Thorium is important in nuclear technology as the naturally occurring fertile nuclide from which neutron capture produces fissile 233 U by the succession of reactions

-Th + 0" —moT

In thermal-neutron reactors 233 U has an important advantage over 235 U or 239 Pu in that the number of neutrons produced per thermal neutron absorbed, rj, is higher for 233 U than for the other fissile nuclides. Table 6.1 compares the 2200 m/s cross sections and neutron yields in fission of these three nuclides. Thorium has not heretofore been extensively used in nuclear reactors because of the ready availability of the 235 U in natural or slightly enriched uranium. As natural uranium becomes scarcer and the conservation of neutrons and fissile material becomes more important, it is anticipated that production of 233 U from thorium will become of greater significance.

Compared with 239Pu, the other synthetic fissile nuclide, 233U, has the advantage that it can be “denatured,” made less available for use as a nuclear explosive, by isotopic dilution with 238 U in a mixture containing less than 12 percent 233 U. Production of a nuclear explosive from such a mixture would require costly and difficult isotope separation (Chap. 14). No similar means exists for denaturing 239Pu, which can be more readily separated from 238U by chemical reprocessing (Chap. 10).

In addition to its potential use in nuclear power systems, thorium has had minor industrial use in Welsbach mantles for incandescent gas lamps, in magnesium alloys to increase strength and creep resistance at high temperatures, and in refractories.

The principal halides of zirconium are listed in Table 7.6, together with the melting points, vapor-pressure equations, and temperatures at which the vapor pressure of each equals 760 Torr. The tetrahalides sublime without melting at atmospheric pressure, like UF6. The lower halides disproportionate before melting.

The halides of greatest practical importance are ZrF4, as a component of fluozirconates; ZrCL), as the feed material for production of zirconium in the Kroll process; and Zrl4, as feed material in the hot-wire process. These will be discussed later in the chapter.

The heat capacity, enthalpy, heat of formation, and free energy of formation of ZrCU are given in Table 7.7.

The tetrahalides are hydrolyzed to Zr02 by steam. They dissolve in water, forming oxyhalides such as ZrOCl2.

3.3 Light-Element (a, n) Reactions

Additional biological hazard in the handling of plutonium recovered from irradiated uranium or of uranium from irradiated thorium arises from fast neutrons produced by (a, n) reaction. Alpha particles from actinide decay react with light elements—lithium, beryllium, carbon, oxygen, etc.—to produce energetic neutrons such as

2Be + ?He-> *?C + in (8.69)

The fast neutrons are very penetrating and may require some hydrogenous shielding for protection of operating personnel. Also, techniques to ensure low concentration of light-element contaminants in the recycled actinide material may be required.

The allowable concentration of light elements in recycled fuel depends on the alpha-decay rate in the material, the energy of the alpha particle, the probability of an (a, n) reaction, the energy and relative biological effectiveness of the neutron produced, and the allowable surface dose rate of these (a, n) neutrons. The average energies of neutrons from (a, n) reactions in light elements are listed in Table 8.13 along with the tolerance flux for these neutrons. Also listed in Table 8.13 is the neutron emission rate per gram of uranium or plutonium metal that would result in a dose of 1 rem per 40-h exposure at the surface of a kilogram of this metal. This dose rate is about 30 percent less than the official tolerance for radiation exposure localized to the hands and forearms of radiation workers.

The rate of neutron generation from (a, n) reactions in a fuel containing alpha-emitting actinides and various light elements is predicted from

where n = the neutron production rate

Xi = the concentration of the light element;

+Based on data by Arnold [Al, A2].

Ф Reaction constant, a.

§ Based on 1010 alphas directly from 228 Th, but includes effect of 228 Th daughters.

a, y = the number of (a, n) neutrons per alpha disintegration per unit concentration of the light element, with / identifying the energy of the alpha particle Aj = the alpha-disintegration rate of actinide /

Values of a for various light elements, calculated from the data of Arnold [A2], are listed in Table

8.14. The values of a for 239Pu and 240Pu are assumed equal for a given light element because of the nearly equal energies of the alphas from these two isotopes. Similarly, the a for 242 Pu should be very nearly the same as that for 233 U.

For a given mass and isotopic composition of plutonium and contaminant concentration, neutron production rate and allowable concentration of a given contaminant can be estimated from data in Tables 8.13 and 8.14. Estimates for 239Pu containing 1000 ppm 238Pu and for 233U containing 100 ppm 232U are given in Table 8.15. By comparison, plutonium undergoing

Table 8.15 (a, n) surface dose from light elements in recycled plutonium and uranium

Light-element concentration (ppm)l to give
1 rem/40-h wk at surface of 1-kg sphere

final purification by the typical technique of oxalate precipitation may contain the contaminant concentrations listed in Table 8.16. Boron, sodium, and calcium are easily found above the allowable concentration. Either neutron shielding or special precautions to maintain low contaminant concentrations are necessary.

A special problem arises when chemical compounds of plutonium with light elements are handled as massive solids. For example, Pu02 fuel will produce above-tolerance fluxes of (a, n) neutrons at surface contact. Even greater neutron production occurs with PuF4, which is an intermediate in the conversion of plutonium compounds to plutonium metal. Approximately 12 neutrons are produced per 106 alphas in 239PuF4, and some thickness of neutron-shield material may be required.

Both the Redox and Trigly process had the disadvantage of adding large amounts of nitrate salts to the high-level wastes. The first process free from this disadvantage to be developed was the Butex process, which used dibutyl carbitol (C4 H9 OC2 H4 OC2 H4 OC4 H9) as solvent and nitric acid as salting agent. The nitric acid was evaporated from the aqueous high-level wastes and reused. The Butex process was developed by a group of British chemists [N4] working at the Chalk River Laboratory in the late 1940s. As possible solvents they evaluated diethyl ether, hexone, triglycol dichloride, dibutyl cellosolve (C4H9OC2 H40C4H9), and dibutyl carbitol. They concluded that hexone and dibutyl carbitol were satisfactory solvents, but that dibutyl carbitol was preferable because of its greater stability toward nitric acid and lower vapor pressure.

Following successful pilot-plant tests at Chalk River [N4], the Butex process was adopted for large-scale separation of plutonium, uranium, and fission products from natural uranium irradiated to low bumup at the Windscale plant of the U. K. Atomic Energy Authority [Н8]. Even after its use in this application was replaced by the Purex process, the Butex process remained in use at Windscale for primary decontamination of high-bumup fuel until the 1970s. Then an explosion, probably due to reaction of nitric acid with solvent, terminated its use.

5.2 Differences from LWR Fuels

Table 10.20 lists the principal differences between irradiated fuel to be reprocessed from an LMFBR and an LWR. The data have been excerpted from Figs. 3.34 and 3.31 and Tables 8.8 and 8.7.

Because some of the sodium coolant used in the LMFBR fuel that may have adhered to the cladding or penetrated leaks in it would react vigorously with water or nitric acid, it is necessary to oxidize all sodium by exposing the fuel to an inert gas containing a controlled amount of water vapor before the dissolution step. LMFBR fuel may not be stored with water cooling till after all sodium has been removed.

Use of stainless steel cladding in the LMFBR instead of zircaloy has little effect on mechanical decladding or on dissolution. Stainless steel, like zircaloy, is not rapidly dissolved by nitric acid of the concentrations used in the Purex process.

Fuel in the core of the LMFBR is operated at a specific power over three times that of the LWR. During the cooling period, the specific power of LMFBR core fuel from radioactive decay remains about three times that of LWR fuel cooled for the same length of time. This

makes shipping, handling, and storage of LMFBR fuel prior to reprocessing much more difficult than LWR fuel.

To reduce the specific power somewhat in reprocessing, it is planned to combine irradiated fuel from the LMFBR core with irradiated fuel from the LMFBR blankets in proportion to the rates at which they are discharged from the reactor. Even so, the specific power of LMFBR fuel cooled 150 days is 1.4 times that of LWR fuel cooled the same length of time.

The bumup of fuel in the LMFBR core is two or more times that of LWR fuel, leading to higher concentrations of fission products, gaseous and solid, and greater radiation effects on cladding and fuel. The average bumup of combined LMFBR core and blanket material is about 15 percent higher than that of LWR fuel.

The concentration of plutonium in combined core and blanket fuel from the LMFBR is more than 10 times that of LWR fuel. This is the most significant difference between the two fuels with respect to reprocessing. Other important differences are the greater amounts of tritium and 1311, the 140 percent greater ruthenium activity, and the 60 percent greater overall specific activity of 150-day cooled LMFBR fuel.

Because of the high plutonium content of spent fuel from the LMFBR, there is strong

&

Figure 10.27 Aqueous phase concentration

at which second organic phase forms. ——-

n-dodecane [W6];——— Ultrasene [S23].

economic incentive to return this plutonium to the reactor with minimum delay for cooling, reprocessing, and refabrication. Consequently, the foregoing comparison of LMFBR and LWR reprocessing conditions for equal cooling periods of 150 days does not tell the whole story. For example, if LMFBR fuel were reprocessed 90 days after discharge from the reactor instead of 150 days, the activity of 8.05-day 1311 would be

2(iso-90)/8.os _ ]75 (10.16)

times greater, and the specific power of fuel from the core would be 1.5 times greater.

The following discussion of reprocessing LMFBR fuels outlines the principal process steps, lists the main problem areas, and discusses possible solutions. Since 1973, international dissemination of reprocessing information has been restricted. This discussion of reprocessing LMFBR fuel is thus less complete and less up to date than would be desired.

Spent fuel elements contain appreciable amounts of tritium, partly produced by fission, partly by other nuclear reactions. About half of the tritium is released from the fuel upon dissolution. The rest is bound to the zircaloy of the hulls and is disposed of with them. The fraction of tritium that is released exchanges with water, forming НТО. The total annual input of tritium in a 1400 MT/year reprocessing plant is about 106 Ci. In West Germany a reprocessing plant of this size is supposed to retain 75 to 80 percent.

The fundamental problem of tritium waste management is that there is no simple way to reduce the volume of tritiated water. There are techniques available to minimize the volume generated in reprocessing, e. g., reuse of tritiated water to feed steam jets, and strict confinement of tritium in the first extraction cycle. These techniques, however, add complica­tions to the process. If, therefore, an inexpensive way were available to dispose of untreated tritiated water, severe generation restrictions would not be appropriate. If, however, expensive methods were to be applied, such as solidification or even concentration by isotopic enrichment, the volume generated has to be limited as much as possible.

Another approach is a suitable head-end process in the reprocessing plant, such as voloxidation (Chap. 10, Sec. 4.3). However, such a head-end process is not yet available technology but requires several more years of development.

There are minor quantities of tritium smeared out over the whole reprocessing flow scheme that will ultimately arise as low-activity condensate with tritium concentrations of the order of 10"4 Сі/liter and 1СГ8 Ci./liter of other radionuclides. It is very likely that this can be released to surface waters.

Basically three options are considered to dispose of tritiated water that is stored in tanks and cannot be released.

Deep well disposal. Injection of tritium-containing liquid into isolated aquifers or depleted oil horizons is the most interesting option. This technique has been used increasingly for almost 20 years to dispose of industrial wastes. In the United States, for instance, some hundred injection wells have been drilled and are actually in operation at depths between 60 m and 3600 m. Although there are still licensing problems, this is a safe and economic way to dispose of tritiated water.

This technique will be tested for tritiated water in the neighborhood of the Karlsruhe Nuclear Research Center in West Germany. An isolated oil lens that is exhausted but located in an oil field still being exploited will be used. Thereby any migrations occurring deep underground will be detected.

Solidification. In principle, any solid that contains firmly bound water may be suitable as a solidification form for НТО. This includes drying agents, such as silica gel, molecular sieves, and calcium sulfate, as well as hydraulic cement and organic polymers. Most experience is available with cement, which has been used to solidify non-high-level waste for quite a while.

Although concrete is a monolithic solid, it is quite porous. In contact with water, about a third of the tritium will be released, mainly by isotopic exchange, in the first month. The release may be retarded by coating the cement. Because of the relatively high leachability, cemented НТО would have to be stored in gastight steel cylinders, probably in a nonaccessible geologic repository.

If it turns out that a more leach-resistant and probably more expensive solidification product has to be developed, it may well become beneficial not only to restrict the volume arising from reprocessing but also to further reduce it by isotopic enrichment prior to solidification. An enrichment process suitable for this purpose must provide a very effectively depleted waste stream.

Ocean disposal. In view of the relatively short half-life of tritium and of the enormous isotopic dilution, sea disposal is another alternative for dealing with tritium waste. Transport will be an economic drawback of this alternative, and political and administrative problems will have to be solved.

3.1 1291

All iodine isotopes except 1291 will have decayed prior to reprocessing as long as a large backlog of unreprocessed spent fuel exists. The 1291 activity per metric ton of heavy metal (30,000 MWd/MT) is only 34 mCi. However, its extremely long half-life of 17 million years makes 1291 a permanent contaminant if released to the atmosphere. In shorter-cooled fuel elements radioactive 1311 will also be present and must be recovered.

Practically all iodine from spent fuel will be released upon dissolution with the dissolver off-gas. There are several scrubbing techniques that remove iodine effectively from the off-gas but do not yield a stable product for long-term disposal.

For permanent fixation of 1291 adsorption on silver-loaded adsorbents, such as zeolites, silica, or alumina, will be the choice [PI, W2]. The process is simple, the bed temperature may be relatively high, the product is a dry solid, the chemisorbed iodine is highly insoluble, and the adsorbent is very efficient in removing both organic and inorganic iodine from gas streams.

The 1291 content of spent fuel with an average bumup of 30,000 MWd/MT heavy metal is 211 g/MT corresponding to 34 mCi. This corresponds to an annual production from a 1400 MT/year reprocessing plant of 300 kg 129I. As there will be some isotopic dilution, an iodine-recovery system could conceivably be required to remove 600 kg of iodine annually. If iodine will be adsorbed on silver zeolite beds ready for final disposal, the total amount of iodine waste is then estimated to be about 5 m3/year with a total activity of 50 Ci. The amount of silver corresponding to 600 kg iodine is about 500 kg. Even though there will be excess silver required, this does not seem an unreasonable silv r consumption in view of the overall reprocessing costs. The world’s silver production was almost 104 tons in 1976. There is, however, some research in progress on regeneration of iodine-loaded silver zeolite and reloading the iodine on a lead zeolite.

When the distribution coefficients are independent of stage number, an equation can be derived for analytical calculation of the number of stages.

For any extractable component with a constant distribution coefficient, Eqs. (4.27) and (4.28) can be rewritten in terms of the constant extraction factor (3:

Equation (4.45) is a form of the Kremser equation, originally derived for countercurrent gas absorption [S3].

The raffinate concentration х, may be eliminated by an overall material balance,

ЩхР ~ Xi) = E(yN — y0) which combines with (4.45) to yield

■>7v -Уо = 0N ~ 1

DxF — y0 0N+1 — 1

Thus, by specifying /3 for the cascade and the ratio Уо/DxF and recovery p for any one of the extractable components, the required number of equilibrium stages N can be calculated from (4.48).

If we have extractable components A and В in the feed to be separated in a simple extraction cascade, the constant distribution coefficients DA and DB result in extraction factors PA and 0B, and the overall decontamination factor fAB is obtained by applying Eq. (4.48) to each of the components, with

For the special case of y0 = 0, Eqs. (4.48) and (4.49) combine to yield

(4.50)

For a specified number of stages N and specified E/F, the decontamination factor for any two extractable components can be calculated from (4.48) and (4.49) or, for y0 = 0, from (4.50).

To illustrate the use of these equations, consider the extraction of zirconium from an aqueous solution of zirconium and hafnium nitrates, as shown in Fig. 4.12. Although the distribution coefficients do, in fact, depend on the concentration of zirconium and hafnium, as shown later in Sec. 6.6, constant distribution coefficients are assumed here for the purpose of this illustration. The specified feed composition and the specified recovery to be obtained are listed in Table 4.6.

The distribution coefficients assumed above are those observed by Нигё and Saint James

Figure 4.11 Number of equilibrium stages in an extraction cascade. (Adapted from Sherwood et al [S3].)

[H4] for an aqueous solution of the feed of concentration 3.5 ЛЧп NaN03 and 3.0 jV in HNO3, in contact with 60 percent TBP in kerosene. Distribution coefficients will be higher at the bottom of the cascade, where the aqueous zirconium concentration is lower; this will be neglected in the present treatment, but will be taken into account in Sec. 6.6.

By applying Eqs. (4.37) and (4,48) to zirconium,

hr= 1.2

from which N = 12.2.

The zirconium-hafnium decontamination factor is obtained from (4.48) and (4.49) with 0Hf“O.12:

The McCabe-Thiele diagrams for this example are shown in Fig. 4.13.

It is interesting to compare the decontamination obtainable for pZl = 0.98 and E/F = 1.0 with that obtainable with an infinite number of stages, corresponding to operation at the same zirconium recovery but at (£/F)mm — From Eq. (4.35):

= 0.817

and with N = », Eq. (4.50) yields

A more effective way to use an increased number of stages in a simple extracting cascade would be to increase the zirconium recovery. This would occur by allowing the slope of the operating line to approach DZr. In the limit of N-*”0, xZr, i “*■ 0 and p zr_>l — Because Dnt<Dzt, the operating line for hafnium can intersect the hafnium equilibrium line only at x > xfif and not at *нг = 0. In this limit of

Source: Adapted from J. Hure and R. Saint James, “Process for Separation of Zirconium and Hafnium,” Proceedings of the International Conference on the Peaceful Uses of Atomic Energy, vol. 8, United Nations, New York, 1956, p. 551.

we find from (4.50) that

To obtain better decontamination a scrubbing section is added to the cascade, as illustrated in Fig. 4.4 and analyzed in the following section.

1.14 World Resources

Table 5.17 summarizes information on the uranium resources and the annual uranium production capabilities of the principal uranium-producing countries of the non-Communist world compiled by the Organization for Economic Cooperation and Development [02] in December 1977. The production costs of $30 and $50/lb U308 do not include exploration costs. The selling price of U308 is highly variable, as it depends on when the sale was negotiated, when the uranium is to be delivered, and whether it is a price for one-time, spot delivery or long-term supply. Typical prices in 1978 were in the range $30 to 50/lb U308. In order of decreasing resources, the four countries with the largest resources are the United States, Canada, South Africa, and Australia.

Reference [02] states:

Reasonably Assured Resources refers to uranium which occurs in known ore deposits of such grade, quantity and configuration that it could be recovered within the given production cost range with currently proven. . . technology. Estimates of tonnage and grade are based on specific sample data and measurements of the deposits… .

Estimated Additional Resources refers to uranium surmised to occur in unexplored extensions of known deposits in known uranium districts, and which is expected to be discoverable and could be produced in the given cost range.

^MT (metric ton) = 1 megagram = 1.102 short tons.

Thorium compounds can be extracted from aqueous solution by many of the immiscible organic solvents that have been used for extraction of uranium (Table 5.14). As with uranium, tributyl phosphate (TBP) is now the universal choice for extracting thorium from aqueous nitric acid solutions and for purifying thorium compounds by solvent extraction. However, in sulfuric acid solutions of thorium compounds or in solutions containing phosphoric acid such as are obtained from acid leaching of monazite, thorium is too strongly complexed to be readily extracted by TBP. To extract thorium from such solutions, processes using other organo — phosphorus compounds or organic amines have been developed, just as they have for uranium. Audsley and co-workers [Al] conducted pilot-plant experiments on extraction of thorium from solutions simulating the composition of sulfuric acid leach liquors from Canadian uranium ores after removal of uranium. They found that di(2-ethylhexyl)phosphoric acid, the solvent used in the Dapex process (Chap. 5, Sec. 8.6), was a satisfactory extractant, provided that ferric iron in the leach liquor (which would extract with thorium) was first reduced to ferrous by reaction with iron filings.

Because of the complications introduced by ferric iron, Audsley et al. concluded that a solvent that would be selective for thorium in the presence of ferric iron and that would not be inhibited by phosphate would be preferable to the Dapex solvent. They concurred with the conclusion of Crouse and co-workers at Oak Ridge National Laboratory [C5], that long-chain primary amines selectively extract thorium in the presence of uranyl, ferric, and phosphate ions. Compounds of this type are now the preferred extractant for thorium in such systems. Application of this so-called Amex process to thorium extraction from monazite is described in Sec. 8.6.

In the United States, practically all zirconium metal is now being made by the Kroll process. This process was an adaptation to zirconium of a similar process for titanium developed by W. J. Kroll. The work of Kroll and metallurgists of the Albany, Oregon, station of the Bureau of Mines culminated in a plant to produce 135,000 kg zirconium/year at the station. A similar plant was operated by the Carborundum Metals Corporation, at Akron, New York. These have been superseded by the plant of the Teledyne Wah Chang Albany Company, at Albany, Oregon, with a capacity in 1978 of 3.4 million kg/year.

The form of the Kroll process used in this plant is believed to be generally similar to the process used at the Bureau of Mines plant, which has been described in detail by Shelton et al. [S3]. The principal steps in the Kroll process as practiced in the Bureau of Mines plant are diown in Fig. 7.11.

Figure 7.11 Kroll zirconium process as practiced at Albany, Oregon.

Production of ZrCl4. Zirconium oxide from the hafnium-separation step was mixed with carbon black, dextrin, and water in proportions 142 Zr02, 142 C, 8 dextrin, and 8 water. The mixture was pressed into small briquettes (3.8 X 2.5 X 1.9 cm) and dried at 120°C in a tray drier. The oxide briquettes were charged to the reaction zone of a vertical-shaft chlorinator lined with silica brick. The charge was first heated by carbon resistance strips until it became conductive. During production, the bed temperature was maintained at 600 to 800°C by an electric current passed directly through the bed. After steady conditions were reached, a reactor 66 cm in diameter produced about 25 kg ZrCLt/h. The ZrCl4 was condensed from the reaction products in two cyclone-shaped aftercondensers in series, and the chlorine off-gas was removed in a water scrubbing tower.

The Wah Chang plant is believed to use an electrically heated fluidized-bed chlorination reactor.

Reduction of ZrCl4. The furnace used for reducing ZrCU to zirconium metal is shown in Fig. 7.12. The outer shell is a stainless steel cylinder 178 cm high and 70 cm inside diameter. An annu­lar trough 5 cm wide and 23 cm deep is welded to the top lip of the cylinder. The top lid of the furnace carries a cylindrical ring that dips into this trough. The trough is filled with a low-melting lead alloy. This arrangement facilitates opening and closing the furnace.

The top lid also carries stainless steel cooling coils, through which air may be circulated, to control top temperatures and prevent loss of ZrCl4 vapor in gases discharged from the furnace. The furnace is provided with three external electric resistance heating elements to provide the heat of sublimation of ZrCU and control temperature distribution.

A stainless steel( reduction crucible rests on the bottom of the furnace. Before a run is started, this is charged with 55 kg of distilled magnesium.

Resting on the top of the crucible is an Inconel can charged with raw ZrCU. The total amount of ZrCU charged here and possibly present on the air-cooling coils from a previous run is 236 kg, an amount that provides a magnesium excess of 10 or 15 percent for the reduction reaction

.Evocuation Tube
в / ^Bleeding Valve

Top Plate Heating Element

eod Seal

eod Seal Heating Element

.Middle — Zone Heating Element Raw Chloride Spocers

etort

Reduction

Baffle

Crucible Lifting Bor

Reduction

Crucible

Lower — Zone Heating Element

Cooling Plugs

ZrCl4 + 2Mg -»• Zr + 2MgCl2

Inconel rods extending into the ZrCL» improve heat transfer.

The detailed operating cycle has been described by Shelton et al. [S3]. A brief summary of the procedure follows. With the reactor first at 300°C and cooling air flowing through the top coils, the furnace is evacuated and flushed with helium three times to remove gases originally present. Temperatures are then raised to 450°C while the internal pressure is kept near atmospheric by bleeding helium. This purges additional gas occluded in the ZrCI* charge. Top temperatures are then reduced to lower the ZrCl4 vapor pressure, while additional helium is fed to hold pressure near atmospheric. The temperature of the reduction crucible containing the magnesium is next raised to 825°C, at which reaction with ZrCLt vapor commences. The ZrCL* transport rate is controlled by the rate at which heat is supplied to the middle-zone heater. The rate is kept as high as possible without raising the temperature in the magnesium reaction zone over 875°C. Completion of reaction is indicated by a fall in pressure, which is countered by supplying additional helium. Heaters are then turned off and the vessel is cooled to 150°C by air blown over the outer reactor surface.

The total cycle time is around 40 h, of which 18 h is for the reduction reaction itself. The product of the reaction is a lower layer of spongy zirconium metal mixed with MgCl2, covered by a layer of frozen MgCl2.

Vacuum distillation of MgCl2. To remove MgCl2 from the zirconium sponge it is necessary to resort to vacuum distillation. Water leaching cannot be used because the finely divided zirconium sponge would become contaminated by oxide corrosion product.

The crucible containing the zirconium sponge and MgClj is transferred to the vacuum distillation retort, shown in Fig. 7.13, where it is supported, upside down, over a perforated, stainless steel funnel. The air in the retort is evacuated, and the crucible is heated to 900 to 920°C to melt the MgCl2, which partially drains off the sponge. Salt still wetting the sponge is distilled at this temperature to a water-cooled jacket inside the retort.

Arc melting. As a last step, the salt-free sponge is fed into an arc-melting furnace and cast into the desired shapes in a water-cooled, copper mold. The furnace atmosphere is helium.

Details of the Rroll process are given in papers by Kroll and his co-workers [КЗ, K4], Shelton [S3], and Lustman and Kerze [LI].

Reduction with sodium. A modified process in which ZrCL was reduced by sodium was used by National Distillers and Chemicals Corporation in Ashtabula, Ohio, during the late 1950s [С1].

Production of hafnium metal. Hafnium metal has been produced from HfOClj by the same methods used in making zirconium. Up to the vacuum distillation step, separate equipment was used than for zirconium, to avoid contaminating the zirconium.