ETAL A Staff-Industry Collaborative Report ELSPETH W. MAINLAND, Assistant Editor in collaboration with D. A. ORTH, E. L. FIELD, and J. H. RADKE Savannah River Plant, E. 1. du Pont de Nemours and
Co., Inc., Aiken, S. C.
PLUTONIUM
At a complex of plants and laboratories on the Savannah Rives near Aiken, S. C., Du Pont produces various materials for the U. S. Atomic Energy Cornmission; one of these is
plutonium
metal
is one of the ten manmade elements heavier than uranium, and as such is a relative newcomer. I n 1940, Seaborg, McMillan, Wahl, and Kennedy at the University of California first produced one isotope, g4Pu23*. The fissile isotope, 94Pu239, was first formed at the same location in 1941. I t was produced in a cyclotron, by neutron bombardment of uranium following the reaction : n-
9zu*38
(%Y)
P szu239 -I
23 min.
82.3 days
239
9 s N p 2 3 9 -8&
This is the isotope of interest for both bombs and reactor fuel. Today, the chain reaction in a nuclear reactor (“pile”) is the only source of neutrons large enough to make practical amounts of plutonium. Spent fuel elements are therefore “plutonium ore”-a mixture of plutonium, unreacted uranium, and products of the fission of uranium-235 that produced the neutrons. The finished product looks much like other metals, with the common silvery luster. But plutonium cannot
exist in large pieces. Violent nuclear chain reaction would promptly disperse them. I n fact, batch sizes in this plant must be limited to small “buttons” that a man can hold in his hand. Plutonium is a very dense metal, as heavy as gold. An impressive property is that a relatively large piece of the metal is warm to the touch-because of the energy given off in alpha-decay. As a radioactive material, plutonium is not particularly hard to handle. Plutonium-239 has a half-life of about 24,000 years, and decays by emitting an alpha-particle. These particles have very low penetrating power, so that shielding is not required after the fission products have been removed. But plutonium is deadly when absorbed by the body. Safety considerations are therefore : 0 0
Prevention of contamination Limiting the amount of material to below critical mass
These considerations lead to some odd contrasts in size and processing methods at Savannah River. T h e plant site occupies over 300 square miles studded with widely separated processing areas. VOL. 53, NO. 9
SEPTEMBER 1961
685
FLOW DIAGRAMS I Eluate is divided into fractions depend-
Ion Exchange Process
The columns are operated on a batch basis, with counter-flow absorption and elution.
ing on plutonium content. A rich portion is taken for the product fraction, while fore and tailing fractions are recycled 0 Nonroutine column operations include
extended product elution for heel removal, refrigerated wash for degassing or reduction of Pu(lV), and fission product elution if the column accumulates too much activity
Absorption
Elution and Reconditioning
0
0 The feed (solvent extraction plant product solution) flows down through the column until a desired load of plutonium has been absorbed
Feed Plutonium, g./l. "03, M NHqOH.'/eHoSOd. M A bsofp t io n ' Batch, g . Pu 7 in. diameter 10 in. diameter Column heel, % of batch Flow rate, ml./min./sq. cm. Losses (absorption and decontamination wash), %
A wash of dilute sulfuric acid down the column removes absorbed uranium and fission products
''
0
Eluting solution flows up through the column and the bulk of the plutonium solution is removed, leaving a heel 0
0 An upflow reconditioning wash restores the low acid condition, making the column ready for the next cycle
a
0.2-1.2 0.2-0.6 0.025-0.05 350-7 750-850 5-15 10-20 10-4-10-1
Fluorination and Reduction Processes to hydrogen fluoride in a furnace at temperatures to 600" C,
Excess hydrogen fluoride, diluted with air, is destroyed with limestone 0
0 The plutonium tetrafluoride is mixed with metallic calcium, and packed in a magnesium oxide crucible
Iodine may be added to act as a "booster" providing additional heat of reaction. This is necessary for small batches 0
The crucible is heated in a pressure chamber by an induction furnace until the 0
686
INDUSTRIAL AND ENGINEERING CHEMISTRY
Elutrian t Reconditioning wash
1.6-4.0 2
Flow rates, ml./min./sq. cm.
E Iu t r ia n t Reconditioning wash
0.2-0.5 0.2-0.8
Product fraction
Pu concentration, g./L
50-70
A m o u n t loaded in addition to the heel.
Key to Operation The desired hexagonal crystal phase can be formed by proper choice of filtration conditions without adding sulfate to aid coagulation. Rate and amount of initial peroxide addition are critical, as is adequate mixing.
e Dry plutonium peroxide cake is exposed
Volume changes
charge fires. Internal temperatures reach at least 1400" C. 0 The plutonium is reduced to molten metal which separates from the slag and solidifies into a "button" at the bottom of the crucible
0 The chamber is dumped, and the solid "button" removed
The plutonium metal is pickled in nitric acid to remove any adhering slag and calcium 0
0 Pickling solution, slag, packing sand, and the crucible material are treated for waste recovery
0 Precipitate is filtered by applying vacuum, and washed with four refrigerated washes of dilute HzOz to remove rema in ing impurities
Fluorination Time cycle, hr. Gas flows, g./hr. HF 02 HF utilization, % Batch size, g.
3.5-12 100-450 0-50 5-52 350-1500
Reduction Time cycle Heating, min. Total (preparations, heating, and cooling), hr. Calcium, % excess Iodine, mole I2/mole Pu Batch size, g.
6-13 3-5 15-50 0-0.3 250-800
-Staff
Key to Operation Countercurrent absorption and elution gives maximum concentration and high throughputs. Basic equilibrium involved is: Puf3 3 W e W WResinh 3H+
*
0 Dry, room-temperature air is passed through the cake 0 Cake moisture is reduced to about 0.2 gram of HsO/gram of Pu by drying with warm air 0 Peroxide content of the filtrate and wash is destroyed by heating
Precipitation Variables Variable P u Feed Solution
Batch Size, g. Pu Pu, g./L H+,M NHzHSOs, M
Range Nominal InuestiValue gated
350-700 700 50-70 65 4.7 4.5-5.0 0-0.3 0.3
Variable Precipitant
+
Range Nominal InvestrValue gated
HZ021 % 50 Volume 50% HzOz added slowly, I. 2.4 Total volume 50% HzOzadded, I. 5.5 Addition rate, slow addition, I./min. 0.53 Agitator speed (5-in. flat paddles), r.p m. 225
30-50
1.8-3.0 4.5-6.5 0.4-0.8 150-350
Temperature
Precipitation, O C. Digestion, O C.
15 6
5-20 2-10
-Industry
Yet a person familiar with conventional metal production might shrug off the final product line as “pilot-plant scale.” This nuclear-age metal is produced in batch processes, using very simple equipment and many hand operations. The facilities a t Savannah River are owned by the U. S. Atomic Energy Commission, and operated by E. I. du Pont d e Nemours & Co. as prime contractor. This company also designed and built both the plant and laboratories. Construction began in 1950, and additions and improvements are still being made. The plant cost about 1.3 billion dollars. Five reactors are in operation; the reactors producing plutonium are fueled with natural uranium and moderated with heavy water. A heavy water production plant and uranium processing facilities service these reactors. T h e spent fuel is processed in two separation areas. The spent reactor fuel elements, containing the plutonium, are processed in heavily shielded buildings called canyons. The Purex process (3) separates the plutonium from uranium and fission products by solvent extraction. The product from this step, a dilute solution of plutonium in nitric acid, containing some residual fission products and uranium, is the feed to the plutonium production section. The plutonium is in the three-valence state, stabilized by hydroxylamine sulfate. This solution must still go through three major process steps to form the metal: 0 Ion exchange for concentration and purification 0 Precipitation of plutonium as the peroxide
Production of plutonium fluoride and reduction to the metal
Key to Operation Fluorination. A reactive cake from the precipitation step will shorten cycle time. Sulfate content below 0.02M in the precipitator feed is the most critical specification. Reduction. A dense cake from the precipitation step will increase possible batch size. Moisture content and sulfate content must be controlled for maximum density.
The chemistry of these processes is not new, but product quality and efficiency of operation depend greatly on equipment used and methods of operation. Du Pont’s experience in attaining consistently successful operation of these final three steps in processing this important product, plutonium metal, is the subject of this article. Operating conditions and procedures that are used represent a compromise among many, often conflicting, considerations of waste loss, product purity, and recycle volumes. The typical operating ranges given in the tables do not represent extremes, run-by-run changes, or uncertainties. Instead, they represent the ranges of conditions that have been handled successfully during several years of operation, and show the flexibirity o f t h e process. I n practice, fixed operating conditions have been maintained for months at a time. VOC. 53, NO. 9
SEPTEMBER 1961
687
Plant Construction a n d Equipment Ion Exchange Columns The ion exchange columns are packed beds. Dimensions and materials of construction are shown in the figures. The resin retainer a t the bottom of the column is fixed; the top resin retainer is movable and spring-loaded. This maintains a compressed bed as the resin expands and contracts during the process cycle, preventing fluidization of the bed which markedly decreases the efficiency of the operation. Two diameters of columns have been used, 7 inches and 10 inches (below). T h e length of the resin bed in the ?-inch unit has been 13 inches or 15.5 inches in two modifications. To maintain nuclear safety, the internal height of the 10-
304SS EVER-OPEN VENT \
9T
inch unit is limited further and the length of the resulting resin bed is 5 inches. Two such columns, separated adequately to reduce interaction, are operated in series with an effective bed depth of 10 inches. For sharp separation of various fractions, the column design has the minimum dead space or mixing volume at the ends of the bed that is consistent with good distribution of solution. The units are designed to exclude gas from the resin beds. An ever-open vent provides a means of escape for gas introduced in the feed and ensures relief of excessive pressurization from any source. The column feed and effluent lines are arranged and vented so that the column is always liquid-full and cannot drain accidentally-. The resin used is Dowex S O W , 10 to 127, cross-linked; 50 to 100 mesh. The particle size provides a compromise between rapid exchange and high flow rate. The resin is graded hydraulically before use and the pressure drop of the graded resin ranges in length from 9 to 15 p.s.i. per foot at standard flow (15 to 20 ml./min./sq. cm.). Grading is important as even a few per cent fines increases the pressure drop greatly. The feed is prefiltered to prevent solids from raising the pressure drop by accumulating in the resin beds: which act as excellent filters. Equipment Failure. About the only failure that can occur in these simple units is a jammed follower plate-a
16
result of improper assembly. The failure is apparent quickly. If the plate jams with the bed in the expanded position, the bed fluidizes during elution and gives low product concentrations and increasing losses. If the plate jams in the contracted position, resin is forced past the retainer when the bed expands. A properly installed column generally will give at least one year of service; the limit is set by slower flow and increasing losses. The pressure drop increases from fractured beads and the exchange capacity gradually decreases as a result of alpha radiation. Column Design. Volume change is a useful unit for design work. This unit represents the volume required for a change in composition of the feed to begin to appear in the effluent. Volume change can be used to standardize flowsheets on columns of different sizes, as long as column lengths are the same. A well-designed, packed bed has a volumc change of SOY0 of the bulk volume of the resin bed. The performance characteristics of packed beds of fine resin are a function only of length, within a wide range of diameters, and laboratory runs have been scaled up successfully by a factor of 500. For example, performance of the two beds (10 inches in diameter, 5 inches deep) in series corresponds to the laboratory performance of a single 10-inch-long column of 1 sq. cm. area. Since processing rates are based on fioiv per unit area, increasing column
\
1.
Expanded polyethylene foam 2. Column top 3. Polyethylene-coated cadmium plate 4. Teflon wiper ringsglass reinforced 5 . Metal wiper springs 6. Stainless steel follower plate spring 7. Column body 8. Column bottom 9. Top screen I 0.TOP support plate 1 1 . Screen support plate (top) 12. Wiper holder 13. Bottom support plate 14. Bottom screen 15.Screen support plate (bottom) 16. Guide tube
Body of 7-inch ion exchange column is Type 304L stainless steel
688
Details of 1 0-inch ion exchange column are shown here
INDUSTRIAL AND ENGINEERING CHEMISTRY
Pu BATCHING
INLET PRECIPITANT\ INLET
t
1/2"
PURGE LINE FURNACE HEAD / [ 'HASTELLOY c
LID
/
IMg01
1
SECTORIAL FRIT
FINGER ( I OF 3)
Peroxide precipitator
area increases the total flow. As the permissible diameter is increased, the column time cycle is therefore decreased. Safety. Both the 7-inch and IO-inch diameters are considerably larger than the safe diameter of a reflected, infinitely long cylinder, which frequently is used for design purposes. T h e larger diameters are based on the "limited safety" concept, with the limitations on the height of the column, interaction, reflection, and possible concentrations considered adequate for safe operation.
Precipitation Equipment
A precipitator (above, left) is made of glass, and cooled with Type 304L stainless steel tubes containing Freon refrigerant. The agitator has a series of flat paddles or marine propellers on a central shaft. A "boat" used for filtering, washing, and drying the cake (above, center) is made of Type 316 stainless steel and contains a 7-inch-diameter frit (porosity F sintered stainless steel 3/ 16-inch thick). The peroxide destruction step indicated on the process block diagram is performed on the filtrate in Type 304L stainless steel tanks heated electrically or with steam. Reagents are fed to the precipitators by gravity. Filtration, washing, and drying are all accomplished by applying vacuum to the underside of the boat. Fluorination Equipment The fluorination boats are made of Hastelloy C, with frits and liners made of platinum, to avoid corrosion and contamination of cakes by corrosion
Stainless steel filter boat filters peroxide cake
products. The shape is similar to that of the filtration boats. After being loaded, the boats are placed on hydraulically operated platforms that are raised to form the furnace bottoms. The boat rests on a gasket of soft Armalon (Teflon felt), which gives a satisfactory seal to the off-gas nozzle. Metalto-metal seals of stainless steel or unlined Hastelloy C corroded excessively. Gold plating did not help. Corrosion increases greatly when cycle times are longer than the minimum. The fluorination furnace enclosures are made of heavy stainless steel and lined with Hastelloy C to provide adequate strength to withstand the vigorous reactions which sometimes occurred in early operation. T h e massive structure lengthens heating and cooling times. Excess H F in the furnace off-gas is diluted with air and absorbed in a bed of limestone.
Reduction Equipment Reduction of the fluoride is accomplished in a magnesium oxide crucible. The crucible is packed into a steel pressiire chamber and is held in position with a thin layer of magnesium oxide sand. A magnesia cap is fitted over the crucible to prevent the slag from spattering onto the furnace head and the upper portion of the pressure chamber. The pressure chamber is placed on an hydraulic lift which raises it into the induction furnace and seals it against the furnace head by means of an annealed aluminum gasket. Complete assembly is shown above, right. The reduction furnace has inert gas (helium or argon) and vacuum fittings so that air can be purged from the pres-
b - 4
1 / 4 " 4
Reduction furnace details
sure chamber. The furnace head valve can be flushed with chloroform or an iodine solvent, if iodine is used in the reduction mixture. The furnace head, valves, and associated piping as well as the induction coil are mounted for convenient maintenance -replacement is occasionally necessary for each of these.
BuildingPlant construction and layout are designed not only for satisfactory operation, but must include provisions for protection against radioactivity and for containment of active materials, if spills occur. High activity solutions are treated in canyons, whose walls serve to protect operating personnel. Plutonium streams which contain little fission product activity do not require shielding, but because of the extreme toxicify of plutonium, all lines and vessels are enclosed in ducts and plastic-paneled cabinets. These cabinets house the precipitation, fluorination, and reduction equipment. The vessels are maintained at a differential of about 2 inches of HzO below the pressure in the cabinets, and the cabinets are about 0.8 inch of HzO below the pressure in the room. These conditions, maintained by using separate exhaust systems for the cabinets and the vessel vents, ensure that all leaks will result in air flows away from operating personnel. All air from vessels, cabinets, and rooms is filtered and monitored before it is discharged to the environment from a 200-foot stack.
VOL. 53, NO. 9
SEPTEMBER 1961
689
Cation Exchange Concen t r a Pion A cation exchange process is the first step in the preparation of metal. Requirements for this isolation step are: Additional decontamination Low waste losses 0 Ability to process dilute feed (as weak as waste) @ A suitable product for subsequent precipitation, both in concentration and chemical composition @ @
tion when a loaded column remains idle, or by reactions of the reducing agents (hydroxylamine and sulfamic acid). This gas can be removed, and any Pu(1V) in the column reduced to Pu(111), with a refrigerated wash of dilute nitric acid and hydroxylamine. Separation of impurities from plutonium by the resin occurs by several mechanisms:
Cation exchange originally was installed as a potential improvement on evaporative concentration to avoid safety and corrosion problems. Ion exchange has operated well; no evaporators ever have been installed for the final plutonium concentration. The exchange units are small fixedbed columns of Dowex 50W resin, with dimensions fixed by requirements of nuclear safety. Process was developed at Oak Ridge National Laboratory
0 Anionic or nonionized impurities, or weakly absorbed cations, pass through the resin unabsorbed 0 Materials that are absorbed less strongly than plutonium are displaced downward during absorption 0 Materials that are held more strongly than plutonium in the elutriant solution are left on the resin 0 Materials complexed preferentially by sulfate are selectively removed from the resin with the decontamination wash
(9).
Steady-State Batch Operation
Process Chemistry
Countercurrent operation of a fixedbed of resin is a simple concentration method that is frequently overlooked in favor of more sophisticated but less efficient methods. Flow direction is the key to satisfactory performance with this plutonium cation exchange system. Traditional cocurrent operation with downflow absorption and elution is not feasible for operations requiring large throughput. For process stability and low losses, the movement of plutonium down the column during absorption and decontamination should balance the movement upward during elution. During absorption, a band of saturated resin grows down the column to a point governed by the plutonium capacity of the resin. Below the loaded band and a short exchange zone, the clean resin serves as a stripping section and reduces losses to a negligible level in the space of only a few inches. The decontamination wash of dilute sulfuric acid causes a further downward movement of the loaded band. During upflow elution, the elutriant solution moves the plutonium back up the stripping section, then is enriched progressively as it flows through the loaded band. At every stage of elution, the product solution leaves the column after contact with the most saturated resin left, so that maximum product concentration is reached. For design work, possible stable cycles can be calculated from resin saturation data and distribution coefficients, as for any ion exchange process. I n the plant comparatively rapid flow rates are actually used to achieve minimum time cycles, causing greater spread during absorption and reduced efficicncy during elution. T h e net effect is that increased
As described in the flow diagram, plutonium is absorbed on Dowex 50 from the dilute nitric acid feed solution produced by the solvent extraction process and is eluted by strong acid. In this plant, the Pu(II1) system is favored over Pu(1V) for several reasons: 0 No further feed adjustment is required for ion exchange, as Pu(II1) is the product of the prior solvent extraction step 0 Sulfate decontamination wash would complex Pu(1V) with high losses 0 Capacity of the resin (hence the column batch size) is larger with Pu(II1) 0 Elution curve of Pu(II1) is sharper, resulting in minimum time cycles and higher concentrations This oxidation state appears to have unique advantages in the peroxide precipitation process
One potential problem with Pu(II1) is its instability with respect to oxidation in strong nitric acid. The reaction can become autocatalytic with vigorous evolution of nitrogen oxide gases. However, the oxidation rate is dependent on concentration. By experience, there is little chance of rapid oxidation if sulfamic acid stabilizer is present and if the peak concentration of plutonium in the strong-acid eluting solution is less than 130 grams per liter. By avoiding concentrations above this level, many months of operation have been maintained without appreciable oxidation. The Pu(II1) is stabilized by hydroxylamine in dilute nitric acid solutions and by sulfamic acid in strong acid. Gas on the resin interferes with fluid flow and with the solution-resin exchange. I t may be formed in the oxidation of plutonium by radiolysis of solu-
690
INDUSTRIAL AND ENGINEERING CHEMISTRY
elutrjant volume is required to maintain stability. Process Experience Absorption. Typical values for some of the variables are shown with the flow diagram. The ranges of feed plutonium and acidity given with the flow diagram have been set by various solvent extraction flowsheets. The wide range in absorption batch size represents the extremes from conservative start-up conditions, with dilute feed and high acid, to strong plutonium and low acid. The nominal load on the 7-inch-diameter column is 700 grams. Much more dilute feed (1 x gram per liter) has been processed during start-up and flushing operations. I n another specialized application, feed up to 6 grams Pu per liter has been standard. The absorption batch size is limited primarily to prevent excessively high plutonium concentrations (oxidizing conditons) during elution, as well as for control of losses. The performance of the peroxide precipitation step following this step is sensitive to the concentration of sulfate, and a result of oxidation of plutonium during elution is the oxidation of sulfamic acid to sulfuric acid. For example, using the 7-inch columns, although there was no apparent oxidation as the batch size was raised from 750 to 775 grams, sulfate in the eluate (which had been below detectable limits) reached 0.02M and interfered with precipitation. Absorption at 20 ml./min./sq. cm. gives fairly sharp absorption bands and reasonable time cycles. The limit here is primarily the pressure drop in the system and the available gravity head. It has been considered undesirable to pump-feed the columns and risk excessive pressurization. The losses, as measured, include the total column waste from absorption and from the decontamination wash. The loss can be set to almost any desired value by adjusting the operating conditions. Losses as low as 1 X lO-*yO
Table
I.
Decontamination Factors for
Volume Changes of Wash
Ru
9 24
2-4 2-4
Uranium U/Pu U/Pu 0.5
4-20 30-80
>lo0
>20
Zr-Nb 1-10 10-70
Wash composition, M H2S04
NH20H.'/2HeSOd Volume changes"
0.19 or 0 . 2 5 0.05 6, 12, 18, or 24 2-3
Flow rate, ml./min./sq. cm. Volume change is 50% of the physical volume of the bed.
have been maintained for extended periods of time, but the low level requires more elutriant, with resultant longer cycles and larger recycles, for any given set of other conditions. For minimum cycle time, the elutriant volume may be reduced until the losses reach an acceptable limit (generally 1 x 10-2%). Decontamination Wash. The volume of the sulfuric acid wash under the conditions specified in Table I is set in a sliding scale, according to the amount of uranium and zirconiumniobium in the feed and the required decontamination. Concentration of the wash is a matter of choice; the O.19M solution gives slightly poorer decontamination than the 0.25M solution but the losses are considerably lower. Below a concentration of 0.19M there is an increasing loss in decontamination power. Standard wash volumes used have been 6 to 24 volume changes. Approximate ranges in decontamination factors which have been obtained are shown. Ruthenium apparently exists in the feed in two forms, one of which does not absorb when the column is being loaded. T h e other form is eluted with the plutonium; the range in decontamination represents the split between these species. None of the absorbed form is removed with the sulfuric acid wash. Decontamination from uranium is obtained by several mechanisms, including break-through and sulfate complexing. I t is absorbed more weakly than plutonium. At low concentrations it is displaced down the column, while a t high concentrations it will break through and appear in the absorption waste. Displacement and break-through points are functions of the uranium, plutonium, and acid in the feed and accouqt for the spread in decontamination values without washing. Complexing of uranium increases decontamination when the wash is used. Decontamination from fission product Zr-Nb without the wash is a case of preferential elution of the plutonium, leaving the Zr-Nb on the resin. Sulfate complexing increases the decontamination when the wash is used. The decontamination range shown apparently is a function of the history of the feed. The two fission products, Zr-95 and Nb-95, are analyzed together and treated as one; they have similar but not identical behavior on the cation exchange resin. Variations in their ratio in the feed therefore give variations in decontamination. Also, off-standard solvent extraction operations apparently form a n abnormal species of Zr-Nb which is not removed readily and can reduce decontamination below the normal lower limit. T h e species has not been identified, but the circumstances under
Handling of large quantities of fissile and radioactive materials add to the safety problem at Savannah River. Hazards can be classified as 0 Nuclear 0 Radiation 0 Ingestion of radioactive materials 0 Normal chemical plant hazards No distinction is made in protecting against these hazards-an atomic blast can be prevented by the same methods used to prevent chemical explosion, or automobile accidents. In all cases 0 A procedure must be safe 0 The operator must follow it Du Pont’s safety record is maintained by insistence on a written procedure covering every operation performed at the plant. Devising and reviewing procedures are the joint responsibility of the plant production personnel, and the plant assistance technical group. If very important factors related either to plant safety or successful processing limits are involved, procedures must also be approved by the Savannah River Laboratory. Nuclear safety, of course, falls in this important category. Throughout, procedures are based on the assumption that one can occasionally expect two consecutive errors in operation. The combined effects of these two errors must not cause a nuclear accident. And a failure in communication is in-
cluded as a possible error. This is the minimum safety factor-in many cases, greater protection is in effect. Since operating personnel occasionally invent brand new mistakes and since process changes must sometimes be made, a continuous re-evaluation is followed. Every violation of procedure and every incident that results in an unexpected concentration of plutonium are investigated. Nuclear safety is incorporated into equipment design as well as operating procedure. For example, where possible, vessels are sized to hold less than the critical amount of plutonium. Equipment is kept as simple as possible to reduce the chance of failure and to lower maintenance. The policy of complete containment of radioactive materials has fed to design of equipment areas in which all operations are performed on one side of a barrier, while all maintenance is performed on the other side. Spills which may occur due to maintenance will not contaminate the control room. A mechanic remains in the active area for only a safe period of time. Protective clothing always is worn, and face masks are used for moderate levels of contamination. At high levels, additional plastic clothing is worn, and pure air is supplied for breathing. Operating procedures must consider that while a mechanic is working on equipment, his body alters the degree of neutron reflection around fissile material.
which it is encountered indicate that it is formed by association with degraded organic material from the solvent. Other decontamination agents have not been tested widely as a replacement for sulfate because of such specific objections as potential precipitation of plutonium, interference with normal waste handling, and corrosion. Dilute fluoride ion has been tested, but it interfered with operation considerably. Elution. T h e composition of the elutriant has not been varied deliberately since start-up. Nitric acid concentration is about the maximum compatible with a Pu(II1) flowsheet, and the sulfamic acid is near the solubility limit. The volume of the elutriant required is based primarily on the waste losses, which are a function of the absorption conditions, column load, volume of decontamination wash, and the flow rates in these steps. T h e volumes shown on the flow diagram cover the range from best-to-worst combinations of these variables. The elutriant flow rate is a compromise between efficiency and time cycle. With the 50- to 100-mesh resin, there
is little difference in elution characteristics at rates below 0.2 ml./min./sq. cm. As the rate is increased to obtain shorter time cycles, the elution curve for plutonium begins to spread. Average concentration of a given volume of product solution decreases, the column heel increases (for a given elutriant volume), and losses increase. The effect of elution rate on losses rapidly becomes greater a t rates higher than 0.5 ml./min./sq. cm., but product concentration is only slightly affected. Reconditioning. The exact composition of the reconditioning wash is not important, and is a matter of convenience in make-up. Use of the same solution for reconditioning and the refrigerated degassing wash simplifies feed preparation. T h e volume of the reconditioning solution has not been varied because it represents the volume necessary to remove strong acid and it is independent of practically all other conditions. The initial part of the reconditioning step is made a t the same flow rate as the elution because it is still moving elutriant ahead of it. Once the acidity
.
VOL. 53, NO. 9
SEPTEMBER 1961
691
ol the solution leaving the column starts to decrease, the rate can be raised. This reduces the time cycle without affecting the process appreciably. Recycle Stream. Concentration of the product fraction and amount of plutonium in the recycle stream are largely a matter of choice. The 50 to 70 grams per liter concentrations represent solutions that are compatible with the subsequent precipitation step. Higher concentrations can be obtained if smaller volumes of the product solution are taken from the peak of the elution curve, with increased recycle. The amount of recycle also is increased when large volumes of elutriant are required to counteract large decontamination washes or other unfavorable conditions. Various combinations give the wid? variation in recycle (flow diagram).
Miscellaneous Operations. ‘The conditions for the refrigerated wash and fission product removal steps are shown in Table 11. T h e primary use for the refrigerated wash is for degassing following delays between process steps (such as overnight or weekend shutdowns) or for known presence of Pu(1V). It causes essentially no loss. Occasionally, accumulated fission products on the column must be removed to reduce the radiation so the unit can be replaced or maintained without shielding. Oxalic acid solution has been very effective for this purpose; ammonium citrate also has been used. Since these agenu also complex plutonium, the plutonium heel on the column must be removed first with a n extended product elution step to prevent loss (IO volume changes of product
Table II. Refrigerated Wash and Fission Product Removal Refrigerated wash Composition, X “08
NH~OH*’/~HZSOJ Flow rate, ml./min./sq. cm. Volume, volume changes Fission product wash Oxalic acid concentration, AI Flow rate, ml./min./sq. cm. Volume, volume changes
0.1 0.05
5-15 10
0.5
0.2
5
elutriant is standard). The volume of oxalic acid solution that is required for decontamination depends on the total activity and form of the fission products. In typical cases, the indicated volume has reduced radiation by a factor of 20.
Precipitation of Plutonium Peroxide The precipitation of plutonium (IV) peroxide with hydrogen peroxide is a standard method of isolating a solid plutonium compound from solution. This system is desirable because : I t provides excellent decontamination from cationic impurities 0 The only foreign material introduced (hydrogen peroxide) can be easily destroyed by heating the filtratr. This simplifies recycling
A disadvantage of the method lies in the handling of dissolved and solid peroxides. These are violently unstable under certain conditions. From prior experience, this precipitation method was selected for plutonium isolation a t the Savannah Kiver Plant, with feed to be supplied by the cation exchange system. Considerable development work was necessary to establish the best precipitation conditions. The plant now produces precipitates which filter and dry rapidly, are dense and even visibly crystalline, and are so reactive and free from impurities that the subsequent fluorination step has been greatly shortened. A flow diagram is presented on pagr 686. Hydrogen peroxide solution is added to batches of plutonium concentrate from the cation exchange step. ‘The resulting precipitate of plutonium peroxide is filtered, washed, and dried to a state suitable for fluorination. Process Chemistry
The chemistry of the precipitation of Pu(IV) peroxide from nitric acid solutions is complex. I t involves the oxidation-reduction behavior of Hz02 and Pu ( I ) , formation of (Pu-O-Pu)+b hydrolysis dimers (in the weaker acid solutions) (5), and the formation of two different crystalline phases of the solid
692
peroxide (6’) (both of which contain coprecipitated anions) (2). Conditions for producing each of the crysralline phases were established by Leary and others (5), who found that an undesirable colloidal face-centered cubic form would precipitate from nitric acid solutions below 2 M in acidity, but that at higher acidities a desirable hexagonal structure would form. ‘This can be explained by the larger proportion of hydrolyzed Pu(1V) inlow acid. These results also tended to clarify the function of coprecipitated sulfate ion. For many years sulfate had been added to the feed for “improving” the precipitate. Sulfate is a n effective coagulant for posixively charged colloids, such as the peroxide precipitate formed in low acid ( d ) . These results also indicated that sulfate had little effect on which phase would precipitate. I n plant operation it was found that the desired hexagonal phase could be formed in the first place through proper choice of precipitation conditions. There was then no need for sulfate addition. The elimination of sulfate proved to be of extreme importance-cakes formed under satisfactory precipitation conditions without sulfate were about six times as dense and filtered six times as fast as those formed under poor conditions with sulfate present to aid coagulation. I n addition, these cakes proved to be more satisfactory for fluorination and reduction. T h e fluorination furnace cycle was shortened greatly when it was no longer necessary to remove the added sulfate from the cake by prolonged high-temperature roasting. Increased density of the peroxide was reflected in increased density of the tetrafluoride. ‘rhe gain was a n increase in batch weight and productivity in the fluorination and reduction equipment, in which the volume was limited.
INDUSTRIAL AND ENGINEERING CHEMISTRY
‘l‘he fact that the Pu in the feed solution to the precipitation step is present initially as Pu(II1) is thought to aid the Precipitation, since oxidation to Pu(1V) by HzOz is not instantaneous. The lower Pu(IV) concentration seems to aid The formation of larger crystals. Actually, precipitates from solutions which initially contained Pu(1V) and no sulfamic acid, while still satisfactory, were not quite as good as those made from normal cation exchange concentrate. Further experimental laboratory work would be rewarding in clarifying the precipitation mechanisms and the complex equilibria. Process Experience The following were of prime importance to the production of a dense, free-filtering cake, suitable for fluorination:
e Agitation and addition rate matched to prevent zones of low acidity or es-
cessive turbulence &t Control of addition rate until the critical phase of the precipitation is complete @ Addition of sufficient total peroxide for good yields Prompt filtration e Prompt and thorough washing Two step-Le., first cold, then warm---air drying Precipitation. Conditions of precipitation are presented with the flow diagram. A batch of concentrate is cooled, and H202 solution is added through a rate-controlling orifice until a predetermined volume has been added ---enough to raise the total peroxide concentration to a t least 2 M . Under thesr conditions, in a n acidity of about 4 M , all traces of the red peroxy-complex arc gone and the slurry is tan c.olored. AI this .point, the remaining volumr o f H z 0 2 is added rapidly to
Staffilmdustry reduce the solubility of the precipitate by raising the total peroxide concentration to about 6M HzOz (about eight times the stoichiometric amount). The temperature is then lowered to 6' C. and the precipitate is digested for 30 minutes prior to filtration. The critical nature of the volume added a t the slow rate cannot be overemphasized ; filtration times increase sharply and density decreases if the peroxide is even a small fraction below the minimum amount. Also, the rate during slow addition appears to have a disproportionate effect. Too fast a rate gives markedly poorer performance, although it is difficult to separate the rate effect from mixing efficiency. The total volume of peroxide added also affects the density, in addition to the expected effect on filtrate loss. The density decreases at the minimum total volume shown. Agitation during the addition of H202 is also of great importance. 'roo high an agitator speed will lead to the formation of fines, and too low a speed (or too fast an addition rate) allows precipitation to proceed in zones of incompletely mixed Ha02 (low-acid, high-peroxide conditions) which can produce the colloidal form of the precipitate. Thus, it is of primary importance to prevent the existence of any region in the precipitator where the acidity falls below 2 M , and to delay rapid additions until a proper foundation has been laid-i.e., a peroxide content of at least 2M (in an acidity of about 4 M ) . The upper limit on agitator speed, above which excessive fines are formed, corresponds to a blade tip-speed of about 5 feet per second for either turbinetype or flat-paddle agitators, both of which create considerable turbulence (this tip-speed limit holds for blades 2.5 inches and 5 inches in diameter). Maximum tip-speed for marine propellers, which give good dispersion with little turbulence, is about 8 feet per second. Temperature conditions are not critical, but refrigeration minimizes decom-
position of the H202 catalyzed by the impurities in the feed. Feeds containing as much as 1.5 grams Fe per liter are handled satisfactorily with the cooling available. Precipitation yields are about 99.770. Decontamination factors are comparable to those of Leary ( 5 ) . Filtration. The cold precipitator slurry is filtered through a porous stainless steel frit. Filtration times vary inversely with the degree of vacuum available on the downstream side of the filter frit. For a vacuum of 15 inches Hg, a filtration time of about 15 minutes is common. If a filtration is long (say an hour) the cake warms up and peroxide decomposition may become excessive. Washing. The wash reduces the concentration of residual HzOZ,"01, and NH2HS03 in the cake, measurably increasing the stability of the cake and the purity of the final metal. Four 1.5-liter refrigerated washes are made with 2% Hz02. The wash liquor is added to the filtrate and both are sent to the peroxide destruction step. Washes of 5% and 1Yo HzOz also have been used ; the residual 5% solution appeared to slightly increase the instability of the cake during further processing, while the 1 solution hydrolyzed the cake somewhat, making it difficult to dry. Drying. Dry room-temperature air is passed through the cake at about 5 SCFM for 2.5 hours; this is followed by dry warm (55 C.) air at about 5 SCFM for 2 hours. This reduces the final average moisture content of the cake to about 0.20 gram HzO per gram Pu, a desirable level for feed to the subsequent fluorination step. (The water content is based on an empirical formula weight of 319 for plutonium peroxide). Constant-rate drying conditions prevail, and the air leaving the bed averages a moisture pickup of about one half of its adiabatic saturation value at the measured air flow. If all the drying is done with warm air, the cake produced is hard, lumpy, and difficult to fluorinate. This is believed caused by bridging between
precipitate particles; plutonium peroxide may dissolve in the warm residual liquid in the cake and reprecipitate as the solution in the cake evaporates. The effect is minimized by drying initially with room-temperature air (down to a moisture content below 1 gram HzO per gram Pu). Peroxide Destruction. The large excess of H202 (about 4 M ) in the filtrate and wash solution must be destroyed for safety in handling this solution. Accordingly, the solution is heated to 50' C., held there for 45 minutes, and then heated to 90' C. briefly to destroy the last traces of peroxide. The two stages of temperature control are for safety; the raw filtrate undergoes runaway decomposition if heated directly to above 60' C. This treatment routinely leaves the filtrate with an € 1 ~ 0 ~ content of less than 0.02%. Low acidity and ihe presence of sulfate both interfere with this step. Safety. A primary safety consideration is the instability of HZOZif it contacts impurities either before or after it enters the process. No piping is connected to drums of H2Oz other than a 6-inch length of pipe with a valve. Solution is dispensed as needed into small, covered stainless steel containers which are moved to the proper addition funnels. To allow for sudden decomposition within the precipitator or the peroxide dei3truction vessels, l1/2-inch vent lines are used and reserve cooling capacity is provided. Because the precipitators and associated equipment are larger than the infinitely safe diameter (about 6 inches) for plutonium, adequate control of the batch size 1s essential to prevent a criticality accident. This is accomplished by using instruments to analyze samples of each batch of concentrate by the gamma-absorption technique ( 8 ) ; pneumatic liquid-level indicators in the batching tanks, the precipitators, and the peroxide destruction tanks; and automatic pneurnatic shutoff controls to prevent overfilling the precipitators
Fluorination and Reduction to Metal Conversion of plutonium peroxide to plutonium tetrafluoride with hydrogen fluoride, followed by reduction of the tetrafluoride with calcium, is a standard method for the production of plutonium metal (7). Most of the controlling variables had been explored thoroughly before the construction of the Savannah River Plant and the existing technology was the basis for the original design of the plant. Many improvements were made to the original operation during the following years. Most of these were matters of technique.
The most significant advance was an understanding of the close relationship between the entire series of isolation operations and the reduction yield. 'The single most important discovery was that the sulfate content of the peroxide cake regulated the ease of fluorination. When sufficient knowledge of the controlling variables was gained, the fluorination furnace cycle was shortened to a fraction of the original time. Also, greater density of the fluoride increased the reduction batch sizes, allowing elimination of the iodine booster.
Process Chemistry
The basic process a t this plant, shown in the flow diagram, was unchanged from the normal sequence. Plutonium peroxide is exposed to hydrogen fluoride in a furnace at temperatures up to GOO0 C. The resulting tetrafluoride is reduced with metallic calcium in an induction furnace. The molten plutonium separates from the slag and solidifies into a "button" in the bottom of the MgO crucible. The over-all efficiency of the process VOL. 53, NO. 9
0
SEPTEMBER 1961
693
may be based on time cycles, batch size, reduction yield, and utilization of HF, and is determined largely by the composition of the peroxide cake. Specifically, the sulfate and moisture content must be controlled properly. The sulfate concentration in the feed to the peroxide precipitation step determines the fluorination time that is required for good reduction yields. Very short cycles can give good yields in the absence of sulfate (a good yield is arbitrarily defined as over 95y0). The exact chemical effect of sulfate has not been established. For example, one indication of excess sulfate is excessive splattering of the reduction charge over the inside of the chamber. In addition to harmful effects on furnace cycle or reduction yield, sulfate in the original peroxide cake decreases the bulk density of the final tetrafluoride, and hence can decrease the possible batch size. The degree of moisture in the peroxide cake at the beginning of fluorination also affects the density; the drier the cake, the greater is the density. However, cake dryness is limited by excessive reactivity in the fluorination step. The gross effect of sulfate on the performance may indicate something of the mechanism of decomposition of sulfamate. The peroxide cake undoubtedly contains some sulfamate ion when precipitated from the 4.7M nitric acid-0.3M sulfamic acid feed solution. Since sulfate effects become apparent a t 0.02M in the feed, sulfamate apparently does not convert to sulfate at the plant furnace conditions. T h e important variables in the reduction operation are:
e Batch size of plutonium
a Amount of calcium 0
Atmosphere of pressure chamber
A large batch size may allow eliminating iodine booster; the amount of calcium influences purity ; and traces of moisture or air depress yield. The work discussed is not a definitive study because of limitations on the equipment and on available analytical techniques. Several items are worthy of further xyork, particularly as the basis for any new installations. The actual role of sulfate should be determined as should the effect of the crystal structure of the peroxide. The actual rate of conversion of good cakes should be measured to establish a true minimum fluorination time. The time is now limited by heating and cooling the massive furnaces-fast-acting furnaces might allow even shorter cycles.
Process Experience Fluorination. Fluorination and reduction data are presented with the flow diagram. T h e influence of sulfate on fluorination may be illustrated by the
694
INDUSTRIAL
required time cycles. T h e original standard cycle required 8 to 12 hours, and included 1.5 hours a t 150” C., 3.0 hours a t 600” C., and associated heating and cooling periods. Performance was good with sulfate concentrations ranging from 0.075M to 0.15M in the precipitation feed. As the sulfate in the feed was decreased gradually to less than 0.02M, it \vas found that the 150’ C. step could be eliminated, and then the GOO’ C. step could be shortened drastically. The rninimum cycle in routine use with low sulfate is 3.5 hours, during which the cake is heated directly to 6003 C. and immediately cooled. Minimum required cycle is quite sensitive to the limit of 0.02M sulfate. For example, this figure is reached due to slow conversion of sulfamate to sulfate when concentrate is stored for more than a day. A slightly longer cycle is required for feed stored over a week end. A high utilization of hydrogen fluoride was obtained from the enhanced reactivity and short furnace cycles obtained with well-prepared peroxide cakes. The maximum utilization of 52y0 (shown with the flow diagram) was obtained with large, sulfate-free batches, which had short time cycles. The minimum utilization shown corresponded to small, high-sulfate batches, which required long cycles. Oxygen aids in the removal of sulfate from the cake: and is used in cases of known high sulfate. But it does not appear necessary with low sulfate. The moisture content of the airdried peroxide cakes must be controlled between 0.20 and 0.35 gram of water per gram of plutonium (based on a n assumed formula weight of 319 for plutonium peroxide) or difficulties occur in the furnace when hydrogen fluoride flow is started. Fresh cakes that are drier than this range frequently react violently and cake is expelled from the boat or even the furnace. Cakes with a moisture content above the limit sometimes decompose to moist, foamy masses that slowly boil out of the boat into the furnace. .4ged, dry cakes lose the extreme sensitivity that characterize the fresh cakes. Washing with alcohol to promote drying is not recommended; alcohol was part of the original flowsheet but caused many violent reactions. Reduction. Calcium alone can be used to reduce plutonium tetrafluoride to pure plutonium metal above some critical batch size determined by the thermal properties of the equipment. Iodine booster may be used to allow reduction of smaller batches or of incompletely converted tetrafluoride. However, iodine causes various complications and corrosion problems. Elimination of iodine ranks with the bigger batch size as a prime reward of increased tetrafluoride density. Reference has been made to the increased bulk den-
AND ENGINEERING CHEMISTRY
sity of the tetrafluoride that resulted from various changes. The bulk density is defined as the grams of plutonium per cubic centimeter in the mixed reduction charge in the pressure chamber; this fixes the possible batch size for reducticn. With a normal calcium excess of 24 to 3oyO,,a plutonium content of over 1 gram per cc. is good. The reduction with calcium also serves as a n auxiliary decontamination step, in accord with the liquid-metal extractions of some pyro-metallurgical systems. The amount of excess calcium was varied in d series or experiments from 15 to 50% over the stoichiometric requirement for reduction and gave measurable changes in the purity of the plutonium. The actual upper limit on the amount of calcium used is set by the heat conservation characteristics of the system. Too much calcium represents a heat sink that can interfere with proper melting and coalescence of the charge. The humidity of the processing line and the atmosphere in the pressure chamber also have large effects upon the reduction performance. Very erratic and poor yields have been obtained when the dew point of the ventilation air esceeded -20’ C. in the cabinets used for operations starting with the drying of the peroxide. The exact effect of the atmospheric moisture and the part of the process that is affected never have been identified. Poor yields also occurred if the pressure chamber containing the reduction charge was not sealed or purged with inert gas. Even small amounts of air cause splattering and poor coalescence. literature Cited
(1) Connick, R . E., McVey, W. H., “Transuranium Elements,” Natl. Nuclear Energy Ser., Div. IV, 14B, Pt. I, p. 445, McGraw-Hill, New York, 1949. (2) Hamaker, J. W., Koch, C. W.,Ibid., p. 666. (3) Joyce, A. W., Jr., Peery, L. C., Sheldon, E. B., Chem. Eng. Progr. Symposium Ser. 56, No. 28, 21-9 (1960). (4) Leary, J. A., Los Alamos Scientific Laboratory, Los Alamos, N. Mex., LA-1913 (October 5, 1955). (5) Leary, J. A , , Morgan, A. N., Maraman, W. J., IND. ENG. CHEM.51, 27 (1959). (6) Mooney, R. C. L., Zachariasen, W. H., “Transuranium Elements,” Natl. Nuclear Energy Ser., Div. IV, 14B, Pt. 11, p. 1442, McGraw-Hill, New York, 1949. (7) Morgan, A. N., Jr., others, “The Los Alamos Plant for Remotely Controlled Production of Plutonium Metal,” Proc. 2nd Intern. Conf. Peaceful Uses Atomic Energy, Geneva, 17, P/531, 537-44. 1958 (8) Thtirnau,
D. H., Anal. Chem. 29, 1772-4 (1957). (9) Tober, F. W., “Concentration and Purification of U, Pu, and Np by Ion Exchange in Nucle arly Safe Equipment,” Proc. 2nd Intern. Conf. Peaceful Uses Atomic Energy, Geneva, 17, P/520? pp. 57444,1958.