(III) or Pu(IV), as used in aqueous separation processes, are stable against oxidation of plutonium by uranium because the potentials for the transitions U(TV) to U(VT) and U(V) to U(VI) are more nearly positive than the plutonium potentials. Plutonium may be reduced from Pu(IV) to Pu(III) without affecting uranium oxidation by choosing a reducing agent, such as Fe2+, whose oxidation potential is less negative than the —0.9819 V required for Pu(IV) reduction and more negative than the —0.338 V that would reduce U(VI).

The data indicate that U(VI) should oxidize Np(III) to Np(IV).

Oxidation-reduction potentials for couples consisting of the actinides or the fission products in acid solution (1 M HC104) are listed in Table 9.7. Potentials for a selected group of oxidizing and reducing agents are listed in Table 9.8. The couples are listed in order of decreasing strength as reducing agents. In the cases where the molecular and ionic species involved in a given valence transition are different in acidic and basic solutions, the acid system (1 M НСЮ4) has been chosen.

The oxidation-reduction schemes of the more important multivalent elements encountered in aqueous fuel reprocessing are summarized in Fig. 9.1.

Rate of oxidation-reduction reactions. Oxidation-reduction reactions that involve only the transfer of electrons from one uncomplexed ion to another in an ionizing solvent are reversible and, for all practical purposes, instantaneous. Equation (9.8) is an example. On the other hand, reactions involving molecular rearrangements, even though thermodynamically possible, may be

very slow or may not go at all. A familiar example of a slow reaction is the gradual approach to the end point in titration of ferrous ion with permanganate ion in acid solution:

8H4 + 5Fe2+ + Mn04‘ -*■ 5Fe34 + Mn24 + 4H2 О (9.34)

The oxidation of an actinide ion from M3+ or M4+ to M024 or M0224, or its reduction from M(V or VI) to M(IV or III), is inconveniently slow, apparently because of the sluggish combination of M and О in the oxidation step or the slow breaking of the M—О bond in the reduction step. Of the three couples involving plutonium, Eq. (9.27) is very slow, whereas (9.26) and (9.28) are practically instantaneous. Further discussion of the rate of oxidation — reduction of plutonium solution appears in Sec. 4.6.

The unusual property of uranium, neptunium, and plutonium of forming volatile hexafluorides has led to extensive work on fluoride volatility processes for separating these elements from irradiated fuel and from each other. Major programs were carried out at Brookhaven, Argonne, Oak Ridge, and European laboratories. These programs have been summarized by Jonke [J2], Barghusen et al. [Bl] and Schmets [S2],

Brookhaven made engineering-scale studies of a process in which uranium metal fuel was dissolved in a liquid interhalogen compound such as BrF3. The reaction was difficult to control; work was terminated after an explosion [В18]. Brookhaven later developed the Nitrofluor process [B19], in which fuel was converted to UF4 and PuF3 by a liquid mixture of HF and oxides of nitrogen. After dissolution, UF4 was converted to UF6 by BrF3 and distilled off. Finally, PuF3 was converted to PuF6 by fluorine and distilled off.

Gas-phase fluorination reactions were studied at Argonne [Bl] and in Europe [С13]. Fuel was first oxidized to U308 and Pu02. The crushed oxides were charged to a fluidized bed of alumina through which gases containing fluorinating agents, F2, C1F3, or BrFs, were passed. Uranium was readily separated as volatile, stable UF6. Separation of neptunium and plutonium was less satisfactory. Although these also form volatile hexafluorides, they are less stable than UF6. Stronger fluorinating conditions are needed to form them, and PuF6, in particular, is so unstable that it tended to decompose and deposit solid fluorides throughout the equipment.

Experience has shown that fluoride volatility processes are most useful when applied either to fuel containing little plutonium and neptunium or to fuel from which these elements have been largely removed by other processes. Oak Ridge National Laboratory has successfully separated and purified multikilogram amounts of irradiated, highly enriched uranium relatively free of plutonium from zirconium-235 U fuel used in submarine reactors [03], from aluminum — 235 U fuel used in research reactors [04], and from the mixture of fused salts used in the aircraft reactor experiment [С2]. More recently, 235 UF6 and 233 UF6 were recovered from the BeF2-7LiF-UF3 fuel melt used in the molten-salt reactor experiment [L3], This work led to design of a process to separate 233U and fission products from the BeF2-7LiF-UF3-ThF4 mixture proposed as fuel for the molten-salt breeder reactor nuclear power system [R9].

In the Aquafluor process [G4] developed by the General Electric Company, most of the plutonium and fission products in irradiated light-water reactor (LWR) fuel are separated from uranium by aqueous solvent extraction and anion exchange. Final uranium separation and purification is by conversion of impure uranyl nitrate to UF6, followed by removal of small amounts of PuF6, NpF6, and other volatile fluorides by adsorption on beds of NaF and MgF2 and a final fractional distillation. A plant to process 1 MT/day of irradiated low-enriched uranium fuel was built at Morris, Illinois, but was never used for irradiated fuel because of inability to maintain on-stream, continuous operation even in runs on unirradiated fuel. The difficulties at the Morris plant are considered more the fault of design details than inherent in the process. They are attributed to the attempt to carry out aqueous primary decontamination, denitration, fluorination, and distillation of intensely radioactive materials in a close-coupled, continuous process, without adequate surge capacity between the different steps and without sufficient spare, readily maintainable equipment [G5, R8].

The processes described in Sec. 4.6 are, in principle, applicable to off-gases from LMFBR reprocessing plants. The problem is the greater iodine activity per ton of fuel processed. This is due to the 60 percent higher specific activity of iodine for 150-day cooled LMFBR fuel compared with similar LWR fuel noted in Table 10.20 and the incentive to reprocess LMFBR fuel with shorter cooling. Oak Ridge National Laboratory has estimated that if LMFBR fuel were to be reprocessed only 30 days after irradiation, an iodine retention factor as high as 1010 would be required. This seems completely unattainable. However, some improvement over the retention factor of 102 (99 percent retention), feasible with the technology described in Sec. 4.6, would be possible if iodine could be stripped more completely from the dissolver solution. Retention factors of 104 or better have been reported for individual silver-zeolite absorbers.

The waste repository will be the final reservoir for all radioactivity generated by nuclear power. It will remain radioactive for a very long time, with some radioactivity even remaining for millions of years. As yet, complete safety analyses of waste repositories are not available. However, several attempts to approach the problem are known, and a number of systematic programs are on their way in various countries.

Proceeding from the assumption that water will be the only vehicle that possibly can carry radioactive material from the repository to people, the following processes must take place to create an actual risk:

1. The geologic containment fails and water is allowed to enter the repository and to find its way to the solidified waste, or brine present in the repository may contact the waste.

2. Radioactivity is released from the repository through contaminated water or brine entering an aquifier which is connected to circulating groundwater.

The magnitude of the consequences will obviously be a function of the radioactive inventory of the waste repository at the time when the sequence starts. As this inventory decreases by natural decay, the consequences will also decrease and will eventually drop below the level of significance.

Types of solvent extraction contactors in general commercial use for nonnuclear applications include [LI, М3, T2]:

1. Spray columns

2. Baffle-plate columns

3. Perforated-plate columns

4. Columns packed with Raschig rings or Berl saddles

5. Mixer-settlers

Some of this conventional equipment has been applied to the solvent extraction purification of natural uranium and thorium. All of these conventional gravity column contactors are less compact than is desirable for reprocessing irradiated reactor fuel. The height of a vertical — column gravity contactor equivalent to a single equilibrium stage of contacting is about 60 to 120 cm, so very tall columns would be needed for the 10 or more theoretical stages needed for some of the separations in fuel reprocessing.

The need for compact contactors in reprocessing irradiated reactor fuel, as well as the need for many stages and small inventory in purifying special organic materials and pharmaceutical products, has stimulated the development of solvent extraction contacting equipment with reduced holdup, reduced stage height, or both. Most of these devices have in common the input of mechanical energy to promote contacting of phases, separation of phases, and/or counter­current flow. Contactors that have been used in reprocessing irradiated reactor fuel are listed in Table 4.12, adapted from a compilation by Davis and Jennings [D1 ]. Additional high- performance contactors that have been used in some nuclear applications and in the pharmaceutical industry include the Fenske-Long extractor, a vertical stack of mixer-settlers, and the Podbielniak centrifugal extractor [Т2].

The contactors used in nuclear applications are described in the following sections.

As an example of the acid-leaching, solvent extraction class of uranium-concentration processes, a description will be given of the process used in the large uranium mill of the Kerr-McGee Corporation at Grants, New Mexico. This has been condensed from a 1960 paper by Hazen

tYellow cake is the name conventionally used for uranium ore concentrates.

[H4] and from Merritt’s [М3] account of operations in 1971. At that time the mill’s capacity was 5000 short tons ore (4500 Mg) per day. In 1978 its capacity was 6200 short tons per day. Figure 5.5 shows this mill.

Leaching operations in the Kerr-McGee mill are described in this section, with reference to Fig. 5.6. Recovery of uranium from leach liquor by solvent extraction with organic amines in the Amex process is to be described in Sec. 8.6.

The ore processed in the Kerr-McGee mill is primarily a sandstone, with uranium minerals in the material bonding the sand grains. The ore contains about 0.2 w/o U308, 0.01 to 0.03 w/o Mo03, and 0.05 to nearly 0.20 w/o V205. Uranium and molybdenum are leached and recovered, with uranium recovery exceeding 97 percent. Some vanadium is also leached, but was not being recovered in 1971. The ore also contains acid-soluble calcium minerals, equivalent to from 2 to 5 w/o CaO, which are the principal consumers of sulfuric acid.

Crushing and grinding. The ore is first crushed dry to particles smaller than 1 in. Crushed ore is then processed in two parallel, identical systems, of which circuit A is shown in Fig. 5.6. In each circuit 2500 t per day of crushed ore is ground with heated water in rod mills until 97 to 98 percent passes 28 mesh, with 70 percent coarser than 150 mesh.

Leaching. Slurry from the rod mills flows in series by gravity through 14 rubber-lined steel leaching tanks 13 ft (4 m) in diameter and 14 ft (4.25 m) high, equipped with turbine agitators. The holding time in the 14 tanks is around 4.5 h. The rod-mill slurry fed to the first tank is mixed with recycle water containing slimes, and steam and sulfuric acid sufficient to bring the temperature in the first tank to 43 to 54°C and the pH to 0.6 to 0.7. Here the most readily dissolved minerals react, and gaseous reaction products such as C02, H2, and H2S are liberated.

To leach the more acid-resistant minerals containing tetravalent uranium, steam is fed to the second tank to bring the temperature to 49 to 60°C, and sodium chlorate NaQ03 is added to bring the oxidation-reduction potential e, measured relative to the calomel electrode, to from —0.47 to —0.51 V. At —0.51 V, the equilibrium ratio of ferric iron to ferrous iron in the solution is 0.52.^ Ferric iron catalyzes the oxidation of insoluble tetravalent uranium to the soluble hexavalent uranyl form:

^Oxidation-reduction potentials and their effect on the valence state of materials being processed are explained in Chap. 9. Because the emf of the saturated KCl-calomel electrode relative to the standard hydrogen electrode is -0.244 eV at 25°C ([М3], p. 70), the relation between the emf e relative to the saturated calomel electrode and the emf E relative to the standard hydrogen electrode used elsewhere in this text is E° = e — 0.244. Thus, the emf in the second tank relative to the standard hydrogen electrode is from -0.714 to -0.754 V.

U4* + 2Fe3+ + 2H20 -*■ UVI02Jt + 2FeJ+ + 4H+

Addition of sodium chlorate to the second tank instead of to the first avoids consumption of this relatively expensive material by metallic iron introduced in grinding or by reducing gases, such as Hj or Ц* S, which are vented from the first tank.

As the hot, acid, oxidizing slurry flows through the remaining 13 tanks, dissolution of the more resistant uranium and molybdenum minerals is completed. After 4.5 h, when the slurry leaves the fourteenth tank, its temperature has dropped to 43 to 54°C, acid has been consumed with pH increased to 0.9 to 1.2, and sodium chlorate has been used up in oxidizing uranium, with the oxidation-reduction potential relative to the calomel electrode in the range —0.41 to —0.43 V. This is sufficiently negative to convert substantially all soluble uranium to the hexavalent state, while leaving only 2 percent of the iron oxidized to ferric, thus minimizing consumption of sodium chlorate.

In some other U. S. mills and in South Africa and Australia, manganese dioxide Mn02, syn­thetic or in the form of the mineral pyrolusite, is used as oxidant. Typical oxidant consumption is 3 lb (1.5 kg) NaC103 or from 3 to 6 lb (1.5 to 3 kg) Mn02 per short ton of ore.

Sulfuric acid consumption depends on the amount of reactive minerals present in the ore. In the United States, acid requirements range from 40 to 120 lb (20 to 60 kg) H2SC>4 per short ton of ore.

liquid-solid separation. Separation of the slurry leaving leach tank #14 into (1) tailings relatively free of uranium-containing liquid and (2) uranium-bearing leach liquor free of suspended solids is shown at the bottom and right of Fig. 5.6. A rough separation of slurry from leach tank #14 at about 150-mesh particle size is made in two 20-in (0.5-m) diameter cyclone separators in parallel. The coarse fraction from the cyclones passes through five rake classifiers in series, where the sand is washed countercurrently with acid-bearing aqueous raffinate from the solvent extraction system, Fig. 5.9. The fines fraction from the cyclone separators and the overflow from #1 rake classifier are combined and washed countercurrently with additional raffinate in six large countercurrent decantation thickeners 120 ft (36 m) in diameter and 17 ft (5 m) deep. Wash ratios are 2.5 to 3.0 in the classifiers and 3.0 to 4.0 in the thickeners. This recycle of raffinate returns H2S04 from the solvent extraction system to the washing circuits and maintains the pH in them at 1.5 or less, thus preventing precipitation of uranium during washing. Tailings from #5 rake classifer are about 75 v/o (volume percent) solids; from #6 thickener, about 31 v/o. The thickeners handle about 25 percent of total solids. Soluble uranium losses in the washed tailings represent only about 0.2 percent of the uranium in mill feed.

Overflow from #1 thickener contains 150 to 200 ppm of solids. This is reduced to 50 to 75 ppm by addition of flocculant and settling in a 60-ft (18-m) diameter by 20-ft (6-m) deep reactor-clarifier. Fine slimes from the clarifier are recycled to leach tank #1.

Overflow from the clarifier is combined with a similar stream from leaching circuit В and passed through six 600-ft5 (58-m2) U. S. pressure filters in parallel, each operated at a feed rate of 300 to 500 gal/min (1.1 to 1.9 m3/min). Filters are precoated with about 0.1 lb (0.05 kg) of precoat per short ton of ore, and 0.35 lb (0.17 kg) of filter aid per ton is added to the filter feed. The filter loading cycle lasts from 4 to 24 h, depending on the solids contents of clarifier overflow. Filter cake is returned to #2 or #3 thickener. Filtered leach liquor containing about 1 g U308/liter is the product of the leaching system and the feed to the solvent extraction system.

A flow sheet of the caustic soda process described by Bearse et al. [B2] is shown in Fig. 6.4. The sand is ground with water in a ball mill until 96.5 percent passes 325 mesh. A wet classifier recycles coarse particles and delivers a slurry of fine particles to a stainless steel reactor. A liquid caustic solution containing 73% NaOH is fed to the reactor. At the beginning of the reaction the slurry contains 1.5 kg of NaOH and 1.7 kg of water per kilogram of sand. The mixture is heated to 140°C, and after 3 h at this temperature the sand is completely reacted. The mixture is then diluted with the wash solution of caustic and trisodium phosphate from a later step and digested at 105°C for 1 h to facilitate later filtration. The resulting hot slurry contains practically all the original phosphorus in solution as trisodium phosphate, and thorium, cerium, and rare earths are present as solid hydrous metal oxides. The trisodium phosphate and unreacted caustic are removed by filtering the slurry through Monel wire cloth, and the metal oxide cake is washed with water. The filtration is carried out at 80°C to keep caustic and trisodium phosphate in solution. The filtrate, which contains about two-thirds of the original caustic soda charged, is evaporated in an open steel kettle until the NaOH

concentration is 47%, corresponding to a boiling point of 137°C. The concentrated solution is cooled to room temperature. More than 95 percent of the sodium phosphate crystallizes out of solution and is removed by filtration. The caustic soda liquor is recycled for sand digestion and for later neutralization steps.

The hydrous oxide cake is brought into solution by dissolving in 37% hydrochloric acid (1.5 kg acid/kg sand) at 80° C for 1 h in a glass-lined vessel. Hydrochloric rather than sulfuric acid is recommended because of more selective precipitation from chloride solutions in the later step. About 2 percent of the weight of the original sand is left as residue, which contains undissolved monazite and rutile (Ti02), an impurity in the sand.

The acid solution and undissolved material are transferred to a neutralizer vessel and diluted with water.

Thorium is separated from the rare earths by selective precipitation of thorium hydroxide at a pH of 5.8. This is effected by neutralizing the diluted chloride solution with caustic recovered from the evaporator. The wet cake is reslurried in water solution, filtered, and again reslurried and filtered to effect a high degree of separation of the thorium precipitate from any occluded rare earth solution.

The percentage of thorium and other constituents of the monazite sand recovered in the precipitate is given in the first column of Table 6.17. The composition of the precipitate is given in the second column.

Rare earths are recovered from the combined decantates and filtrates by further neutraliza­tion with NaOH. The hydroxide precipitate is removed by filtration.

The process used in Brazil [B6] is generally similar.

Also shown in Table 8.2 are the effective thermal cross sections for the individual nuclides, calculated for the neutron spectrum of a typical PWR and including the contributions from resonance absorption. The cross sections are multiplied by the atoms per fission-product pair to obtain the effective cross sections per fission-product pair listed in Table 8.2. Although the total effective cross section of 89.2 b/fission-product pair is calculated for the mixture of radionuclides existing 150 days after fuel discharge, it is a good approximation for the effective

^Acts both as an acid and as a base.

8 Effective thermal cross sections for a typical neutron spectrum of a PWR, including contribu­tions from nonthermal resonance absorption.

1 Total yield of element whose principal radionuclides are listed above

cross section for all fission products other than 135Xe at the time of fuel discharge. Except for 135Xe, the shorter-lived species that are also present at the time of discharge do not exist in sufficient concentration to contribute appreciably to neutron absorption. Neutron absorption in 13SXe is usually treated separately, by the techniques discussed in Sec. 6.3 of Chap. 2.

The elemental contribution to neutron absorption by fission products tends to follow the effective fission yield of the elements, but with exceptions for several individual elements. The rare earths neodymium, promethium, samarium, europium, and gadolinium, as well as xenon and cesium, are the important neutron-absorbing elements resulting from the high-mass fission-yield peak, and rhodium and its near neighbors are the important neutron absorbers from the low-mass peak.