The hydrofluorination reaction to convert U02 to UF4,

U02 + 4HF ^ UF4 + 2H20

is exothermic. It proceeds rapidly at 500°C, but the equilibrium mixture of H20 and HF contains around 35 percent HF. At 300°C, nearly complete utilization of HF can be obtained, but the reaction rate is slow. Problem 5.3 illustrates calculation of the HF content of the equilibrium mixture of HF and H20 at these two temperatures from free-energy data.

In U. S. plants hydrofluorination is carried out in two stirred fluidized-bed reactors in series, with counterflow of solids and gases. The bed to which U02 is fed and from which exhaust gases are discharged runs at 300°C, partially converts U02 to UF4, and reduces the HF content of the effluent gases to around 15 percent. The bed to which anhydrous HF and the partially converted U02 are fed runs at 500°C and converts more than 95 percent of the U02 to UF4. To prevent caking of the fluidized beds, it has been found necessary to provide each reactor with a vertical-shaft, slow-speed stirrer to scrape the reactor walls. Production rates around 700 to 900 kg/h are obtained in 0.75-m-diameter reactors. Effluent gases are filtered to remove entrained solids, cooled to condense aqueous HF, and scrubbed to remove the last traces of HF.

In the Comurhex plant at Malvesi [B5], reduction of U03 and conversion of U02 to UF4 are carried out in a single L-shaped, moving-bed reactor. Reduction takes place in a vertical section and hydrofluorination in a horizontal section. Practically complete utilization of HF is obtained.

Zirconium is the eleventh most abundant element in the earth’s crust, which contains 0.028 percent of this element. It is more abundant than copper, lead, nickel, or zinc. Zirconium minerals always contain from 0.5 to 2 percent of chemically similar hafnium, which seldom occurs naturally by itself.

The principal natural sources of zirconium and hafnium are the minerals zircon (Zr, Hf)Si04, and baddleyite (Zr, Hf)02.

1 PRODUCTION AND PRICE

Table 7.1 gives the annual production of zirconium concentrate by the principal producing countries of the non-Communist world, excluding the United States.

U. S. production in these years was around 150,000 short tons. Thus, Australia and the United States are the principal zirconium-producing nations. Most of their production was from dredging of black sands on beaches and in stream beds, where zircon has been concentrated hydraulically along with other relatively dense minerals such as rutile (Ti02), ilmenite (FeTi03), and monazite (Chap. 6).

Principal U. S. producers of zircon concentrates in these years were E. I. du Pont de Nemours and Company and Titanium Enterprises, Inc., with operations primarily in northern Florida and southern Georgia.

In the 1960s, hafnium-free zirconium had been produced by several U. S. companies

including Amax, Inc., National Distillers and Chemicals Corporation, Columbia-National Cor­poration, and Wah Chang Corporation, but in 1978, the only U. S. producer of hafnium-free zirconium sponge was Teledyne Wah Chang Albany Corporation, with annual capacity of 7.5 million lb [N2]. Western Zirconium then announced plans to produce 3 to 4 million lb/year. In France, Pechiney Ugine Kuhlmann was increasing capacity to 4 million lb/year. Indian zirconium production capacity was around 0.1 million lb/year. Hafnium-free zirconium has also been produced in England, Canada, Japan, West Germany, and the Soviet Union.

Prices in 1974 were [A2]

Zircon concentrate Zirconium, hafnium-free Sponge

Sheets, strip, bars Hafnium Sponge Bar and plate

The principal sources of activity in irradiated and chemically purified thorium are 234 Th and its short-lived daughter 234mPa, and 228 Th and its daughters. Beta and gamma activity from these

tCorresponding to a weekly total of 0.100 rem for continued exposure.

nuclides constitute the greatest danger in external exposure; neutrons from (a, n) reactions with light contaminants are relatively unimportant in this regard. Prediction of activities due to 234Th is similar to the analyses of 237U activity in Sec. 2.2.

Nuclides in the 234 Th chain reach equilibrium concentration during irradiation exposures of a few months or greater, with the concentrations given by

Nоз^оз ~ МогОоіФ (8.28)

and ^04X04 = Маіо0зФ (8.29)

where 002 is the equivalent thermal cross section for (n, 7) reactions in 232 Th, and is greater than the true thermal value to allow for absorption of resonance neutrons. By combining Eqs. (8.28) and (8.29), the equilibrium concentration of 234Th is

■Wot _ ОсдОозФ2 ,й ,пч

JV02 Л04Л0З

If the 232Th is irradiated in a neutron flux with a negligible component above 6.37 MeV so that no 232U-228Th are formed, postirradiation cooling can reduce the beta activity to a tolerable level. Even if 228Th is present, preprocessing decay of 234Th may be useful to aid beta decontamination of the separated thorium product. From Eq. (8.32) the time required for the 234Th-234Pa beta activity to reach the beta activity of natural thorium of 4.37 X 1СГ7 Ci/g is given by

Tc = 34.8 In (6.17 X 1O2OOo20o302) days (8.33)

where аф are expressed in reciprocal seconds. ‘

For the uranium-thorium-fueled reactor of Fig. 3.33, am = 6.1 b, о0з = 520 b, and 0= 1.2 X 1014 rt I (cm2 ‘s), resulting in 234Th-234Pa beta activity at discharge of

®^ = 1.2X 10-2 Ci/g 232 Th Л’02

The time for this to decay to the equilibrium beta activity of the 232Th daughters is

Tc = 356 days

Cooling for this length of time will ensure that in chemical reprocessing thorium can undergo total beta decontamination to twice the level of natural 232 Th. The decontamination can be verified with total beta monitoring. For shorter cooling times beta discrimination techniques must be used to ensure that long-lived beta contaminants are not present in the separated thorium.

218 Th is also present in Eradiated thorium and is accompanied by beta-emitting daughters in its decay chain. These daughters are removed from thorium in chemical reprocessing, but they appear again in the separated thorium, growing with a time constant of about 4 days. Thereafter, the beta activity in the separated thorium approaches the level in secular equilibrium with 228 Th. It is therefore important that monitoring for beta decontamination of thorium separated in fuel reprocessing be carried out promptly after the separation is performed.

In aqueous solution curium exists in the oxidation states Cm(III) and Cm(IV). Solutions of Cm(IV) can be prepared only by dissolving CmF4 in a solution containing a high concentration of fluoride to form the stabilized complexes CmFs~ and/or CmF6 2". As shown by the oxidation — reduction potentials of Table 9.6, Cm(IV) is a strong oxidizing agent. In the absence of sufficient complexing for stability, it is rapidly reduced to Cm(III) under the influence of the alpha activity from curium decay. Trivalent curium forms complexes with СГ, N03", S04 2~, and C2 04 2~ which are less stable than those with americium, but the CNS’ complex of curium is more stable than that of americium. Organocomplexes and chelates are also formed [K2].

242 Cm is recovered from irradiated Am02/Al cermets by dissolution of the aluminum in hot NaOH, followed by dissolution in 6 M HC1. The curium and americium are separated from fission products by anion exchange from 11 M LiCl and elution with 12 M HQ, and the curium is then separated from the americium by selective elution from a cation-exchange column with lactic acid solution [H4].

244Cm is recovered from irradiated Pu/Al alloys and Am02(Pu02)/Al cermets by dissolution, extraction of plutonium with TBP in и-dodecane, extraction of americium and curium from the aqueous raffinate with 50 percent TBP in kerosene, purification of the americium and curium fraction by extraction with tertiary amines, and separation of americium by precipitation of the double carbonate K5 Am02(C03)3 [G5]. A high-pressure ion-exchange system for the separation of curium, americium, and rare earths from feed solutions of dilute HN03 has been applied at the Savannah River Laboratory [HI, L5].

Unlike irradiated U02 fuel, which dissolves readily in hot nitric acid, irradiated Th02-U02 fuel dissolves only very slowly and incompletely in this reagent. After extensive research, Oak Ridge National Laboratory concluded [B14] that the best reagent for producing a nitrate solution from Th02-U02 fuel was a mixture of nitric and hydrofluoric acids. This reagent has two serious drawbacks:

1. Mixed nitric and hydrofluoric acids react also with stainless steel and zircaloy, so that both the cladding and the stainless steel dissolver itself are attacked

2. Dissolution of Th02 is much slower than that of U02 in nitric acid.

Oak Ridge found [El] that corrosion of 304L and 309SCb stainless steel by the mixture of HN03 and HF could be reduced to an acceptable level by addition of aluminum nitrate, A1(N03)3, to the mixture of HN03 and HF without decreasing the rate of solution of Th02 by more than 20 percent. The aluminum nitrate acts by complexing the fluoride ion. The composition recommended for the solvent is 13 Af HN03, 0.05 M HF, 0.1 M A1(N03)3. Reddick [R6] has summarized the extensive research that has been conducted on dissolution of Th02 in HN03-HF-A1(N03)3 mixtures. This procedure still has drawbacks. Parts of a stainless steel dissolver exposed to hot HN03-HF vapors, which contain no complexing A1(N03)3, are not protected and will corrode. Aluminum nitrate is only partially effective in preventing reaction of HF with zircaloy cladding. Aluminum nitrate increases the volume of nonvolatile solids in the waste.

The rate of dissolution of Th02-U02 fuel is higher the lower the density of the fuel and the smaller the particle size. For example, with 200 percent stoichiometric excess of reagent, from 25 to 40 h were required to dissolve completely pellets 0.66 cm in diameter having 90 to 95 percent of theoretical density, whereas in 5 h, 99 percent of such fuel dissolved when the fuel was first crushed to under 100 mesh [ВІЗ]. Fuel with only 60 percent of theoretical density dissolved almost 10 times as fast as fuel having 90 percent density.

A broad spectrum of processes to solidify HLW has been considered in various countries over the last two decades. The more urgent the need for an operational process has become, the more has this spectrum narrowed. Attention is now focused worldwide on a few types of vitrification processes with a strong preference on those for borosilicate glass, and on a fluidized-bed calcination process. The latter yields granules of calcine as the primary product. It must be consolidated for final disposal, preferably by mixing with molten glass. Other products under investigation are considered long-term developments rather than present technology.

Besides the alternatives concerning material and shape of the product discussed in the last paragraph, there are a few characteristic alternatives among the process parameters that may serve to classify solidification processes.

Glass melting. Glass melting may be performed either in a continuously fed melter with discontinuous discharge of the melt into a storage canister or directly in the storage canister (in-can melting). The continuous melter may be a joule-heated ceramic melter or a furnace — heated metallic melter.

In-can melting is the simplest choice as far as the melting device is concerned. No replacement or repair of a melter is necessary, and the potentially troublesome melt drain is avoided. On the other hand, the capacity of a canister is limited compared to a melter. Therefore parallel melting units are required with a complex technique to divert the feed from one canister to another.

Among continuous melters, the ceramic melter is favored over the metallic one, usually made of Inconel, because of its better corrosion resistance. The ceramic melter will have a longer life than the metallic one, and it may be the only practical device with sufficient corrosion resistance for processing high-temperature glasses if they are desired for the sake of improved long-term stability. On the other hand, remote replacement of a bulky ceramic melter is a more difficult task than that of a metallic crucible.

Joule heating is practically a requirement when a ceramic melter is employed. This means dissipating electrical energy in the molten glass between immersed electrodes. Joule heating has been shown to be feasible with sufficiently refractory electrode material such as molybdenum or even tin oxide. An auxiliary heating system has to be provided for initial start-up and for restarting.

A separate melter, particularly a ceramic one, is more flexible with respect to the feeding technique than a canister, mainly because of its greater surface. Furthermore, a continuous melter leaves the option either to fill a canister with the molten glass or to be coupled to a glass-shape forming device. This, for example, can produce beads to be embedded in a metal matrix.

Feed to die melter. The feed to the melter or to the canister may be liquid waste or a calcine with the glass frit either added to the waste or as a separate stream to the melt.

Liquid feeding saves the separate calcination step, which requires considerable engineering effort and which may be the source of a number of operational problems. Liquid feeding presents problems as well, such as capacity limitation due to the higher heat demand per unit weight of glass and a chance of unsteady boiling of the liquid fed onto the frozen but still very hot surface of the melt. However, liquid feeding seems feasible and, apparently, requires less sophisticated technology.

If calcination is to be performed solely to feed a melter, a fine powered calcine is desired, although not suitable as such for transport or interim storage. Granulated calcine can also be fed to a glass melter, but this will be considered a calcine consolidation treatment rather than a vitrification process.

To obtain a calcine powder, two techniques have been developed to the demonstration stage, the spray process and the rotary kiln process. Both have specific problems, such as the replacement frequency of the spray nozzle and the general reliability of a large rotating tube. Nevertheless, both have received intensive development and have proved to be feasible.

To obtain a granulated calcine a fluidized-bed process is available on a technical scale. A crucial point of this process is the treatment of dusty off-gas, which is created in large amounts by fluidizing the bed.

Denitration. Denitration may be performed thermally on calcination and/or melting. No separate denitration equipment is then required. The penalty is that the off-gas contains nitric oxides and that ruthenium volatility may be promoted by the oxidizing environment. However, there is still debate about the significance of the latter effect.

Denitration may also be achieved in solution by chemical means prior to feeding the calciner or the melter. The major part of nitric acid and nitrates will be destroyed, forming N2/N20, NO, and N02 depending on the reducing agent and the conditions of the chemical reaction. The nitrates are converted to oxides or oxide hydrates, forming a suspension that can easily be transferred to the calciner or melter. The most common denitrating agents to be applied are formic acid, formaldehyde, and sucrose.

Off-gas purification. As a high-temperature process, any type of vitrification process will have to have a very effective off-gas cleaning system. In fact, besides the remote operation and maintenance technique, off-gas treatment will be among the most important waste-processing problems to be solved. The off-gas may contain volatile fission products, such as ruthenium and cesium, as well as aerosols and dust. Multistage systems will be required with wet and dry cleaning procedures to obtain an off-gas sufficiently clean for release to the atmosphere.

Specific vitrification processes. An installation that has to serve a 5 MT/day reprocessing plant will have to have a capacity of about 150 liters/h, corresponding to a specific HLW volume of 600 liters/t of heavy metal and 80 percent load factor. As yet, none of the vitrification processes has been operated on this scale and with full radioactivity. In fact, there is only one process that is now demonstrated on a technical scale with highly radioactive waste and about 25 percent of full capacity, the French AVM process. The others are in different stages of development and still awaiting the hot demonstration phase. Consequently, design and operation data available are preliminary, and the following discussions of individual vitrification processes will not go deeply into the details but rather emphasize the principles of the processes.

U. S. vitrification processes. In the United States [М2] development efforts are focused on two processes, a spray-calcine/in-can melting process and a ceramic melter process that may be
coupled with a spray calciner, a fluidized-bed calciner, or liquid feeding (Figs. 11.11 and 11.12). No separate chemical denitration is provided in either case. Both processes are to produce borosilicate glass cylinders. Work is in progress at Battelle Pacific Northwest Labora­tories in Hanford.

In the spray calciner, liquid waste is pumped to a nozzle at the top of the calciner where it is atomized by pressurized air, producing droplets with diameters less than 70 дт that are dried and calcined in-flight in the 700°C-wall-temperature spray chamber. Sintered stainless steel dust filters collect a portion of the powder with a mean diameter of 10 дт. They are periodically cleaned by a reverse pulse of air. Calcine from the spray chambers and filters drops directly into the melting canister. Frit is fed to the cone of the calciner.

Two problems typical of such a device have been largely eliminated. Spray-chamber fouling has been overcome by improved feed atomization and by use of vibrators mounted radially on

ATOMIZING AIR

Figure 11.11 Spray calciner/in-can melter. (Courtesy of Battelle Pacific Northwest Laboratories.)

the spray chamber. The operating life of the nozzle has been significantly increased by using an alumina insert in the nozzle.

A spray calciner designed for a 5 MT/day reprocessing plant has an incoming waste flow rate of 118 liters/h with 75 g/liter solids to be combined with an off-gas system recycle of 8 liter/h to a total calciner feed of 126 liter/h with 70 g/liter solids. The product stream is 15.6 kg/h calcined solids. The demand on atomizing air is 0.85 m3 /min.

The canister is placed in a multizone furnace and coupled directly to the output of the calciner as shown in Fig. 11.11. The canister is heated to 1050°C and calcine and frit, which are metered continuously at a rate proportional to the calcine generation rate, are fed to the canister. Once the calcine-frit mixture in a zone has melted to a glass, heating is stopped and cooling is initiated to prevent excessive exposure of the canister to high temperatures. After the canister is full, feed is diverted to another canister in a parallel furnace.

The spray calciner is also used to feed a ceramic melter as shown in Fig. 11.12. In January 1975 an engineering-scale ceramic melter was started. The melting cavity is 0.36 m wide, 0.76 m long, and 0.30 m deep. The outside dimensions are 1.28 m wide, 1.36 m long, and 0.89 m high. This melter has a capacity of 60 kg/h glass when fed with calcine corresponding to a specific melting rate of 200 kg/m2 surface. The ceramic melter was inspected after about 11 months of continuous operation and only minor corrosion of the refractories and the electrodes was detected.

The same melter was used with liquid feed. This means that liquid HLW is transferred to a mix tank where the frit is slurried into the waste liquid. The slurry is then fed directly into the melting cavity and covers the molten glass. Flooding the entire surface with 40 to 80 mm of the slurry is preferred because particulate entrainment in the off-gas stream is less than with the slurry falling directly on the melt surface. With liquid feed the capacity of the melter is reduced very roughly by a factor of 5.

To cover the required capacity range for full-scale operation with liquid feed as well as

with calcined feed, a larger melter has been built. It has a melting cavity that is 0.86 m wide,

1.22 m long, and 0.71 m deep with a glass depth of 0.48 m. The overall size is 1.95 m wide,

2.13 m long, and 1.62 m high. This melter has a surface that is about four times larger than

that of the smaller one. Figure 11.13 shows this melter equipped for liquid feed.

For this new device, which has been operational since 1977, a new technique has been designed for periodic on-off drain control. The overflow is permanently open, but the melt flows only when the whole melter is tilted by a few degrees. When the canister is to be replaced, the melter is tilted back to a horizontal position, thereby interrupting the melt flow.

An alternative calcination process derived from the Idaho Waste Calcining Facility to be employed in connection with a vitrification unit has been developed to take advantage of the excellent heat transfer and solid mixing properties of fluidized beds. Silica is used as bed particles and is continuously fed into the bed at the rate needed in the final glass.

The German PAMELA process. In West Germany it has been decided to concentrate all development efforts on a modified PAMELA process, PAMELA II. The original PAMELA process, developed by Gelsenberg AG and Eurochemic [C3, Gl], is a liquid-feed/ceramic melter process with chemical denitration, producing phosphate glass. From the glass, beads of about 5 mm diameter are formed and embedded in a lead matrix. The product is called vitromet. This process has been developed up to a semitechnical scale, cold as well as hot. The advantages of this process are considered to be relative simplicity of phosphate glass preparation, suitability of a particulate product for quality-control procedures, and favorable mechanical and thermal properties of vitromet as discussed before.

PAMELA II, to be built by DWK (Deutsche Gesellschaft fiir Wiederaufbereitung von Kembrennstoffen) as a demonstration plant at the Eurochemic site near Mol (Belgium), differs from the original PAMELA with respect to the product. Phosphate glass will be replaced by borosilicate glass and the plant is to be operated on two product versions, glass blocks and vitromet. Construction of PAMELA II will profit a lot from two former process developments in West Germany: VERA, a spray-calcine/ceramic melter process carried out by Kem — forschungszentrum Karlsruhe, and FIPS, a drum dryer/in-can melting process carried out by Kemforschungsanlage Jiilich. The PAMELA II plant will vitrify the Eurochemic Purex waste (LEWC), with a specific activity of about 200 Сі/liter, and a specific heat rate of 0.7 W/liter.

According to the present design, the liquid waste is transferred from a process storage vessel by air lift or steam jet to the denitrator. The denitration is performed batchwise with one batch of 720 liter waste per day. Then formaldehyde solution (37 w/o) is metered into the waste. It will destroy the nitric acid and much of the nitrates. The effect of the denitration step has not yet proven in detail. A final decision as to whether the additional effort is justified is still pending.

The denitrated and concentrated waste is transferred to a mixing vessel where 140 g

borosilicate glass frit per liter of waste is added. The slurry is fed on top of the molten glass in a ceramic melter. The ceramic melter will have a surface area of 0.8 X 0.8 m. The depth of the glass melt is 0.4 m. The continuous feed rate is 30 liters/h. There will be two alternatives for the melt drain, one to fill a storage canister and another to produce glass beads.

For the glass block production, the melt is drained periodically from the melter by means of a bottom drain. This bottom drain uses joule heating as well as medium-frequency heating. It can be frozen with air cooling. The glass is cast into storage canisters. For the production of beads continuous draining is needed. Because of the low flow rates desired (2 liters melt/h), an overflow drain will be more suitable than a bottom drain. The glass melt leaves the drain as droplets.

Beads for vitromet production are prepared by means of a slowly (0.5 to 3 г/min) rotating stainless steel disk with a diameter of 700 mm. The droplets hit the disk and solidify to beads of about 5 mm diameter and 0.08 cm3 volume with a flat bottom. The beads are transferred to an intermediate storage vessel for product control and mass balance and then via a metering vessel to the final canister. The bead production is shown in Fig. 11.14. The canister is heated at 350 to 400°C and can be vibrated to achieve a close packing of the beads. When the canister is filled with beads, molten lead is introduced through a central pipe extending to the bottom. After some cooling the canister is sealed. Then it contains 67 v/o (volume percent) glass and 33 v/o lead.

The PAMELA II demonstration plant at Mol will have a capacity of about 30 liters HLW/h, corresponding to a scale-up factor of about 5 related to a 5 MT/day reprocessing plant. It is scheduled to be in operation in 1985.

The British HARVEST process. Another vitrification process is the HARVEST process [C3, М3], an improved version of the former FINGAL process. It is a pot process or, in the categories of this chapter, a liquid-feed/in-can melting process. A full-scale, fully radioactive plant is scheduled to be in operation at the Windscale site in 1986.

Concentrated radioactive waste solution, together with glass-forming chemicals such as silica and borax, are fed into a stainless steel vessel held at a temperature of 1050°C by a multizone, resistance-heated furnace. Evaporation, denitration, sintering, and glass formation occur steadily during the filling cycle and the feed rate is kept constant to ensure that the free liquid level rises at a rate equal to the rate at which glass is formed.

With the FINGAL process, the off-gases from the first, that is, the glass-making, vessel were passed through a second and a third vessel that contained filters to trap the particulate material and volatile ruthenium. At the end of the process cycle, when the first vessel was filled with glass, it was removed to storage and the vessel from the second position containing the primary filter was moved into the furnace and the filter was incorporated in the glass. A new vessel with a new filter was put in the middle position. The third vessel was only to provide a backup filter and did not require frequent replacement. Although this filter system was very efficient, it will not be used in the present HARVEST process. This is because of the filter size in a full-scale plant, because of problems that will arise when it fails blocking the entire off-gas system, and because of the necessity of handling additional pipe connections. The HARVEST off-gas system will rather follow the more conventional pattern of most other solidification processes.

The French AVM process. The French vitrification process at Marcoule is the first one in the world that is now effectively operating on a routine industrial basis after an exceptionally smooth period of test operation. With a team of 21 workers distributed among six shifts, AVM (Atelier de Vitrification de Marcoule) is operated continuously and produces one 150-liter glass block per day. It is used to solidify the backlog of military waste and future waste from natural uranium fuel produced at Marcoule.

AVM is a continuous rotary-kiln calciner/induction-heated melter process [B6, C3]. The

development of AVM was based on extensive experience with the pot process P1VER, which has been operated on a pilot-plant scale with full radioactivity for several years. Figure 11.15 shows the basic flow sheet of AVM.

The plant has two 15-m3 tanks for receipt of the liquid to be solidified. The liquid is cooled and agitated to avoid any buildup of solid residues. It is blended with additives before being fed to the calciner, to prevent caking.

The calciner, which receives a feed of 40 liters/h, comprises a tube of wrought Uranus-65 which has been machine-finished. The ends of the tube are fitted with graphite-ring air seals. These end fittings lie on “fore-and-aft” movable trolleys. The tube lies on easily removable roller bearings, has a slight slope, rotates at 30 г/min, and is heated by four • separate heaters arranged in zones.

The solution is fed through the upper end fitting and dried in the first half of the tube. The dry product, which is calcined in the second half of the tube at a temperature of 300 to 400° C, leaves by gravity through the lower end fitting and passes to the melting furnace, which is fed, through another connection, with small batches of glass frit.

The presence of a free rod inside the tube and the use of a chemical additive produces a more consistent calcined product and prevents material sticking to the inside surface of the tube. Great care was taken in the design of this component, in particular with respect to the quality of the output. It had been successfully tested in a 6000-h cold operation.

The melting furnace consists of a ceramic melting crucible heated to a temperature of around 1150°C by five induction heaters. The molten glass is allowed to build up in the furnace for a period of 8 h, and then a glass plug in the bottom of the furnace is melted through the use of two additional induction heaters and the glass is poured into the stainless steel canister. The canister is 50 cm in diameter and 1 m tall. It takes 3 h to fill with about 150 liters of glass.

In the original AVM design, an Inconel-crucible was used for glass melting, which requires more frequent replacement than a ceramic one.

The off-gas system ensures that the bulk of active material escaping from the furnace is trapped in a countercurrent water-scrubbing column and recirculated directly to the calciner. Fur­ther off-gas treatment includes a condenser, an absorption column, and a washing column. The low-activity liquid from this section of the plant is recycled to the adjoining reprocessing plant.

The main cell of the vitrification plant is provided with a 2-t bridge crane, eight shielded windows, and 14 manipulator positions. Every component in the plant is designed for remote disconnection and removal to an adjoining maintenance cell.

On-site engineered storage in air-cooled underground vaults is provided for the glass canisters. The storage facility has a 10-year capacity related to the AVM output.

The AVM plant is designed to produce glass blocks with a heat rate of up to 400 W/liter. The basic design of AVM is considered appropriate for the construction of a further plant to serve the La Hague reprocessing center. The intention is to build a vitrification plant of about twice the Marcoule capacity—probably with two parallel lines-to produce glass with a heat rate of up to 100 W/liter from the waste of oxide fuel reprocessing in the present plant (UP 2). This should be available to start glassmaking by about 1983. Follow-on vitrification plants of about the same size will be built for the two new oxide-reprocessing plants planned at La Hague, UP-ЗА and UP-3B. As the first of them is being assigned to the reprocessing of fuel from foreign customers and with contracts that provide for the return of waste in solidified form, the availability of proven technology for vitrification has assumed special importance.

Fluidized-bed calcination. The fluidized-bed calcination process has been developed at the Idaho National Engineering Laboratory (INEL), where in 1962 the Waste Calcining Facility (WCF)

started operation. Since then about 107 liters of liquid waste from the reprocessing of aluminum — and zirconium-alloy fuels have been calcined to produce about 1000 m3 of granular solids.

In the fluidized-bed calcination process, exemplified by the WCF, pneumatically atomized waste solution is sprayed at a gross rate of 375 liters/h into a 1.22-m-diameter by 1.37-m-deep fluidized bed of solidified waste granules maintained at 400 to 500°C. A recycle stream of off-gas scrubbing solution representing 20 to 30 percent of the total feed rate is added to the calciner feed stream. Inlet fluidizing velocities, based on only the fluidizing air flowing through the empty cross-sectional area of the calciner vessel, of 0.18 to 0.36 m/s are generally used, and a freeboard of about 2.3 m, supplemented by a louvered baffle, is provided for deentrainment of solids from the off-gas within the calciner vessel. The bed height is maintained at a constant level above the feed-spray nozzles by adjusting the rate of withdrawal of the product. Process heat is provided by in-bed combustion of kerosene with oxygen.

During operation waste is blended with required feed additives and fed by air lift and gravity to the calciner. The feed is atomized by air through spray nozzles located on the wall of the calciner vessel. The primary solidification mechanism is the evaporation of atomized liquid droplets on the fluidized-bed particles. A portion of the atomized liquid also evaporates to a dry powder before striking the surface of a bed particle. Therefore, the calciner produces a mixture of powdery solids and granules in the size range 0.05 to 0.5 mm.

Calcination of the waste solution to granular solids is accompanied by the release of large amounts of water vapor and gaseous products. These vapors and gases, along with the air employed for fluidizing the bed, atomizing the feed, and purging connecting lines, sweep a portion of the bed material-mainly, fine particles—into the off-gas piping. The initial separation of these solids from the gas takes place in a cyclone, the collected solids being combined with the primary product from the fluidized bed and transported pneumatically to product-storage bins in an underground vault. The gas then flows to a wet-scrubbing system that includes a quench tower, a venturi scrubber, and a deentrainment cyclone. In the scrubbing system, condensing takes place, which provides a scrubbing-solution recycle flow back to the calciner feed tanks at a rate sufficient to keep the dissolved solids content of the scrubbing solution well below the saturation level. For a final treatment, the off-gas is passed through four silica-gel beds in parallel and then through three off-gas filters in parallel. The silica-gel beds were installed primarily to remove gaseous ruthenium compounds, the only fission-product compounds in the wastes, other than tritium, that volatilize at the calcination temperatures.

Four solids-storage facilities have been placed in operation. The first and second facilities have been filled, and the third is presently being filled. The bins are cooled by atmospheric air, which flows through prefilters, down an inlet duct to the bottom of the vault. Air then flows upward through the vault by natural convection and out of the vault through a 15-m cooling-air stack. A forced-air cooling system was installed in the first storage facility, but has not been needed. The cooling air can be shut off, and high-efficiency filters can be installed, should radioactivity be detected.

In 1976 pilot-scale testing with simulated commercial high — and medium-level waste feedstock composition was conducted to demonstrate the feasibility of the process for this type of waste. Currently expected commercial waste compositions do not seem to present major problems in fluidized-bed calcination [М2]. A conceptual flow sheet for fluidized-bed solidifica­tion of commercial waste is shown in Fig. 11.16.

The specific volume of calcine will be about 40 liters/MT of heavy metal for combined HLW and MLW, corresponding to that to be expected from the AGNS plant. For final disposal, the product from the fluidized-bed calcination will have to be consolidated by melting with a glass flux. If it is to be stored for extended periods directly in sealed canisters, the calcined solid will have to be stabilized (denitrated, dehydrated) at approximately 900°C.

Fluidized-bed calcination is the only solidification process where long-term operation experience is available. Thereby it is probably the most readily available solidification process.

The solidification processes—vitrification and calcination—whose principles have been described in the last two sections are summarized in Table 11.11.

In many solvent extraction systems, addition of solutes to the aqueous phase increases the distribution coefficient of extractable components. Data in Table 4.2 and Fig. 4.6 show how addition of nitrates to an aqueous solution of uranyl nitrate increases the distribution coefficient of uranyl nitrate between the aqueous phase and diethyl ether [F2]. The increase in distribution coefficients with increased nitrate concentration is explained as follows: Analysis of

the ether phase shows that uranium is extracted in the form of un-ionized uranyl nitrate. Addition of nitrate ion tends to increase the concentration of un-ionized uranyl nitrate by shifting the equilibrium to the right, and thus

U022+ + 2N03- ^ U02(N03)2

converts more of the uranium to an extractable form. It may be noted that uranyl nitrate acts as a self-salting agent, probably also by displacement of this equilibrium. In addition, readily hydrated cations, such as Ca2+ and Al3+, tie up much of the water in the aqueous phase, and thus increase the effective concentration of uranyl nitrate.

When the organic complexing agent in the solvent is nearly all combined as extracted complexes, further increase in concentration of the complex-forming metal ions in the aqueous phase wih cause the distribution coefficient for metal extraction to decrease. This phenomenon has been observed for uranyl nitrate [G3, Ml, М2] and for zirconium and hafnium nitrates [H4] when extracted by TBP in kerosene.

Table 4.3 gives distribution coefficients for uranyl nitrate between aqueous nitric acid and 40 percent TBP in kerosene observed by Goldschmidt et al. [G3], At each nitric acid concentration, the uranium distribution coefficient decreases with increasing uranium concen­tration. This can be attributed to the following overall reaction equilibria [М2]:

U02 2*(aq) + 2N03~(«?) + 2TBP(o) ^ U02(N03)2-2TBP(o) and H>q) + N03‘(aq) + TBP(o) =* HN03 • TBP(o)

with the following equilibrium constants:

К = [UO:(NQ3)2-2TBP(0)]

U [U02 [N03 -(«?)]2 [TBP(o)]2 { J

Table 4.4 Distribution coefficients for Zr(N03)4 between water and TBP in kerosene

Aqueous phase: 3.0 M HNO3

3.5 M NaN03

Organic phase: 60 v/o TBP (2.19 M)

Moles Zr per liter

—————————— Distribution

Aqueous Organic coefficient

For the purposes of this chapter, activity coefficients of unity are assumed, so that the bracketed quantities in Eqs. (4.15) and (4.16) become identical with molar concentrations.

Assuming that at equilibrium all aqueous uranium is in the form of uranyl ion and all organic uranium is in the form of the U0j(N03)j ^TBP complex, the uranium distribution coefficient is

„ _ |UO,(NO,),-2TBP(o)] „ …

——— №’*(«,)]———— <4Л,)

and combining Eqs. (4.15) and (4.17),

Dv = Kv [NO* -(«7)J 2 [TBP(o)J 2 (4.18)

The TBP concentration appearing in this equation is that of uncombined TBP. For a given total amount of TBP in the organic phase, the uncombined TBP is lower the higher the concentration of uranium, and the uranium distribution coefficient should decrease as the uranium concentration increases. This is confirmed by the experimental data in Table 4.3.

Similar saturation effects are apparent from the data for the extraction of Zr(NOs)4 with TBP in kerosene [H4], as shown in Table 4.4.

The properties of UF6 summarized in this section have been taken primarily from a comprehensive report by DeWitt [D3].

Vapor pressure, triple point, and critical point. Table 5.10 gives the triple-point pressure and temperature of UF6 measured by Brickwedde et al. [B6], the critical pressure and temperature reported by Oliver et al. [01], and values of the vapor pressure at temperatures between —200°C and the critical point from the following sources:

1. Solid, below 0°C, Eq. (5.4) fitted by Llewellyn [LI] to his measurements between —15°C and the triple point:

log, o Ртогг = — 75.0 exp — 1.01 log10 T + 13.797 (T = r, °С + 273)

V (5.4)

2. Solid at 0°C, measured by Weinstock et al. [W2].

3. Solid, between 0°C and triple point, Eq. (5.5) fitted by Oliver et al. [01] to their measurements between 0°C and the triple point:

logro Pro* = 6.38353 + 0.0075377Г — , Гт^іб (5’5)

4. Liquid, between triple point and critical point, values quoted by DeWitt [D3] from a sixth-order polynomial fitted by Brooks and Wood [B7] to data of Oliver et al. [01]. These appear to be the most reliable and consistent data among the many cited by DeWitt. Values below — 15°C are of order-of-magnitude reliability only.

Density. Values of the density of liquid UF6 between the triple point and 160°C given in Table 5.10 are from measurements of Wechsler and Hoge [Wl]. The critical density is the estimate of Oliver et al. [01]. The solid density at 25°C is calculated from x-ray diffraction data. Solid UF6 formed when the liquid freezes or vapor condenses is much less dense because of contraction on solidification.

UF6 vapor is unassociated, but displays van der Waals-type departures from the ideal gas law. Weinstock et al. [W2] proposed Eq. (5.6) as an equation of state for UF6 vapor:

, , 3. _ 4.29 lp,, ,,

Pg (g/cm ) ^ _ 1(376(9оор/Гз) ( • )

where p is the pressure in atmospheres and T is in kelvins.

Thermodynamic properties. Table 5.11 gives thermodynamic properties of UF6 under its vapor pressure, selected from the comprehensive review of DeWitt [D3]. The original data sources were as follows:

Heat capacity and enthalpy of solid and liquid: measurements by Brickwedde et al. [B6].

Molal heat of vaporization: from the Gapeyron equation

dp dT

Change of vapor pressure with temperature: dp/dT from the measurements of Oliver et al. [01] summarized in Table 5.10.

Molal volume change AV: from

y_ 352 _ 352 Pg Pc

with pG from Weinstock’s Eq. (5.6) and Pc from Table 5.10.

Enthalpy of vapor: enthalpy of condensed phase plus heat of vaporization.

The value of 12,965 for the molal heat of vaporization at О К was found by Weinstock et al. [W2] to provide the best correlation of measurements of vapor pressure, density of each phase, and heat capacity of each phase through the Gapeyron equation.

Table 5.12 gives thermodynamic properties of UF6 in the ideal gas state at 1 atm pressure, denoted by the superscript (°). Values of C£, the free-energy function —(G° —Я£)/Г and the entropy S° are taken from DeWitt’s [D3] citation of calculations by Bigeleisen et al. [B3] from spectroscopic data. The enthalpy of UF6 in the ideal gas state relative to the solid at 298.15 К was evaluated from

Barm and Knacke [B1 ] give for the heat of formation of solid UF6 at 25°C

AHCi298 = -523,000 cal/g-mol (5.12)

Hence the heat of formation of UF6 in the ideal gas state at 25°C is

298 = -523,000 + 11,807 =-511,193 (5.13)

The free energy of formation of solid UF6 at 25° C from the elements, from the same source, is

ДССі298 = —492,252 cal/g-mol (5.14)

The heat capacities of the two solid phases of thorium and the liquid metal and the heats of transformation and fission are given in Table 6.6.

1.5 Thermal and Electrical Conductivity

The thermal conductivity of thorium metal is given in Table 6.7. The electrical conductivity of thorium metal is very dependent on its impurity content. Chiotti [C3] found that at room temperature the resistivity of thorium metal containing 0.2 w/o (weight percent) carbon was 37 X 10~* fi-cm, and that of metal containing 0.03 w/o carbon was 18 X 10’6 f2-cm. An extrapolated value for carbon-free thorium metal is 13 to 15 X 10’6 fl’cm. The temperature coefficient of resistivity is 3.6 to 4.0 X 10-3 per °С.

The alkali-fusion process was developed by the Ames Laboratory of the U. S. Atomic Energy Commission [B2] to provide a method for producing zirconium salts that did not need the high temperature of an electric furnace. A flow sheet for this process is shown in Fig. 7.5. In this process, zircon sand is mixed with from 1.0 to 1.5 times its weight of sodium hydroxide, and the mixture is heated in a furnace at 565°C. The sodium hydroxide melts at 318°C, and as its temperature rises it reacts with the zircon:

4NaOH + ZrSi04 -*■ Na2Zr03 + Na2Si03 + 2H20

Figure 7.4 Production of Z1CI4 from zircon.

Steam is evolved, the mix becomes viscous, and finally is converted to a fragile, porous solid (“frit”) when the temperature reaches 530°C. After cooling, this solid is ground and leached with water, which extracts the Na2Si03. The residue then is leached with acid, which dissolves the Na2Zr03. The final residue consists of unreacted zircon, which may be recycled. Any desired zirconyl salt can be made by using the appropriate acid in the final leaching step.

Ibis process would appear to be especially suitable for preparing feed for separating hafnium from zirconium by solvent extraction from an aqueous solution.

1.9 Fluosilicate Fusion

Fhiosilicate fusion has been used in the Soviet Union [SI] to produce feed for separation of hafnium from zirconium by fractional crystallization of K5MF6. Zircon is ground to pass 200 mesh and mixed with potassium fluosilicate and potassium chloride (to act as promoter). The mixture is sintered in a rotary furnace at 650 to 700°C. The following reaction takes place:

ZrSi04 + K2SiF6 -*■ KjZrF6 + 2Si03

The sinter is cooled, crushed to pass 100 mesh, and leached at 85°C with 1 percent HC1. The product is filtered at 80°C, then cooled, to crystallize K2ZrF6(+K2HfF6), which are filtered off and washed with water.