<<

PLEASE DO NOT RE , MOVE FROM LIBRA '. £ I'.] RY RI8927 , 2 n 27 %

Bureau of Mines Report of Investigations/1985

Cobalt Recovery From Leach Solutions

By T. H. Jeffers and M. R. Harvey

UNITED STATES DEPARTMENT OF THE INTERIOR r I Report of Investigations 8927

Cobalt Recovery From Copper Leach Solutions

By T. H. Jeffers and M. R. Harvey

UNITED STATES DEPARTMENT OF THE INTERIOR Donald Paul Hodel, Secretary

BUREAU OF MINES Robert C. Horton, Director Library of Congress Cataloging in Publication Data:

Jeffers, T. H. (Thomas H.) Cobalt recovery from copper leach solutions.

(Report of investigations; 8927) Bibliography: p. 12. Supt. of Docs. no.: [28.23:8927. 1. Cobalt-., 2. . 3. Copper. 1. Harvey, M. R. (Malcolm R.). II. Title. III. Series: Report of investigations (United SUites. Bureau of Mines) ; 8927.

TN23. U43 ['fN799. C6] 6228 [66H'. 733] 84·600288

" I l'

CONTENTS

Abstract .•••...••••...••..•••...••..•.•..•.... !t •••••••••••••••••••••••••••••• 1 Int roduc tion •••••••••• " ...... 2 Description of the resource •••••••••••••••••••••••••••••••••••••••••••••••••• 3

Resin selection •••••••••••• I) ••••••••••••••••••••••••••••••••••••• It ••••••••••• 3 Process description •••••••••••••••••••••••••••••••••••••••••••••••••••••••••• 4 Experimental results •••••••••••••••••••••• ~ •••••••••••••••••••••••••••••••••• 5

Ion exehange ••••••••••••••• it ••••••••••••••••••••••••••••••••••••••••••••••• 5 I Res in loading •••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• 5 I Selective impurity removal ••••••••••••••••••••••••••••••••••••••••••••••• 7 , Res in elution ...... 7 ! , , and aluminum impurity removal •••••••••••••••••••••••••••• ,•••••• 7 ! Preliminary tea ts ...... 8 D2EHPA semi continuous circuit •••••••••••••••••••••••••••••••••••••••••••• 8 Cobalt- separation ••••••••••••••••••••••••••••••••••••••••••••••••••• 9 I Preliminary tes ts ...... 9 Final iron and zinc impurity removal ••••••••••••••••••••••••••••••••••••• 10 Cyanex 272 semicontinuous circuit •••••••••••••••••••••••••••••••••••••••• 10 Summary and conclua ions ...... 11 References •••••• " ...... 12

ILLUSTRATIONS

1. Copper with cobalt recovery circuit •••••••••••••••••••••• 3 2. Simplified flow diagram for cobalt and nickel recovery from copper processing solution ••••••••••••••••••••••••••••••••••••••••••••••••••••• 4 3. Typical two-stage profile •••••••••••••••••••••••••••••• 6

TABLES

1. Loading and elution results using XFS-4195 in 4-ft-high columns ••••••••• 5 2. Metal extractions and resin loadings using various flow rates and solu­ tion volumes in 4-ft-high columns ••••••••••••••••••••••••••••••••••••••• 6 3. Selective elution of iron, zinc, and aluminum impurities ••••••••••••••••• 7 4. Process streams resulting from recovery of cobalt from copper leach solutions .••••.••••••••.••.••.•.••.•••••..••.•.••.•••.••••.••••.••• "'.... 11 UNIT OF MEASURE ABBREVIATIONS USED IN THIS REPORT

ft foot mg/L milligram per liter

g gram min minute

giL gram per liter mL milliliter

gpm/ft 2 gallon per minute per mL/min milliliter per minute square foot pct percent h hour vol pct volume percent in inch

Ib pound

'I' .1 COBALT RECOVERY FROM COPPER LEACH SOLUTIONS

By T. H. JeHers 1 and M. R. Harvey 1

\ I I

ABSTRACT

Significant amounts of cobalt, a strategic and critical metal, are present in readily accessible spent copper leach solutions. For exam­ ple, recovery of cobalt at two major u.S. copper operations could pro­ duce about 1,300,000 lb Co annually, about 13 pct of domestic consump­ tion. However, techniques such as solvent extraction and precipitation have not proven cost effective in separating and recovering the cobalt from these low-grade domestic sources.

The Bureau of Mines has devised a procedure using a chelating ion­ exchange resin to extract cobalt from a pH 3.0 copper leach solution containing 30 mg/L Co. Cyclic tests in 4-ft-high by 1-in-diam columns gave an average cobalt extraction of 95 pct when 65 bed volumes of solu­ tion were processed at a flow rate of 4 gpm/ft 2 of resin area. After an impurity scrub, eluates contained, in grams per liter, 0.4 Co, 0.2 Ni, <0.001 Cu, 1.3 Fe, 2.8 Zn, and 0.008 AI. Solvent extraction procedures to remove impurities, reject nickel, and concentrate cobalt produced a cobalt solution containing 80 g/L Co, 0.05 g/L Ni, and not more than 1 mg/L Cu, Fe, Zn, or AI. Based on data from commercial electro­ winning operations, this solution appears suitable for production of metallic cobalt.

'Chemical engineer, Salt Lake City Research Center, Bureau of Mines, Salt Lake City, UT. 2

INTRODUCTION

Cobalt is a strategic and critical met­ Several processing schemes utilizing al because of its use in jet engines, solvent extraction and precipitation have tool steels, and catalysts. In many ap­ been developed for recovering cobalt from plications, substitution of other metals acidic sulfate solutions. Flett and West is not possible. Currently, the United (2) used di-2-ethylhexyl phosphoric States does not produce any primary co­ to extract cobalt from sulfate solutions, balt and depends entirely upon imported Ogata, Namihisa, and Fujii (3) used 2- supplies and a small recycling industry. ethylhexyl hydrogen 2-ethYlhexylphosphon­ In 1982, the United States imported about ate. Solvent extraction of cobalt using 90 pct of its cobalt supply, much of it these or other reagents for the copper from Africa (l).2 leach streams would not be effective be­ cause of poor selectivity and excessive Although domestic cobalt reserves are solvent losses incurred while extracting limited, significant amounts are present cobalt from very large volumes of solu­ in some copper leach solutions produced tion. Likewise, precipitation of cobalt by leaching of low-grade directly from the acidic sulfate solu­ copper . The total quantity of co­ tions (4-5) would be ineffective owing balt available in these solutions is not to selectivity problems and filtration known, but cobalt recovery from two recy­ requirements. cle streams located at major U.s. copper operations could produce about 1,300,000 The Bureau of Mines has investigated lb of cobalt annually. This would sat­ ion-exchange cobalt recovery from copper isfy 13 pct of domestic consumption leach solutions. The process consists of and help relieve the need for imported ion exchange to extract cobalt from the cobalt. high-volume copper leach streams, result­ ing in small volumes of cobalt-rich elu­ Presently, cost-effective technology is ates. These eluates are then processed not available to separate and recover using solvent extraction to remove coex­ cobalt from these low-grade solutions. tracted impurities and to further concen­ The complex solutions contain copper, trate the cobalt. A cobalt sulfate solu­ nickel, iron, zinc, aluminum, magnesium, tion suitable for cobalt manganese, and other elements, in add­ is produced. This process has signif­ ition to cobalt. Although significant icant advantages compared with processes amounts of cobalt may be recoverable, recovering cobalt from primary sources. cobalt solution concentrations are only Since the cobalt in the copper leach so­ 15 to 30 mg!L. Since economic consider­ lution has already been solubilized, the ations dictate that copper leach stream often difficult and expensive dissolution i flows of several thousand gallons per step is avoided. Also, existing support minute must be processed, pH and temper­ facilities are available, and thus the ature adjustments would not be practical. initial capital investment for site de­ Also, the addition of reagents to the velopment would be minimal. streams to enhance cobalt extraction would be costly and could affect affili­ Only very limited studies have pre­ ated leaching and copper recovery viously been conducted using ion exchang­ operations. ers for cobalt-bearing solutions (6). These ion-exchange resins were generally unacceptable for use with dilute copper 2underlined numbers in parentheses re­ leach solutions because of their poor fer to items in the list of references at selectivity for cobalt and low cobalt the end of this report. loadings. 3

DESCRIPTION OF THE RESOURCE

Solutions from two copper leaching op­ downward through the are, leaching out erations were studied; one contained 30 materials that accumulated in the solu­ mg/L Co and the other 15 mg/L Co. The tion. This leach liquor was then col­ 30-mg/L solution, used in the majority of lected in a reservoir and processed using the semicontinuous testing, was copper cementation with scrap iron to remove cementation plant effluent containing, in most of the copper. At this point, solu­ grams per liter, 0.03 Co, 0.035 Ni, 0.06 tion was removed for testing in the co­ Cu, 2.0 Fe, 0.2 Zn, 4.5 AI, 7.2 Mg, and balt recovery system. Because large 0.4 Mn. The solution pH was 3.0. amounts of solution were required in cer­ tain phases of the test program, synthet­ The copper leach solutions were produc­ ic solutions were sometimes used. How­ ed by dump leaching of low-grade ores ever, continual cross-checking showed with dilute sulfuric acid. A schematic similar test results regardless of wheth­ of this process is shown in figure 1. er actual processing solutions or syn­ The acid was allowed to slowly percolate thetic laboratory solutions were used.

RESIN SELECTION

The first objective was to identify an elution characteristics of the loaded extractant capable of removing cobalt resin, and (5) resin regeneration. Both from the copper leach solutions. Over batch tests in small flasks on an 200 commercially available ion-exchange orbital-type shaker and small column resins, activated , and chemical tests were conducted. Based on these reagents were tested; the major criteria tests, four promising ion-exchange resins considered were (1) cobalt capacity, (2) were chosen for further study: Rohm and kinetics of the extractant, (3) selectiv­ Haas IR-904, Amberlite XE-318, and Dow ity for cobalt over other ions, (4) resins XFS-4195 and XFS-43084. 3

Additional column tests were then con­ ducted in a l-in-diam by 12-in-high col­ umn, and Dow resin XFS-4195 was selected as the most promising cobalt extractant. The properties of this weakly basic che­ lating resin have been described by Jones and Grinstead (7). The cobalt loading of resin XFS-4195 was significantly greater than that of the other resins tested, and the kinetics of cobalt sorption were sat­ isfactory. A second major advantage of this resin was that solution containing only 20 g/L H2S04 eluted the cobalt, while either 50 g/L H2S04 or 2N NH40H was needed for complete nickel elution. This permitted the use of split elution tech­

i Scrap 'f---*'l niques and simplified cobalt-nickel sepa­ . iron ration later. Also, copper was eluted using 2N NH40H, which was an advantage since it provided a means of producing

3Reference to specific products does FIGURE 1. - Copper dump leaching with cobalt not imply endorsement by the Bureau of recovery circuit. Mines. 4

acidic eluates free of copper. Addition­ cycle. Although XFS-4195 was not highly ally, the ammonium readily con­ selective for cobalt over copper, nickel, verted the resin to the base form, which iron, or zinc, both magnesium and man­ gave higher cobalt loadings than the acid ganese were completely rejected by the form of the resin in the next loading resin.

PROCESS DESCRIPTION

After the selection of XFS-4195 resin, A three-column system was used for the studies were initiated to recover cobalt ion-exchange studies. In operation, two from copper solutions. The cobalt recov­ columns, the and scavenger units, ery system consisted of three major unit were loaded in series. Meanwhile, the operations: ion exchange to extract the third column, which was loaded in an ear­ cobalt, purification of the column elu­ lier cycle, was eluted. At the end of a ates using solvent extraction to remove processing cycle, the scavenger column coextracted impurities, and a second sol­ became the lead column, the eluted column vent extraction operation to separate the became the scavenger column, and the lead cobalt and nickel and produce an electro­ column was eluted. Downflow loading was lyte suitable for cobalt electrowinning. used throughout the testing period. Elu­ A simplified flow diagram of the process tion was also carried out in a downf10w is shown in figure 2. mode using various amounts of acidic and ammoniacal eluant solutions.

Copper processing Sulfuric acid solution eluont

Roffinate (Cobalt-nickel)

Iron, zinc, aluminum (Cobalt, nickel, iron, (Return to Cobalt solution zinc, aluminum) leach solution) Depleted feed Nickel (Return to solution leach solution)

SEPARATION PRIMARY ION EXCHANGE IMPURITY REMOVAL AND RECONCENTRATION FIGURE 2 •• Simpl ified flow diagram for cobalt and nickel recovery from copper processing solution.

i ,I I; 5

Acidic eluate containing cobalt, some returned to the copper leaching circuit.) nickel, iron, zinc, and aluminum was Ammoniacal eluate containing nickel and passed into a solvent extraction circuit copper was processed for recovery of a where its pH was adjusted to 2.5. Iron, nickel-copper residue. Meanwhile, the zinc, and aluminum impurities were remov­ cobalt-nickel raffinate was directed into ed with di-2-ethylhepyl phosphoric acid a second solvent extraction circuit where (D2EHPA), manufactured by Mobil Chemical the pH was adjusted to 5.0, and Cyanex Co. (In a commercial operation, these 272, manufactured by American Cyanamid impurities, along with depleted feed from Co., was used to separate the cobalt and the ion-exchange columns, would be nickel and produce a cobalt .

EXPERIMENTAL RESULTS

ION EXCHANGE coextracted. The elution results were satisfactory, especially after the first Each of the three ion-exchange columns, few cycles. Copper was initially diffi­ 1-in-diam by 4-ft-high, was filled with cult to elute, and a small amount remain­ 460 mL of resin to a depth of 3 ft. The ed on the resin after each cycle; how­ ion-exchange work was divided into three ever, the remaining residual copper distinct phases: resin loading, selec­ reached an equilibrium in the later cy­ tive impurity scrubbing, and resin cles, and a balance was achieved between elution. the copper loaded and eluted.

Resin Loading After baseline conditions were estab­ lished, several additional cycles were The three-column system was filled with run to determine the effects of flow rate resin and operated to establish baseline and volume variations in the solution conditions. Seventeen loading and elu­ processed for each cycle. Tests were tion cycles were conducted, using a flow conducted using flow rates of 3 to 5 rate of 3.0 gpm/ft 2 of resin area and 50 gpm/ft2 and solution volumes of 50 to 80 bed volumes of pH 3.0 copper leach solu­ bed volumes. Extractions and resin load­ tion in each cycle. Elution was accom­ ings are presented in table 2. The co­ plished using three bed volumes of 20 giL balt extraction was 95 pet when 65 bed H2S04' followed by four bed volumes of 2N volumes of feed were processed at 4 2 NH 40H at a flow rate of 1.0 gpm/ft of gpm/ft2. Increasing the flow rate to 5 resin area. The results for the 17 cy­ gpm/ft 2 or the solution volume to 80 bed cles are shown in table 1. The overall volumes decreased the cobalt extraction extraction represents the extraction ob­ by several percent. As expected, cobalt tained by pumping the 50 bed volumes of loading on the resin increased as the feed through both the lead and the scav­ volume of solution increased. However, enger columns. the loading at 80 bed volumes was only slightly higher than that obtained at 65 Good extraction of cobalt from the cop­ bed volumes because of the lower cobalt per leach solution was obtained although extraction at 80 bed volumes. The over­ nickel, copper, iron, and zinc were all recovery and resin loadings for

TABLE 1. - Loading and elution results using XFS-4195 in 4-ft-high columns

Description Cobalt Nickel Copper Iron Zinc Overall extraction •••••• pct •• 97 95 100 12 93 Resin loading •••••• g/L WSRI •• 1.3 2 3.6 9.4 10.7 Elution efficiency •••••• pet •• 100 100 92 100 100 IIWet settled resin. 6

TABLE 2. - Metal extractions and resin loadings using various flow rates and solution volumes in 4-ft-high columns

Flow rate, Solution gpmlft 2 processed, Cobalt Nickel Copper Iron Zinc volumes OVERALL EXTRACTION, PCT 3 50 97 95 100 12 93 4 50 94 90 100 9 85 5 50 89 88 100 10 82

4 65 95 94 100 10 90 4 80 84 85 100 9 81 RESIN LOADING, GIL WSR' 3 50 1.3 2.0 3.6 9.4 10.7 4 50 1.3 2.0 3.6 7.4 10.0 5 50 1.1 1.7 3.5 7.4 9.2

4 65 1.6 2.5 4.7 10.1 12.9 4 80 1.7 2.8 5.8 11.6 14.9 lWet settled resin. nickel, copper, iron, and zinc impurities leach solution, cobalt loading of the followed the same trends as the cobalt. fresh resin was 0.44 g Co per liter of Based on these results, a flow rate of 4 WSR or 4.8 pct higher than that of the gpm/ft 2 and a solution volume of 65 bed used resin. Similar nickel, copper, volumes per cycle were selected for fur­ iron, and zinc loading results were ther test work. noted.

A typical extraction profile using these conditions is shown in figure 3. 24 .----.-----,----,-----,----,----~ The scavenger load column was quite ef­ o U fective in removing cobalt from the ef­ Feed: 27 mg/L Co fluent of the lead load column. After 65 ~ 20 Feed rate: 4 gpm/ft2 0> bed volumes of feed were processed, the E lead column contained 1.6 g Co per liter z of wet settled resin (WSR) while the 0 16 scavenger column contained 0.2 g Co per t:c n:: liter of WSR. Attempts to saturate the I- z lead column with cobalt to attain a high­ w 12 u er loading resulted in a significantly z 0 lower overall extraction. u 8 I- - Z During the first 150 loading and elu­ W ::J tion cycles, no change in cobalt or im­ -1 l.L 4 purity loadings on the resin had been l.L noted in the day-to-day operations. In W order to confirm these results, a small o ~~~~ __L- ___L __ ~ ____~ __~ sample of used XFS-4195 resin was with­ 10 20 30 40 50 60 70 drawn from one of the columns and compar­ EFFLUENT, bed volumes ed with a fresh resin sample. The cobalt loading of the used resin was 0.42 giL FIGURE 3. - Typical two-stage cobalt extrac­ WSR. When contacted with pH 3.0 copper tion profi Ie. 7

Selective Impurity Removal Several elution procedures were tested using variations in reagent concentra­ Before each loaded column was eluted, tions and flow rates. The above proce­ it was washed with two bed volumes of a dure proved to be the most effective dilute sulfuric acid solution to selec­ since a copper-barren eluate product was tively remove iron, zinc, and aluminum obtained that had a cobalt-nickel ratio impurities. Tests were conducted to de­ of 2:1. Eluant acid concentrations termine the most effective wash procedure greater than 20 giL H2S0 4 decreased the by passing solutions of various acid cobalt-nickel ratio because of increased strengths through loaded resin columns at nickel elution, while less concentrated a flow rate of 1 gpm/ft2. Results are eluants resulted in only partial cobalt shown in table 3. The best iron, zinc, elution. Flows greater than 1 gpm/ft 2 of and aluminum removal without excessive resin area resulted in incomplete resin cobalt loss was achieved with a 9-g/L elution. No advantage in elution effici­ H:zS0 4 wash. Only 3 pct of the cobalt re­ ency was gained with flow rates less than ported to the wash solution while 39 pct 1 gpmlft 2. of the iron, 19 pct of the zinc, , and 90 pct of the aluminum were eluted from the The 10-pct bleed stream used to control resin. The wash solutions' containing 1, impurity levels in the recycled ammoniac­ 5, or 13 giL H2S04 were not as effective al eluate was processed for nickel and because of lower impurity removals or copper recovery. The stream was boiled greater cobalt loss. to drive off the and the residue dried to yield mixed nickel and copper TABLE 3. - Selective elution of iron, . When the bleed stream con­ zinc, and aluminum impurities, tained 2.9 giL Ni and 8.1 giL Cu, the re­ percent sulting dried residue contained 11.2 pct Ni and 29.7 pct Cu. This material could H 2S0 4 be either disposed of as solid waste or conc, Cobalt Nickel Iron Zinc Aluminum sold if a favorable market existed. giL 1 ••••• 0 0 4 1 8 IRON, ZINC, AND ALUMINUM 5 ••••• 1 1 19 13 91 IMPURITY REMOVAL 9 ••••• 3 1 39 19 90 13 •••• 14 2 56 29 88 Although excellent cobalt recovery was obtained from the copper processing solu­ Resin Elution tion, the resulting column eluates con­ tained considerable amounts of nickel, After the impurity wash, the columns iron, zinc, and aluminum impurities. A were eluted using three bed volumes of 20 literature search and several series of giL H2S0 4 followed by four bed volumes of tests did not identify a procedure for 2N NH40H. A typical acid eluate contain­ selectively recovering cobalt from the ed, in grams per liter, 0.4 Co, 0.2 Ni, eluates; thus, the research was directed <0.001 Cu, 1.3 Fe, 2.8 Zn, and 0.008 AI. toward removing the impurities. Since This eluate, which contained essentially the volume of eluate was considerably all of the cobalt, was saved as the prod­ less than that of the copper leach solu­ uct and sent to the impurity removal cir­ tion feed stream, precipitation and sol­ cuit for further processing. The ammoni­ vent extraction as well as ion exchange acal eluate was recycled, and a 10-pct were evaluated for the impurity removal bleed stream was used to control nickel step. The column eluates used in the im­ and copper levels. After equilibrium was purity removal studies contained, in established, the ammoniacal eluate con­ grams per liter, 0.4 Co, 0.2 Ni, <0.001 tained, in grams per liter, <0.001 Co, Cu, 1.3 Fe, 2.8 Zn, and 0.008 AI. 2.9 Ni, 8.1 Cu, 0.01 Zn, and 0.01 Al. 8

Preliminary Tests mixing times or increased aqueous­ to-organic (A:O) flow ratios resulted in Selective precipitation of nickel, lower impurity removals and an increase iron, zinc, and aluminum impurities was in emulsions. attempted using sodium , sodium hydroxide, sodium , and ca1ci.um The first extraction stage removed hydroxide. Addition of these reagents to about 60 pct of the iron and 10 pct of attain pH's of 2 to 5 removed many of the the zinc. The raffinate from this stage impurities, but none of the reagents were was then increased to pH 3.5 using 8N completely selective and some of the co­ NaOH, and since much of the iron had been balt was also precipitated. Also, the removed, only small amounts of precipi­ iron precipitates were often difficult to tates were formed. The equilibrium pH filter. Next, several ion-exchange res­ was maintained at 3.5 in the second ex­ ins and solvent extract:l.on reagents were traction stage using direct caustic addi­ tested under a variety of conditions to tion, and excellent iron, zinc, and alum­ remove impurities from column eluate. inum extractions were obtained while After considerable testing, solvent ex­ little cobalt was coextracted. Raffin­ traction with D2EHPA was identified as a ates from the second extraction stage suitable method. typically contained 0.001 to 0.003 giL Fe and Zn and <0.001 giL Al. Cobalt and D2EHPA Semi continuous Circuit nickel extractions were less than 1 pct because of the countercurrent nature of After several series of shakeout tests the circuit. Cobalt extracted by the or­ using D2EHPA were completed and extrac­ ganic in the second extraction stage at tion and stripping isotherms were gener­ pH 3.5 was subsequently stripped when the ated, semicontinuous testing in a coun­ organic encountered the pH 2.5 feed in tercurrent circuit was initiated. A two­ the first extraction stage. Iron and stage extraction, two-stage stripping zinc crowding also contributed to low co­ unit using D2EHPA was operated for 8-h balt extractions. periods to remove the iron, zinc, and aluminum from column eluates. This cir­ The loaded organic reagent was stripped cuit also removed traces of copper and in the two-stage stripping section using magnesium that occasionally were contain­ an A:O flow ratio of 1:1 and 15 min of ed in the eluate. A 15 vol pet D2EHPA, 5 mixing time in each stage. The stripping vol pet tributyl phosphate (TBP), and 80 solution was 200 giL H2S04 containing vol pct kerosine mixture was used. TBP several grams per liter of iron and zinc. was added as a modifier and greatly im­ A 10-pct bleed stream controlled the iron proved phase disengagements in the ex­ and zinc levels in the strip solution. traction settlers. D2EHPA concentrations Zinc and aluminum were completely strip­ greater than 15 to 20 vol pet gave poor ped from the loaded organic, but iron phase disengagements, even in the pres­ stripping was incomplete. For example, ence of TBP. aft~r 10 cycles, equilibrium had been reached; the loaded D2EHPA contained 2.5 Precise pH control was the key to suc­ giL Zn and 7.8 giL Fe. All of the zinc cessful processing of the column eluates. was stripped in each cycle, while only The pH of the eluates coming off the col­ 1.3 giL Fe was stripped. This pattern umns was generally 1.6 to 1.8, which was was repeated in subsequent cycles, and adjusted to 2.5 using either lime or so­ thus only 17 pct of the iron contained by dium hydroxide. Adjustment of the pH to the D2EHPA was stripped by the 200-g/L values above 2.5 caused iron precipita­ H2S0 4 solution. However, the partially tion and heavy emulsions in the first ex­ stripped D2EHPA still had sufficient cap­ traction stage. The flow rates of both acity to allow good impurity extractions the eluate feed and organic were 4 from the incoming eluate feed. Iron' mL/min, and the retention time in each stripping efficiencies were poor even extraction mixer was 15 min. Shorter with acid solutions containing up to 350 9

or 400 giL H2S0 4 , Several synergistic only 8 min. The 8-pct NaOH solution used combinations of D2EHPA and trioctyl phos­ in the conversion stage contained 3 pct phine oxide (TOPO) were tested to facili­ mannitol, which chelated the iron and tate iron stripping, but stripping effic­ prevented iron hydroxide precipitation. iencies were still poor. This solution was neutralized with the acid strip bleed stream and returned to One advantage of the high acid require­ the copper leaching circuit. In summary, ment for iron stripping was the oppor­ both the acid and sodium forms of D2EHPA tunity to selectively strip and recover appear to be viable systems for impurity zinc from the loaded D2EHPA. The loaded removal. organic was contacted with 50 giL H2S0 4 , and nearly 100 pct of the zinc was strip­ COBALT-NICKEL SEPARATION ped but essentially none of the iron. Acid sulfate solutions having zinc-iron After removal of iron, zirtc, and alum­ ratios in excess of 500:1 were produced inum impurities from the column eluates, in this manner. These solutions contain­ the raffinates contained, in grams per ed 30 to 40 giL Zn and only 0.03 to 0.06 liter, 0.4 Co, 0.2 Ni, <0.001 Cu, 0.002 giL Fe and may be suitable for zinc re­ Fe, 0.002 Zn, and <0.001 Al. At this covery using electrowinning or precipita­ point, tests were conducted to determine tion techniques. if a cobalt or mixed cobalt-nickel pre­ cipitate could be prepared. Sodium car­ Although excellent iron, zinc, and alu­ bonate and in combina­ minum extractions were obtained from the tion with sodium hypochlorite were added column eluates using the above circuit, to the cobalt-nickel raffinates until process upsets occurred at times because pH's in the range of 7 to 9 were achiev­ of difficulties with the pH control. ed. The resulting precipitates were fil­ Precipitates occasionally formed in the tered, dried, and assayed. second extraction stage; these precipi­ tates choked the pH electrode and caused A cobalt product containing 19.4 pct Co false readings. To avoid these problems, and 15.1 pct Ni was obtained when sodium a second D2EHPA system was used with a carbonate and sodium hypochlorite were sodium form of D2EHPA, as described by used to reach a final pH of 8.5; however, Cook and Szmokaluk (8). This system, de­ cobalt recovery was only 82 pct. The veloped by the Pyrites Co., Inc., did not best mixed cobalt-nickel precipitate was need in-stage caustic addition for pH obtained using sodium carbonate addition control since sodium rather than hydrogen to a pH of 8.5. This product contained ions were exchanged for extracted iron 19.8 pct Co and 12.4 pct Nt. The respec­ and zinc. Our laboratory unit sequen­ tive cobalt and nickel recoveries were 72 tially consisted of two extraction and 88 pct. Although these tests demon­ stages, two stripping stages, one sodium strated that intermediate products could conversion stage using 8 pct NaOH, and be prepared, the low grades and recover­ two wash stages using 3 pct NaCI to scrub ies made this route unattractive. Re­ excess sodium hydroxide from the D2EHPA. search was therefore directed toward re­ As with the acidic D2EHPA system, essen­ covery of a higher grade cobalt product. tially complete iron, zinc, and aluminum extractions were realized from the pH 2.5 Preliminary Tests column eluates. The raffinates from this system contained only 0.001 giL Fe, 0.003 Separation of the cobalt and nickel and giL Zn, and <0.001 giL AI. Over 99 pct further concentration of the cobalt was of the iron remaining on the stripped necessary before a high-grade cobalt D2EHPA was removed from the organic in product could be attained. Several ion­ the sodium conversion stage. This was an exchange resins and solvent extraction advantage since excellent impurity remov­ reagents were tested for this purpose; als were obtained at A:O flow ratios of only Cyanex 272 showed promise. Batch up to 3.7 and mixer retention times of shakeout tests with this organophosphorus 10

reagent indicated that excellent selec­ Cyanex 272 Semicontinuous Circuit tivity for cobalt over nickel could be achieved at pH's of 5 to 5.6. The test After the final impurity removal, a results also showed that iron and zinc semicontinuous countercurrent system con­ were readily coextracted over a wide pH sisting of two extraction and two strip­ range, demonstrating the importance of ping stages was used to recover the co­ removing all impurities from the column balt. The organic mixture contained 16 eluates prior to cobalt extraction. Pre­ vol pct Cyanex 272, 10 vol pct nonylphe­ liminary stripping tests were also con­ nol, and 74 vol pct kerosine. Raffinate ducted, and test results showed that co­ from the impurity removal circuit was ad­ balt could be completely stripped from justed to pH 5.0 using sodium hydroxide, loaded Cyanex 272 using a solution con­ and an A:O flow ratio of 2:1 was used. taining 75 gIL Co and only 3 gIL H2S0 4" The mixing time in each extraction stage This was important since low acid and was 10 min. The equilibrium pH of the high cobalt concentrations are critical second extraction stage was maintained at for effective cobalt electrowinning (9). 5.6 by direct caustic addition to the These results indicated that Cyanex sol­ mixer vessel. Under these conditions, 99 vent extraction and electrowinning could pct of the cobalt was extracted by Cyanex likely be interfaced. 272 while the nickel extraction was less than 1 pct. The resulting raffinate con­ Final Iron and Zinc Impurity Removal tained, in grams per liter, 0.004 Co and 0.2 Ni. In laboratory testing, this raf­ Although the raffinate from the D2EHPA finate was discarded, but in an operating impurity removal system contained only 2 plant it could be either returned to the mglL each of iron and zinc, further puri­ copper leach circuit or processed for fication was necessary to remove the last nickel recovery using sodium carbonate traces of these contaminants. Prelimin­ precipitation. When the second stage pH ary tests showed that Cyanex 272 readily was maintained above 5.6, increasing extracted iron and zinc at pH 3.7 prefer­ amounts of nickel were coextracted with entially to cobalt. A one-stage extrac­ the cobalt, while lower pH's decreased tion, one-stage stripping unit was there­ the cobalt extraction. Likewise, A:O fore operated using 16 vol pct Cyanex 272 flow ratios greater than 2:1 or mixing containing 10 vol pct nonylphenol as a times less than 10 min also decreased co­ modifier. Kerosine was used as the dilu­ balt extraction. ent. The A:O flow ratio was 1:1, and the mixing time in the extraction stage Loaded Cyanex 272 containing about 0.7 was 15 min. Under these conditions, the gIL Co was stripped with a cobalt-rich iron and zinc levels were both reduced to solution containing 50 gIL Co and 3 gIL less than 1 mg/L while only 4 pct of the H2S0 4, Essentially all of the cobalt was cobalt was extracted. The loaded Cyanex stripped from the organic reagent, using reagent was stripped using a 20-g/L H2S04 an A:O flow ratio of 1:1 and 15 min of solution at an A:O flow of 1:1 and 15 min mixing time. After several cycles, the of mixing time. A 10-pct bleed stream of resulting pH 2.5 enriched cobalt solution the circulating strip solution controlled contained, in grams per liter, 80 Co, the cobalt, iron, and zinc levels. This 0.05 Ni, <0.001 Cu, <0.001 Fe, 0.001 Zn, bleed stream was recycled to the D2EHPA and <0.001 AI. Based on data from comr system and resulted in an internal cobalt mercial electrowinning operations, this recycle of 4 pct; thus, the raffinate enriched cobalt solution appears suitable directed to the cobalt-nickel separation for cobalt electrowinning (lQ). A sum­ circuit contained 0.4 gIL Co, 0.2 gIL Ni, mary of the various process streams and a and less than 0.001 gIL each of iron, history of the cobalt concentration in zinc, copper, and aluminum. each stream are shown in table 4. 11

TABLE 4. - Process streams resulting from recovery of cobalt from copper leach solutions, grams per liter

Copper Raffinate Enriched cObalt leach Column from solution from solution eluate impurity Cyanex 272 removal circuit Cobalt ••••• 0.03 0.4 0.4 80 Nickel ••••• .035 .2 .2 .05 Copper ••••• .06 <.001 <.001 <.001 Iron ••••••• 2 1.3 <.001 <.001 Zinc ••••••• .2 2.8 <.001 .0.1 Aluminum ••• 4.5 .008 <.001 <.001

Because the amount of cobalt-enriched cobalt deposits were too small to allow electrolyte produced was quite limited, determination of the plating characteris­ only very small-scale cobalt electrowin­ tics. However, current efficiencies of ning tests were conducted during this in­ about 90 pct were obtained, and the vestigation. Metallic cobalt was nickel, copper, iron, and zinc contents e1ectrowon from the Cyanex 272 strip sol­ of the electrowon cobalt were typically utions using lead and stainless <0.1 pct. steel cathode blanks, but the resulting

SUMMARY AND CONCLUSIONS

• Dow resin XFS-4195 effectively ex­ as the product for further processing. tracted cobalt from complex copper recy­ The ammoniacal eluate was recycled and cling solutions containing only 30 mglL contained, in grams per liter, <0.001 Co, Co. Cyclic ion-exchange tests in a 2.9 Ni, 8.1 Cu, 0.01 Zn, and 0.01 Al. three-column system gave cobalt extrac­ tions of 95 pct when 65 bed volumes of • Solvent extraction using D2EHPA and solution were processed at a flow rate of Cyanex 272 effectively removed iron, 4 gpmlft 2. zinc, and aluminum impurities from the column eluates, resulting in raffinates • A dilute sulfuric acid solution con­ containing no more than 1 mglL of copper, taining 9 giL H2S0 4 scrubbed 39 pct of iron, zinc, or aluminum. the iron, 19 pct of the zinc, and 90 pct of the aluminum from the loaded resin • Concentrated cobalt sulfate solu­ prior to elution. tions containing, in grams per liter, 80 Co, 0.05 Ni, <0.001 Cu, <0.001 Fe, 0.001 • A two-stage elution procedure using Zn, and <0.001 Al were obtained by treat­ 20 giL H2S0 4 followed by 2N NH 40H was ing raffinates from the impurity removal used. The first stage produced an acidic circuit with Cyanex 272 at a pH of 5.0 to eluate containing, in grams per liter, 5.6. The enriched cobalt solutions ap­ 0.4 Co, 0.2 Ni, <0.00 1 Cu, 1.3 Fe, 2.8 pear suitable for cobalt electrowinning. Zn, and 0.008 Al. This eluate was saved 12

REFERENCES

1. Kirk, W. Cobalt. Ch. in BuMines (5-acetyl-4-hydroxy-l,3-phenylene) -1,2- Minerals Yearbook 1982, v. 1, pp. 249- ethanediyl With Bivalent Ions. J. Indian 257. Chern. Soc., v. 58, No.5, 1981, pp. 491- 493. 2. Flett, D. S., and D. W. West. Im­ proved Solvent Extraction Process for 7. Jones, K. C., and R. R. Grinstead. Cobalt-Nickel Separation in Sulfate So­ Properties and Hydrometallurgical Appli­ lution by Use of Di-(2-ethylhexyl) Phos­ cations of Two New Chelating Ion Exchange phoric Acid. Paper in Complex Metal­ Resins. Chem. and Ind. (London), Aug. 6, lurgy '78, (Int. Symp.). Inst. Min. and 1977, pp. 637-641. Metall., London, 1978, pp. 49-57. 8. Cook, L. F., and W. W. Szmokaluk. 3. Ogata, T., S. Namihisa, and T. of Cobalt and Nickel Sulphate Fujii. Separation of Cobalt and Nick­ Solutions by Solvent Extraction Using Di­ el in an Aqueous Solution. Jpn. Pat. (2-ethylhexyl) Phosphoric Acid. Paper in 80 18,547, Feb. 8, 1980. Solvent Extraction (Proc. Int. Solvent Extr. Conf). Soc. Chern. and Ind., Lon­ 4. Huggins, D. A., and N. C. Nissen. don, v. 1,1971, pp. 451-462. Precipitation of the of Nickel, Cobalt and/or Zinc Filterable From Acid 9. MussIer, R. E., and R. E. Siemens. Solutions. Fr. Pat. 2,198,897, Apr. 5, Electrowinning Nickel and Cobalt From 1974. Domestic Processing Preliminary Laboratory-Scale Results. BuMines RI 5. Jennings, J. R. Treating Aqueous 8604, 1982, 20 pp. Solutions To Bring About a Separation of Nickel Values From Cobalt Values. Can. 10. Aird, J., R. S. Celmer, and A. V. Pat. 1,025,670, Feb. 7, 1978. May. New Cobalt Production From R. C. M. 's Chambishi Roast-Leach-Electrowin 6. Patel, M. N., and J. B. Patel. Process. Min. Mag., v. 142-143, Oct. Ion-Exchange Properties of Poly 1980, pp. 320-336.

,zU.S. GPO: 1985-505-019/20/003 INT.-BU.OF MINES,FGH.,PA. 27888