Recovery of Uranium From Wet-Process Phosphoric: J. David J. and B
Recovery of Uranium From Wet-Process Phosphoric: J. David J. and B
Phosphoric Acid
A two-cycle process for recovering uranium by extraction with di(2-ethylhexyl)phosphoric acid plus trioctyl-
phosphine oxide in an aliphatic diluent has been developed and demonstrated successfully in bench-scale
mixer-settler tests. The uranium is recovered from the solvent and concentrated by reducing it to the less
extractable tetravalent form with ferrous iron contained in reduced process raffinate. This solution is oxidized
and processed in a second cycle with the same extractant. In the second cycle the uranium is stripped from
the solvent with ammonium carbonate solution and recovered as a high-grade (>97% U3Oe) product. Overall
recovery of uranium is about 95%. In the bench-scale tests, losses of the relatively expensive solvent to the
aqueous raffinate were satisfactorily low (<0.1 ml/l.). Chemical reagent costs for the process are estimated
a t less than $ 1 .OO per Ib of recovered u308.
I n the early 1950's a solvent extraction process was developed only a small fraction of the total iron is in the reduced ($2)
and utilized for the recovery of uranium from the wet-process state, and about 50% of the uranium is present as V(1V). For
phosphoric acid produced by treatment of Florida phosphate effective extraction of uranium with the synergistic extractant,
rock (Cronan, 1959; Greek e t al., 1957; Long e t al., 1955). the uranium must be in the hexavalent form. Conversion to
Conimercial application of this process was short-lived, how- this form can be accomplished easily by adding mole of
ever, because of the discovery of large amounts of relatively NaC103 per mole of ferrous iron to the acid liquor. Tests have
low-cost uranium in the western United States and a number also indicated that the uranium can be oxidized by bubbling
of troublesome problems inherent in the process. air or oxygen through the liquor a t 60-70°C.
During the intervening years the phosphate fertilizer in- Process Flow Sheet. T h e proposed process consists of
dustry has grown rapidly, and the amount of u308 dissolved two cycles (Figure 1). I n the first cycle the oxidized acid is
in Floridiaii wet-process phosphoric acid has been estimated cooled to 40-4j°C, and the uranium is extracted with 0.551
as presently approaching 2000 tons annually (Keiinedy, 1967). D2EHPA-0.125JI' TOPO in an aliphatic diluent. Uranium is
This consideration, in conjunction with an increasingly opti- recovered from the solvent by contacting it with a phosphoric
mistic outlook for the uranium market, has greatly renewed acid solution containing ferrous iron which reduces the ura-
interest in the wet-process acid as a significant source of ura- nium to the less extractable tetravalent state and effects its
nium. transfer to the aqueous phase. T o borrow a small volume of
I n late 1967 a program was initiated a t the Oak Ridge Na- raffinate from the extraction system for this purpose is con-
tional Laboratory to develop an improved recovery process venient (and inexpensive). For best results an inert atmo-
(Ferguson, 1968, 1969, 1970; Hurst e t al., 1969). A large num- sphere should be provided in the stripping units to prevent air
ber of potential extractants were screened for their ability oxidation of the strip solution. Under the proper conditions
to extract uranium from pure phosphoric acid. The ordinary uranium strip solutions containing about 12 grams of uranium
uranium extractants such as tributyl phosphate and long- per liter (about a factor of 70 richer in uranium than the origi-
chain alkyl amines cannot be used because they do not have nal acid) are readily obtained. These solutions are highly
enough extraction power in the phosphoric acid system. The
synergistic extractant combination of di(2-ethylhexy1)phos-
phoric acid (D2EHPA) plus trioctylphosphine oxide (TOPO)
dissolved in a high-boiling aliphatic diluent showed the most Table 1. Results of Analysis of Wet-Process
promise for process use. Since that time we have developed Acid Samples
and tested, on a bench scale, a complete process flow sheet concentration, g/l.
Constituent Compony A Compony B
using this extractant. The results have been highly attractive,
and further engineering evaluations are recommended. Po4 5.0-6, O M 5.4-6. O X
Ua 0.14-0.17 0.16-0.19
Experimental Fe(I1) 0.3-0.8 0.3-0.6
Phosphoric Acid Samples. Eleven samples of wet- Fe(tota1) 7-10 6-12
V 0.1-0.3 0.1-0.3
process acid were obtained from two commercial phosphate
A1 3-6 3-4
plants (identified as Companies A and B), for testing (Table 2-4 2-4
Ca
I ) . These samples ranged from 5-6M in Hap04 and contained so4 19-31 27-33
0.14-0.19 grams of uranium per liter. Major impurities were F 21-30 26-29
iron, aluminum, calcium, sulfate, and fluoride. I n fresh acid
a I n fresh acid the uranium is only partially oxidized. The
U P ) gradually oxidizes to U(V1) as the solution is aged.
To whom correspondence should be addressed.
122 Ind. Eng. Chem. Process Der. Develop., Vol. 1 1 , No. 1 , 1972
WET PROCESS-
-
EXTRACTION , 1
.t
REDUCTIVE
( 4 StOQIS, STRIPPING
f - 40.-45.C) (2- 3 stapes,
5o.c )
t PRODUCT
SOLUTION
EDUCTION
R A FFINPLTE
( t o Fertilizer
Production1 I SOLVENT** RECYCLE
I
t
b’ SCRUBBING
( 2 - 3 stages,
25 -45OCI
. =- CARBONATE
STRl PPING
( 2 sto 0s
35-48.6)
- (NH4)r COS
SOLUTION
RECYCLE
TO PHOSPHATE
ROCK LEACHING
CI RCUlT
*0.5 M 1)2EHPA--0.125 M TOPO-Amrco 450
** 0.3 M I)2EHPA--0.075 M TOPO--Amsco 4 5 0
Figure 1. Process flow sheet for recovery of uranium from wet-process phosphoric acid
Ind. Eng. Chem. Process Des. Develop., Vol. 1 1, No. 1 , 1972 123
ORNL-DWG-71-164
D ORGANIC PHASE, 0.18 1! DZEHPA-0.05 1! T W O - N D D
eWm
+.10'-
z
-w
LI.
W
u
z
;
I-
L
loo-
f
3
D2EHPA CONCENTRATION Id-
- I I I I I tion
-
.* 7-
W
I-
- Coefficients from this liquor after dilution to 5.3M HaPOl
and 4.83f &Pod, respectively, were 4.4 and 5.8. The phos-
P 5-
phoric acid concentration of wet-process acid produced in the
Y
LA
- United States is typically in the range of 5.5-6M. Prior to
8 - shipment the acid is usually evaporated to a concentration
z of about 12M. Obviously, from the data of Figure 5, uranium
0
u
I- could not be recovered with D2EHPA-TOP0 after the con-
d
I- 2- centration step.
X
W
Figure 6 shows the extraction of uranium with 0.5M D2-
3 EHPA-0.125M TOPO from Company A and B acids a t about
f
a 4OoC. The data for the two experiments were essentially
K
3 equivalent and hence are plotted as a single isotherm. &IC-
I-
- Cabe-Thiele diagrams indicate that about 96% of the ura-
r
IO
I
20
I
30
I
40
1
50
I
60
I
70
nium could be recovered in four ideal extraction stages by as-
TEMPERATURE ( "C ) suming that an aqueous/organic flow ratio of 2 was used for
operation. By the assumption of a uranium concentration of
Figure 4. Effect of temperature on uranium extraction from 0. I 7 g/l. in the feed liquor, loading of the solvent extract would
Company A acid with 0.2M D2EHPA-0.05M TOPO-NDD be about 0.33 grams of uranium per liter.
Stripping of Uranium.Uranium is effectively stripped by
contacting the solvent with 61M H3P04 solution that contains
10-20 grams of Fe(I1) per liter to reduce the uranium to the
coefficient increased by approximately a factor of 2 when the less extractable tetravalent state. Raffinate from the extrac-
temperature was decreased from 60-40OC. Without cooling, tion system makes an ideal strip solution since it is suitably
the wet-process acid feed to solvent extraction would have a concentrated in H3P04 and, after reduction with iron metal,
temperature of about 6OOC. Industrial representatives have contains a suitable concentration of Fe(I1). (Raffinate rather
stated that the acid can be cooled easily and inexpensively than the original feed liquor is preferred since the former has
to about 4OoC, but cooling below this temperature would be been cleaned up to some extent on passage through the extrac-
relatively expensive. tion system, and its use results in less accumulation of inter-
The uranium extraction coefficient decreases rapidly as the facial solids in the stripping system.) Moreover, fluoride pres-
H3P04 concentration is increased. In extractions from Com- ent in the acid catalyzes the reduction rate (Baes, 1956).
pany A acid (6M &Pod) a t 25OC with 0.2M D2EHPA-0.05M In process operation, intrastage recycle of the aqueous
TOPO, the uranium extraction coefficient was 2.2 (Figure 5). phase from the settler to the mixer is provided to give an
124 Ind. Eng. Chem. Process Des. Develop., Vol. 1 1, No. 1, 1972
0 COMPANY A ACiD;43*C Table II. Rate of Reduction Stripping of Uranium
A COMPANY B ACID;3B°C Organic : 0.5111 D2EHPA-O.125M TOPO-Amsco 450 con-
taining 0.45 g of U(V1) per liter
Aqueous: Cranium-barren webprocess acid; 6 J l H3P04, 11.4
g of Fe(I1) per liter
Procedure: Single mixer-settler; operation a t 50°C under
argon atmosphere
Residence
Flow rate, ml/min time in Uranium
Organic Aqueous mixer, min stripped, %
30 60 1 90
15 30 2 91
10 20 3 94
5 10 6 95
transfer of U(V1) to the strip solution. pure oxygen (or oxygen in air) , however, is relatively slow:
Carbonate Stripping. Uranium is effectively stripped
from D2EHPA-TOP0 by contact with ammonium carbonate
2Fez+ + 1/202+ 2 H + -+ 2Fea+ + H20 (2)
solutions. The solubility of uranium in the solution is inversely This may be accounted for by the extremely low solubility
dependent on the ammonium salt concentration of the solu- of oxygen in the solution, particularly a t temperatures above
tion. Operation of the stripping system with a relatively con- 7OoC and a t atmospheric pressure. Also, the oxidation of
centrated (2-3121) ammonium carbonate solution results in Fe(I1) to Fe(II1) with molecular oxygen in acidic solutions
direct precipitation of ammonium uranyl tricarbonate (AUT) reportedly proceeds rather slowly because the oxygen mole-
in the stripping system (Hurst and Crouse, 1960). The AUT, cule is first converted to hydrogen peroxide which is slow in
which settles rapidly in the aqueous phase and is readily fil- oxidizing the ferrous iron (Latimer, 1952).
tered, can be decomposed thermally to yield U308. The filtrate Results of batch tests made to study the effect of tempera-
is recycled for further stripping. This stripping method is used ture on the oxidation efficiency indicated that the best tem-
in the second cycle of the process flowsheet. perature for pure oxygen was about 70"C, whereas that for
air was about 90°C. I n these tests the oxygen (or air) was bub-
Results
bled through a fine frit into a liter of stirred solution which
Single-Cycle Process. I n an earlier process design (a initially contained about 10 grams of Fe(I1) and 10 grams of
single-cycle process) , the uranium was stripped from the initial U(1V) per liter. The half time for oxidation of the ferrous iron
extract (after scrubbing with water) with dilute (1M) am- was about 16 min with an oxygen flow of 25 cc/min a t 70°C
Ind. Eng. Chem. Process Des. Develop., Vol. 1 1, No. 1, 1972 125
I I I I I I 1 I I the extraction stripping system. The wet-process acid con-
tains soluble organic matter (probably humic acids), some
of which is extracted into the solvent phase. The humic acid
concentration in the solvent increased with solvent cycling
up to about 0.15 g/l. and then leveled off.
Typical stage data are shown in Figure 8. Uranium recovery
from Company A acid at 40°C was about 96% in four extrac-
tion stages with an aqueous/organic flow ratio of 2. The ex-
tract, which contained about 0.32 gram of uranium per liter,
was stripped with a small volume of extraction system raffi-
nate reduced with iron metal. The working volumes of the
extraction and stripping mixers were 170 and 300 ml, respec-
tively. No attempts were made to determine the maximum
0.1 1 I I I 1 1 1 I I 1
1 throughput capacity of these units, and higher throughputs
may have been possible.
0 20 40 60 80 100 120 140 160 180 About 95% of the uranium was stripped in two stages a t
TIME ( m i n ) 50"C, yielding a product solution that analyzed 11.8 grams
of uranium per liter. Intrastage recycle of the aqueous phase
Figure 7. Oxidation of Fe(l1) to Fe(l1l) in first-cycle product from the settlers to the mixers was provided in the stripping
solution with oxygen or air in stirred reactor
section to give an aqueous/orgauic phase ratio of about 2 in
the mixers. Although results nere satisfactory in these tests,
and about 37 min with an airflow of 125 cc/min at 90°C (Fig- the use of two stripping stages is marginal. Addition of a third
ure 7). Doubling the oxygen or airflow rates did not signifi- stage appears desirable to decrease the amount of uranium re-
cantly increase the oxidation rates which apparently were not cycled in the solvent to the extractor and to allow preparation
mass transfer limited in these tests. of even more concentrated strip product solutions.
Oxidation of the solution with sodium chlorate may be The first-cycle product solution was oxidized with oxygen
preferable. I n this case oxidation is complete within a few a t 70°C on a batch basis, and then the uranium was extracted
minutes at 50°C by use of only a small excess over the stoichio- a t 25°C with 0.3?11D2EHPA-0.075119 TOPO. More than 99%
metric requirement of '/e mol/mol of Fe(I1) plus '/a mol/mol of the uranium was extracted in three stages, producing an
of U(1V). extract that contained more than 9 grams of uranium per liter
Process Demonstrations. Continuous demonstrations of (Figure 9). I n addition to uranium the extract contained 1.8
the first-cycle process were made with wet-process acid sam- grams of phosphate and 0.13 grams of iron per liter. Three
ples from Companies A and B in a small mixer-settler test water scrub stages, operated a t an organic/aqueous ratio of
array at liquor flow rates of about 3 l./hr. The solvent was 5, removed more than 98y0 of the phosphate but little iron;
0.5M D2EHPA-0.125M TOPO-Amsco 450. No significant most of the iron precipitates in the stripping unit and con-
change in performance was detected for periods of operation taminates the AUT product, but the degree of iron contamina-
equivalent to about 40 complete cycles of the solvent through tion is low.
-C 0.04 U
- 0.007 U
EXTRACTION (4OOC)
0.10 U
0.02 U
0.19 U
0.04 U
0.32 U
0.07 U
-
. -
STRIPPING ( 5 O O C )
0.075 U
11.8 U 3.5 U
0.016 U
9 Fe (It) 13 Fe(II)
-+
t
(0.7 ml/min)
RAFFINAT E
Figure 8. Demonstration of first-cycle flow sheet
Numbers in blocks show concentration in grams per liter a t steody state
126 Ind. Eng. Chem. Process Des. Develop., Vol. 1 1 , No. 1, 1972
0.3HDZEHPA-0.075 M TOPO-AMSCO 4 5 0 ( I O m l / m i n )
I
3.0U 9.4U
2.8 PO4 1.8 PO4
).I9 Fe 0.13 Fe
9.7 u
m0.07 PO, Q04 PO4 0.03 PO4 - 0.015U 0.009U
4.IU 1.7U
-
ri
70"s 30"s
0.OClU 0.8U 3.9u c 9.5 PO4 0.44P04 0.12 PO4 h 116C0, 66C03 --I
0.02 PO, 0.07P04
0.1 Fe
AUT
SLURRY I-
t WATER
'O2 (NH,), COS
R A FF I NATE OXIDIZED ( 2 ml /min)
F I RST-CYCLE AQUEOUS (0.7m11rnin)
RECYCLE
Almost all of the uranium was removed from the extract 0.7 ppin hlo. This product would be a highly suitable feed for
with ammonium carbonate solution in the first of two strip- uranium refineries that produce UFs.
ping stages. The first styipping unit was designed to facilitate Phase Separation. Phase separation in all systems of both
continuous precipitation and discharge of the AUT precipi- cycles was satisfactory. I n the first-cycle extraction system,
tate. Intrastage recycle of the slurry from the settler through separation of the phases is rapid and clean, provided mixing is
a filter and back to the mixer was provided for continuous done in the organic continuous mode-Le., dispersion of aque-
removal of ilUT. ous droplets in the organic phase. In the laboratory mixer-
Makeup ammonia and carbon dioxide were bubbled directly settlers this !vas accomplished by suspending the miser in the
into the aqueous phase of the first-stage settler to maintain settler above the organic-aqueous interface. Separated organic
the ammonium carbonate concentration a t about 2-11. The phase in the settler circulated into the mixer through a hole in
concentration of uranium in the recycle solution $vas about the bottom of the mixer centered below the uplift turbine
4 g '1. Dilute (-~0.5M) ammonium carbonate solution was fed agitator. The dispersion passed into the settler through a hole
a t an extremely lon flobv rate to the second stripping stage; in the side of the mixer.
the function of this stage was primarily to scrub entrained Solids accumulated a t the aqueous-organic interface but did
uranium from the solvellt stream. The flow rate of this solu- not interfere with the operation. Although a grav-black band
tion was set to balance the loss of water from the stripping of solids appeared soon after startup, the thickness of the band
system, which occurs 011 conversion of the D2EHP-4 to the reached a certain level and then increased only slowly with
hydrated ammonium salt. No precipitation occurred in the extended operation. For commercial operations some pro-
second stage. vision would be needed for the periodic removal of solids.
The AUT precipitate analyzed 12.7% "3, 44.9y0 U,and Both the reductive stripping system and the carbonate strip-
34% cos after being washed (with a small volume of lJ1 ping system were operated with aqueous continuous mixing;
",OH and Amsco 450 diluent to remove eiitraiiied solvent) this is essential for satisfactory phase separation in the latter
and then air-dried. Calcination of the air-dried precipitate system.
for 2 hr a t 600°C yielded a product that analyzed 97.5% U30e, The amount of organic phase entrained iii the aqueous raffi-
0.5% Fe, 0.06% Pod, 0.5% cos, 25 ppm Ti, 40 ppm V, and nate leaving the extractor usually ranged from 0.2-0.3 gal/
Ind. Eng. Chem. Process Des. Develop., Vol. 1 1 , No. 1 , 1972 127
thousand gal of acid. Passing the raffinate through a 0.5-hr Blake, C. A., Jr., Homer, D. E., Schmitt, J. M., U.S. At. Energy
Comm., Rept. ORNL-2259, 1959.
holdup tank reduced this loss to less than 0.10 gal/thousand Cronan, C. S., Chenz. Eng., 66 (91, 108-11 (1959).
gal of acid. Ferguson, D. E., US.At. Energy Comm., Rept. ORNL-4272,
Chemical Reagent Costs. Total reagent costs for the 1968.
Ferguson, D. E., ibid., ORNL-4422, 1969.
process, based on the continuous bench-scale tests, are con- Ferguson, D. E., ibid., ORNL-4572, 1970.
servatively estimated to be less than $1.00 per lb of Us08 Greek, B. F., Allen, 0 . W., Tynan, D. E., I n d . Eng. Chem., 49,
(Table 111). Since the major cost is that of the relatively ex- 628-38 (1957).
Hurst, F. J., Crouse, D. J., U.S.At. Energy Comm., Rept.
pensive solvent, the economic success of the process depends ORNL-2952, 1960.
on keeping solvent losses a t a tolerable level. Although solvent Hurst, F. J., Crouse, D. J., Brown, K. B., ibid., ORNL-TM-
2.522. -1969.
- -
loss by entrainment in the aqueous raffinate in tests in the I
The value of yi,the mean activity coefficient of an electrolyte in aqueous solution, varies with the ionic
strength p. If the charges on the anions and cations present are, respectively, L- and L+, then the term
l/Z,Z-
yi is readily calculated. When this term yTtl/'+'- is plotted against p for various strong electrolytes
at 25"C,a curve family forms such that, given a single experimental value of y* at a known ionic strength
for an electrolyte, its value of y+ a t any other ionic strength can b e approximated. For use, when no such
experimental point is available, a method applicable to a number of electrolytes is described for pre-
dicting T~ at an ionic strength of 2. By this empirical procedure, approximate values of y+ can b e predicted
at any concentration for many strong electrolytes at 25'C.
M u c h of t'lie experiment'al data available for Ti, the arbitrarily selected electrolytes plotted against concentrations
activity coefficients of strong electrolytes a t various concen- from 0. to 2.0m. Inspect,ion indicates that the change of yh
trations in aqueous solutions, is well summarized in the stan- with molality differs widely from one electrolyte to the next.
dard references (Harned and Owen, 1958; Robinson and The object here is to present a simple graphical method for
Stokes, 1965). The Debye-Hilckel equat'ion applies to t,hese representing the act,ivity coefficients of strong electrolytes in
data below about 0.01m, while the modified equations of aqueous solution, applicable from low concentrations to
Scatchard and others, involving empirical constants to be de- saturated solutions. To use this method, a value of y i for the
termined for each electrolyte, apply a t higher concentrations. electrolyte in question must be known a t a concentration well
Pitzer and Brewer (1961) have presented a procedure for in- above those to which t,he Debye-Hiickel equation applies.
terpolating y L values for a number of electrolytes. When such an experimental value is not available, a procedure
These activity coefficient data in general are frequently pre- for predicting yi a t a known concentration is presented. This
sented as in Figure 1, showing curves of y* a t 25OC for several procedure appears applicable to chlorides. bromides. iodides,
acetates, nitrates, chlorates, and hydroxides. In combina-
1 To whom correspondence should be addressed tion, these relat,ionshjps make it possible to predict yh values
128 Ind. Eng. Chem. Process Des. Develop., Vol. 1 1, No. 1 , 1972