WHITNEY H. MEARS, ROBERT V. TOWNEND, ROBERT D. BROADLEY, A. D. TURISSINI, and RICHARD F. STAHL General Chemical Research Laboratory, General Chemical Division, Allied Chemical Corp., Morristown, N. J. ,
Removal of Some Volatile Impurities from Uranium Hexafluoride The experimental data reported can be used to design a distillation column capable of refirling crude uranium hexafluoride to high purity. SoIubiI ities, I iquid-va por e quiI ibria, a nd dist iI Iat ion ex pe riment s a re a c curate enough for engineering calculations
-
IN
experimental work with a glass still vanadium oxyfluoride, from a nearly saturated solution, was separated from uranium hexafluoride; the low. solubility of the oxyfluoride did not interfere. The authors have purified crude uranium hexafluoride, removed vanadium impurities by a combination distillation method involving selectively condensing, solidifying, and separating solid vanadium oxyfluoride from the vapor, and thus obtained specification-grade uranium hexafluoride.
Preparation of Materials The uranium hexafluoride was specification-grade material from the U. S. Atomic Energy Commission, Oak Ridge, Tenn. Vanadium oxyfluoride was prepared by the reaction of vanadium pentoxide with gaseous fluorine at about 450' C. (8),and degassed a t liquid nitrogen temperature by standard techniques. I t contained 40.9 to 41.3 weight % vanadium (41.I% theoretical). Vanadium pentafluoride was synthesized by the reaction between reagentgrade vanadium metal and gaseous fluorine a t about 500' C. (4). The samples, after degassing, averaged 35.0 weight % vanadium (34.9% theoretical). Molybdenum hexafluoride, tungsten hexafluoride, and antimony pentafluoride were prepared by General's Baker and Adamson Division and degassed before use. Infrared studies indicated them to be substantially pure, compared with published absorptiqn spectra.
by distilling weighed amounts of pure uranium hexafluoride and the desired solute into Fluorothene tubes under vacuum. These were thermostated a t 70' C.; a single liquid phase, free of solid, was taken as evidence of complete solubility. Solubility of antimony pentafluoride in uranium hexafluoride could not be determined in Fluorothene tubes, which tended to fracture, though stable with either component alone. Therefore, it was measured in glass, intensively dried by baking and evacuation to prevent the self-regenerating reactions : 2Hz0 4HF
Solubility of Fluorides in Uranium Hexafluoride. Exact solubilities of the fluoride impurities in uranium hexafluoride were not required, if they were greater than 10 weight %. Therefore, no attempt was made to extend the data if this quantity dissolved completely in uranium hexafluoride at 70' C. Unless otherwise stated, solubility was measured
+
The white uranium hexafluoride turned greenish and melted to a greenish black solution when placed in contact with the antimony pentafluoride, itself a waterwhite liquid. This was probably caused by some chemical reaction. Vanadium Oxyfluoride. Solubility was measured in three ways. In the first, several Fluorothene tubes contain; ing uranium hexafluoride and excess vanadium oxyfluoride were equilibrated at 70' C. for 1 week, with hourly shaking each day. Then the oxyfluoride excess was allowed to float to the top of the
Solubility of Fluorides in Uranium Hexafluoride Solu-
Cornpound
Physical Measurements
+ UFs 4HF + U02Fz + Si02 +. SiFc + 2 H 2 0
VOFa
VFI
bility (70' C.),
Wt. %
Remarks
0.6
0 . 7 7 5 O C., from liquidvapor equilibrium 0.8 From freezing point lowering, probably a high value
> 10
M ~ F ~S 2 . s 380 wF6 UOzF2 SbFa
>30 < 0.1 >40
c.
Chemical reaction observed
solution, the tubes standing undisturbed for 2 days and being checked visually for completeness of solid separation. Finally, they were quenched in liquid nitrogen and their bottom sections, containing only the quenched solution, were sawed off, dropped into water, and analyzed for uranium and vanadium. A solubility of 0.6 weight yo vanadium oxyfluoride was found. I n another measurement, an excess of vanadium oxyfluoride was added to pure uranium hexafluoride in the vaporliquid apparatus and brought to equilibrium at 75' C. Samples of liquid phase analyzed for vanadium indicated the solubility to be 0.7 weight yo. Finally, the freezing curve of a sample of the liquid phase from the above test was determined using a calibrated platinum resistance thermometer and recorder. The freezing points of two such saturated solutions were found to be 62.94' f 0.06' C. [literature value, 64.02' f 0.05' C. (73) for pure uranium hexafluoride]. Using 4570 cal. per gram mole ( 7 7 ) for the heat of fusion of the hexafluoride and the standard one-term equation (7) for the freezing point lowering of an ideal solution, the solubility of vanadium oxyfluoride was calculated as 0.8 weight %. The value from this third technique may be somewhat high because of the possible presence of other impurities such as traces of hydrofluoric acid, but it confirms that from analytical methods. Liquid-Vapor Equilibrium. In proper design of the uranium hexafluoride distillation column, liquid-vapor equilibrium relations had to be determined for the major impurities expected in the crude uranium hexafluoride. When this work was begun, the literature (5,6, 74, 75) indicated that molybdenum hexafluoride (7, 7 4 , tungsten hexafluoride ( 2 ) , and chromyl fluoride (5) would be more volatile than uranium hexafluoride, and vanadium pentafluoride and vanadium oxyfluoride would be less volatile. Publications appearing during this VOL. 50, NO. 12
DECEMBER 1958
1771
work (3, 16) reported new vapor pressures which confirmed the experimental observation that vanadium pentafluoride also was more volatile than uranium hexafluoride. The relative volatility of vanadium oxyfluoride was still in question. Liquid-vapor equilibrium studies were carried out in which uranium hexafluoride formed one component, while molybdenum hexafluoride, vanadium pentafluoride, or vanadium oxyfluoride was the second. Experimental Method. An allMonel apparatus, constructed for this purpose (Figure l ) , consisted of a cylindrical static equilibrium chamber fitted with sampling valves at top and bottom. By means of these and suitable accessory Monel piping, the apparatus could be charged and liquid or vapor samples withdrawn for analysis, all manipulations being carried out under vacuum. The interior of the equipment was preconditioned with chlorine trifluoride. The apparatus was charged with the required amounts of pure uranium hexafluoride and of the desired impurity by standard vacuum transfer techniques. The apparr?tus and contents were then immersed in an oil thermostat a t 75' C. until vapor-liquid equilibrium was established. Several samples of the liquid and vapor phases were withdrawn alternately, under vacuum, into Fluorothene tubes and analyzed for the desired impurity. For optimum results, a slow rate of sample removal is necessary. Molybdenum Hexafluoride-Uranium Hexafluoride. I n this system, at equilibrium at 75' C., the average analysis of the liquid phase was 585 =k 15 p,p.m. of molybdenum; of the vapor phase, 983 i 22 p.p.m. Thus, at this concentration, the molybdenum was enriched in the vapor phase by a
factor of 1.7. The ideal or Raoult's law value is calculated to be 1.6, using the vapor pressure equation for molybdenum hexafluoride ( 7 4 ) : %logPmm,= 7.409 - 1394.9/T°K and the known vapor pressure of uranium hexafluoride. Thus, at these dilutions, a solution of molybdenum hexafluoride in uranium hexafluoride can be considered to tehave ideally, for engineering calculations. Vanadium Oxyfluoride-Uranium Hexafluoride. When vanadium oxyfluoride was studied at 75" C., equilibrium was obtained between a liquid phase containing 37 i 5 p.p.m, of vanadium and a vapor phase containing 122 =t 11 p.p.m., the vanadium oxyfluoride being enriched in the vapor phase by a factor of 3.3. These enrichment results cannot be correlated directly by a Raoult's law calculation using the available vapor pressures of vanadium oxyfluoride (9, 72) and uranium hexafluoride at 75' C., because the former is normally a solid a t this temperature. I n equilibrium solution, the partial pressure of the oxyfluoride is a function of the vapor pressure of its pure, supercooled liquid at this temperature. T o obtain the experimental enrichment factor of 3.3 when uranium hexafluoride at 75' C. has a vapor, pressure of 1592 mm. (13), the ideal vapor pressure of supercooled liquid vanadium oxyfluoride would have to be about 5250 mm. a t the same temperature. The vapor pressure of the metastable liquid can be approximated in another way. Because a saturated solution of the oxyfluoride is in equilibrium with its solid at the same temperature, the partial pressure of the dissolved material equals the sublimation pressure of the solid at the same temperature. If ideality is
PIPE C
A- SAE MALE COMPRESSION FITTING ( I/2"BRASS) BOTTOM MACHINED FLAT. B- HOKE VALVES (I/4"MONEL) TEFLON PACKED. c ALL PARTS EXCEPT A
-
LIQUID
LEVEL
Figure 1. Vapor-liquid equilibrium studies were carried out in a cylindrical chamber fitted with sampling valves Note large vapor volume
1772
INDUSTRIAL AND ENGINEERING CHEMISTRY
assumed and Raoult's law applied, the relation holds that PvOFa sol,d NVOFs
dissolved
x
P V O F t metastsble liquid,
where .\'VOFa is the mole fraction of vanadium oxyfluoride in uranium hexafluoride a t saturation. Given a solubility of 0.7 weight yo oxyfluoride and a sublimation pressure of 131 mm.at75' C., the pressure of the metastable liquid is computed to be 6670 mm. in reasonably good agreement with the value from liquid-vapor equilibrium, in view of the assumption made and the experimental consistency. The measured sublimation pressure of vanadium oxyfluoride may be somewhat high. Vanadium Pentafluoride-Uranium Hexafluoride. For this system at 75" C., a vapor phase containing 570 =!= 17 p.p.m. of vanadium was found to exist in equilibrium with a liquid phase containing 180 rt 8 p.p.m.; the vapor enrichment factor for vanadium was 3.2. Utilizing the literature vapor pressure of 2675 mm. at 75" C. (76) an ideal enrichment factor of 1.7 is calculated. As a result of conductivity measurements, Clark and Emeltus (3) have postulated association in liquid vanadium pentafluoride : 2VFs + VFa'
+ VFs-
If such is the case, the ideal enrichment factor would be expected to be lower than that obtained experimentally. because in the dilute solutions studied, vanadium pentafluoride would be unassociated with a higher partial pressure than that calculated by application of Raoult's law to the vapor pressure of the pure associated liquid. Infrared Spectra. To investigate the identity of the various fluorides dissolved in uranium hexafluoride, their infrared spectra were observed in the 2- to 16micron region with a Baird recording spectrometer. The gases were measured in a Fluorothene cell to which sodium chloride windows were attached with Apiezon wax As some attack on the windows occurred, occasional repolishing was necessary. The infrared samples were first degassed' to remove silicon tetrafluoride and other noncondensables and then admitted to the cell via the gaseous phase; in this manner the volatile impurities were concentrated. In Figure 2 are shown the spectra of vanadium pentafluoride, vanadium oxyfluoride, and molybdenum hexafluoride. The first contains some of the second as an impurity. Figure 3 presents the curves for a pure and a typical crude uranium hexafluoride sample prepared from a uranium trioxide concentrate. That the crude's main impurity is vanadium oxyfluoride is shown by the characteristic absorption of the triplet a t 9.5 microns. Absorption a t 13.5 to 13.6 microns appears to be due to both molybdenum hexafluoride absorbing at 13.4 microns and vanadium oxyfluoride
t
NUCLEAR TECHNOLOGY
-
‘6
7
8
9
IO
II
12
13
14
40
16
WhVE LENGTH IN MICRONS
Figure 3. Infrared spectra of pure and contaminated uranium hexafluoride The main impurity of the crude is vanadium oxyfluoride
0
I
7
I
I
I
I
0 9 10 I1 WAVE LENGTH
I
I
I
12 13 14 IN MICRONS
I
15
lo
16
Figure 2. Infrared spectra of pure compounds often found in uranium hexafluoride Vanadium pentafluoride contains some vanadium oxyfluoride as impurity
absorbing a t 13.6 microns. The original crude concentrate contained O.49Yc vanadium and 0.19% molybdenum.
Distillation Vanadium Oxyfluoride-Uranium Hexafluoride Solution. In view of the low solubility of the oxyfluoride in the hexafluoride at 75’ C., and of the 3 to 1 vapor-liquid enrichment ratio found for this solution pair, a question arose as to whether the vanadium oxyfluoride in the still column would separate on the still plates as a solid, and, if so, would travel up or down the column. The hexafluoride can be handled in glass if the glass is properly dried (70). Therefore, an all-glass perforated-plate still was constructed to observe distillation of this solution pair. The still proper was thermostated to 75’ C. while water a t 68’ to 70 O C. was circulated through its reflux condenser jacket. Drying was accomplished by heating the entire apparatus to 300’ C. for an extended period under a vacuum of mm. In carrying out the distillation run, the still was charged with 284 grams of uranium hexafluoride containing 0.5 weight yc oxyfluoride. At the start of the run, the solution was clear, with a slight yellow cast. During the 21/2-hour reflux at 20 to 24 p.s.i., no solid particles were visible on the still plates or pot. Some etching of the pot showed slight hydrolysis of the hexafluoride.
During the run it became evident that crystals were forming on the wall at the top of the reflux condenser. At the end of the run, these were found to be flat, diamond-shaped plates [33y0vanadium (80% vanadium oxyfluoride) and 6% uranium (9% uranium hexafluoride) the latter probably caused by contamination on shutdown of the still]. At this time, the charge was removed by distillation and analyzed for vanadium, 45 If 5 p.p.m. being found. A 16-mm.color film was made of the run. Plate operation, liquid phase condition, and the crystals are readily seen. Distillation of Crude Uranium Hexafluoride. A small column of 10 theoretical plates was constructed of 1-inch Monel pipe, 4 feet long and packed with Podbielniak Helipak, with provision for product withdrawal from the top of the reflux condenser via vapor phase. Crude process uranium hexafluoride was distilled at a pot temperature of 66’ to 69’ C. and a pressure of 11.5 p.s.i. Successive vapor phase samples were slowly withdrawn from the condenser over a 24-hour period. Concentrations of vanadium in the initial fractions were higher than in the original material. However, conventional distillation does not remove vanadium oxyfluoride from the hexafluoride to the extent required.
Acknowledgment The authors express their appreciation to S. R. Orfeo for assistance in some
of the liquid-vapor equilibrium measurements, and thank C. F. S d i t h for excellent photographic work on the glass still.
literature Cited (1) Am. Inst. Physics, “Temperature, Its Measurement and Control in Science and Industry,” p. 257, Equation 4, Reinhold, New York, 1941. (2) Barber, E., Cady, G., J . Phys. Chem. 60, 505-6 (1956). (3) Clark, H. C., Emeltus, H. J., J. Chem. Sac. 1957, pp. 2119-32. (4) Emeltus, H. J., Gutman, V., Ibid., 1949,p. 2979. (5) Engelbrecht, A., Grosse, A,, J. Am. Chem. Sac. 74,5262-4 (1952). (6) Ephraim, F., “Inorganic Chemistry,” 4th ed., p. 356, Nordeman, New York, ? 943. (7) Gmelin, L., “Handbuch der Anorganische Chemie,” 8th ed., System 53, p. 150, Verlag Chemie Weinheim/Borgstr., n2c
.( 1jr.J.J.
e
(8) Haendler, H. M., 0th rs, J . Am. Chem. Sac. 76,2177 (1954). (9) Littv. A. H. F.. General Chemical
Division, Allied Chemical Corp., unpublished data, on work at Argonne National Laboratories. (IO) McGill, R . M., U. S. Atomic Energy Comm., Tech. Information Service, AECD-3713 (July 2, 1951). (11) Masi, J., J . Chem. Phys. 17, 755-8 (1 , 949). -. -~ ,(12) Mears, W. H., Orfeo, S. R., unpublished data. (13) Oliver. G.. Grissard. G.. J . Am. Chem. ‘ Sac. 75. 2827 i1953). (14) Ruff, O., Ascher, E., Z. anorg. u. allgem. Chem. 196, 413-20 (1931). (15) Ruff, O., Lickfett, H., Ber. 44, 2539 (1944). (16) Trevorrow, L. E., others, J. Am. Chem. Sar. 79, 5167-8 (1957). RECEIVED for review April 7, 1958 ACCEPTED September 8, 1958 Division of Industrial and Engineering Chemistry, Symposium on Preparation and Recycle of Feed Materials for Nuclear Fuel, 133rd Meeting, ACS, San Francisco, Calif., April 1958. I
VOL. 50, NO. 12
,
DECEMBER 1958
1773
