Table of Contents
- Experimental Techniques and Equipment
- Experimental Results
- Kinetics of Extraction
- Kinetics of Stripping
- Continuous Circuit
- Survey of Extractants
- SX of Ni & Cu in Laterite Ammoniacal Leach Liquors
Nickel and copper were co-extracted from ammoniacal ammonium sulfate leach liquors (pH-9.5) with LIX 64N extractant in a laboratory-size mixer-settler continuous circuit. Three extraction stages resulted in >99-pct extraction of nickel, substantial copper extraction, and essentially no cobalt extraction from leach liquors. A crowd stage reduced the ammonia loading on the solvent by maximizing nickel loading. Residual ammonia was removed from the loaded solvent with a pH-controlled scrub. Nickel and copper loaded on the solvent were separated by selective stripping. Typical nickel strip circuit operation using electrolyte containing 90 grams per liter (g/l) Ni and 11.9 g/l yielded >99.9 pct Ni stripped from the solvent in five stages. The copper loading on the recycled solvent was controlled in a copper strip stage using electrolyte containing 25 g/l Cu and 160 g/l H2SO4.
The United States produces less than 10 pct of the primary nickel it consumes each year. However, the U.S. Geological Survey has estimated U.S. resources of nickel to be over 15 million short tons of contained nickel. Present domestic consumption of nickel is about 200,000 short tons per year.
A significant amount of the domestic nickel resources are nickel silicate and laterite deposits in southern Oregon and northern California which contain from 0.5 to 1.2 pct Ni, 0.01 to 0.05 pct Cu, and 0.06 to 0.25 pct Co. Genererally, commercial processes cannot efficiently and economically recover Ni, Cu, and Co from low-grade domestic laterites.
The Bureau of Mines, as part of its program to provide technology that can help ensure an adequate supply of minerals and metals to meet national economic and strategic needs, is developing a method to recover Ni, Cu, and Co metals from low-grade domestic laterites. This report describes the separation of nickel and copper from cobalt through solvent extraction from liquor produced by the reduction roast and leaching unit operations of this method. The coextracted nickel and copper are subsequently separated by selective stripping. Electrolytes of both nickel and copper are produced by the stripping circuits that are suitable for the recovery of both metals by electrowinning techniques. The method being developed is shown diagrammatically in figure 1.
The laterite ore is subjected to selective reduction at about 525° C with carbon monoxide. This reduces the nickel and iron to ferronickel, and the copper and cobalt to the metallic state. The reduced material is next leached with a solution containing 100 g/l NH4OH and 300 g/l (NH4)2SO4 (at pH-10.5) in a controlled-oxidation leach. This solubilizes the Ni, Cu, and Co as ammine complexes. In general, the nickel ammine formed is Ni(NH3)x (II), where X ranges from 2 to 6. Similar complexes occur for copper and cobalt, with the exception that cobalt is primarily Co (III).
Two different leaching approaches have been evaluated. In one approach, the leach solution is contacted with reduced ore at about 15 pct solids in a single-pass leach, resulting in from 1 to 2 g/l Ni in solution. In the other approach, the leach solution is recycled through a total of four leaching contacts. This produces a leach solution with from 5 to 6 g/l Ni. Thus, the metal recovery method had to include testing for the recovery of nickel and copper from liquors of two different concentrations. Typical leach liquors are shown in table 1.
Solvent extraction coupled with electrowinning is an attractive approach for the extraction, separation, and metal recovery because it generally offers (1) leach reagent regeneration, (2) continuous processing, (3) pure metal recovery, and (4) minimum pollution potential. Possible disadvantages to this type of approach for metal recovery are (1) relatively high capital cost and (2) high electrical consumption.
General Mills Chemical, Inc. had applied LIX 64N to the separation of nickel and copper from ammoniacal ammonium carbonate solutions. These studies showed that LIX 64N would not extract Co(III), thus affording a separation route for nickel and copper from cobalt. Kennecott’s F.I.X. process also proposed the separation and recovery of nickel and copper from ammoniacal ammonium carbonate leach liquor with LIX 64N. The S.E.C. Corp. had commercially applied LIX 64N to the separation of nickel and copper from a copper tankhouse bleed stream (pH adjusted 8.5 to 9.0).
Ritchey, Ashbrook, and Joe studied the separation of Co, Cu, and Ni from sulfate solution by solvent extraction. These studies dealt with solutions having pH ranges of either 2 to 6 or 11 to 12. Either di- (2 ethyl- hexyl) phosphoric acid (DEHPA) or a mixture of DEHPA and LIX 63 was used as the extractant in these studies. The work at high alkalinity had shown that not greater than 40 g/l (NH4)2SO4 could be tolerated. Later work by Ritcey and Lucas showed that copper could be extracted from a 300-g/l (NH4)2SO4 solution between pH 6.5 and 8.0 using LIX 63; nickel and cobalt remained in the raffinate. Although the referenced experimental work with LIX 64N was made with carbonate solutions, with low-sulfate solutions, and/or at different pH ranges than required in the method being developed by the Bureau of Mines, it was decided to start the investigation with this extractant.
Preliminary tests using LIX 64N to extract nickel and copper from the unique leach liquor produced by the method being developed (300 g/l (NH4)2SO4 and high pH) indicated encouraging results. Therefore, an investigation was pursued to develop a solvent extraction system that would efficiently extract, separate, and recover nickel and copper.
Experimental Techniques and Equipment
The laboratory evaluation of LIX 64N for the separation and recovery of nickel and copper began with a series of tests to determine the maximum loading capacity of nickel and copper on the solvent. Equilibrium isotherms for extraction and stripping were then constructed. These were used to predict the number of stages necessary to obtain a desired efficiency or recovery. The kinetics of extraction and stripping were also investigated. A laboratory- size mixer-settler continuous circuit was used to determine the actual behavior of the system under various conditions.
The maximum loading capacity of metal ions on the solvent was obtained by a “repeating contact” technique. Organic-to-aqueous volume ratios (O/A) of either one or one half were used. To start with, fresh aqueous solution and solvent were contacted by vigorous shaking in a separatory funnel to equilibrium, on a wrist-action shaker. Shake-out tests had determined that 15 minutes of shaking was adequate to insure equilibrium was reached. After phase separation, the aqueous phase was removed and replaced with a fresh aliquot of aqueous solution. The procedure was repeated until analyses of raffinate samples showed that extraction of the appropriate metal ions had ceased. The solvent was then analyzed for the appropriate metal, and this provided the value for the maximum loading capacity under the conditions used in that test. The ammonia loading on the solvent was determined by stripping an aliquot of the appropriate solvent with a 50-g/l solution (O/A = 1). The resultant strip solution was then analyzed for ammonium sulfate. The equivalent ammonia loading on the solvent was then calculated.
Equilibrium isotherms for extraction and stripping were determined by contacting the appropriate aqueous and solvent solutions to equilibrium at a range of O/A ratios, generally 10:1, 5:1, 2:1, 3:2, 1:1, 1:2, 1:5, and 1:10. At each ratio fresh aliquots of the feed solutions were used. The contacting was again achieved by vigorous shaking in a separatory funnel, on a wrist- action shaker. The feed solutions were heated to -40° C in a constant- temperature bath before contact. Adequate time was allowed to insure equilibrium had been reached; for extraction, 15 minutes was used for the intermediate ratios, and 30 minutes for the extreme ratios. Contact times used for the stripping isotherms were 30 minutes at O/A ratios of 1:2, 1:5, and 1:10. At the other ratios 1-hour total contact times were used with reheating after 30 minutes. An operating temperature of 40° C is expected for the nickel-copper solvent extraction circuit when it is tied in with the liquor feed and electrowinning streams of a continuous process. Unless otherwise stated in this report, all tests were made at about 40° C.
In the extraction isotherms both phases were analyzed for nickel after the phases were equilibrated. However, it was found that nickel analyses of the resultant strip solutions were not sensitive enough in the range of this work (85 to 105 g/l Ni) to use in determining the equilibrium curves. Therefore, the strip solutions were analyzed for sulfuric acid, with the change in acid concentration being related to the change in the nickel content. The organic phases were analyzed for nickel as in the extraction isotherms determinations.
Tests were made to define the kinetics of the extraction and stripping of nickel. “Kinetics” in this context refers to the rate of mass transfer under specified conditions, and where appropriate was expressed as the percentage approach to equilibrium in a given time. An initial test (stripping of nickel from loaded solvent at 20° C) was conducted in the mixer compartment of a 180-ml mixer-settler unit. It then became apparent that this study should include testing at elevated temperatures. Therefore, all further testing was conducted in a 600-ml (baffled) beaker that was suspended in a constant-temperature bath. The beaker was fitted with a 3.2—cm-diameter turbine impeller. The tests were begun by agitating (1,700 rev/min; tip speed 169 m/min, or 556 ft/min) a measured quantity of solvent in the mixer or beaker. A measured quantity of aqueous phase was then added; simultaneously a stopwatch was started. After the chosen time interval, the agitation was stopped, and as soon as sufficient clear phase was generated (-25 ml), samples of the appropriate phase were withdrawn for analysis. The reported times do not include the 25 to 50 seconds required for phase separation before sampling took place. For a given set of conditions the process was repeated, starting with fresh solutions, with agitation for different time intervals.
A laboratory-size mixer-settler continuous circuit was used for much of the evaluation testing. The circuit generally consisted of three extraction stages, one crowd stage, two pH-controlled scrub stages, five nickel strip stages, and one copper strip stage. In each mixer-settler stage the mixer compartment had a volume of 620 ml and the settler compartment had an area of 94.3 cm². The mixer compartments were fitted with agitators having 3.8-cm- diameter turbine impellers and were generally operated at an average speed of about 1,600 rev/min, resulting in a tip speed of about 192 m/min (630 ft/min). The mixer-settler agitators were set for a speed to maintain steady pumping from one cell to the next and to provide complete mixing of the two phases, while avoiding such violent agitation as to cause emulsions and excessive phase disengagement times. Immersion heaters were used to maintain an average temperature of about 40° C during the runs. Solution feeds were provided and controlled by positive-displacement diaphragm pumps. Where aqueous-phase recycles were used, the turbine impeller in the mixer compartments provided the pumping action. Rotameters were used to indicate the recycle rates, and tubing pinch clamps were used to control the liquid flows. When organic phase recycle was used, diaphragm pumps controlled it.
This research is based upon the ability of LIX 64N to extract nickel and copper from ammoniacal solutions. The LIX 64N reagent can be considered to operate on a hydrogen ion cycle, forming metal-organic chelates. The reactions involved in the extraction process may be represented as
Ni(NH3)6++ + 2OH¯ + 2RH7 + 4H2O ↔ R2Ni + 6NH4OH,
and Cu(NH3)4++ + 2 OH¯ + 2RH + 2H2O ↔ R2Cu + 4NH4OH.
Effect of pH
Evaluation of the LIX 64N extractant began with the development of maximum loading capacity versus pH curves (fig. 2) for both nickel and copper. Synthetic solutions containing either 5 g/l Ni, 5 g/l Cu, or 1.2 g/l Ni were used in this testing. The solutions also contained 300 g/l (NH4)2SO4, and the pH8 was adjusted with either NH4OH or H2SO4. The solvent for this testing contained either 25 vol-pct LIX 64N with 75 vol-pct Kermac 470B9 or 12 vol-pct LIX 64N with 88 vol-pct Kermac 470B. The 25-vol-pct-LIX 64N solvent was used with the 5-g/l-Ni and 5-g/l-Cu solutions, while the 12-vol-pct-LIX 64N solvent was used with the 1.2-g/l-Ni solution.
The data in figure 2 show that in the pH range tested, copper has a higher loading capacity and a larger
useful pH loading range than does nickel on LIX 64N. Nickel loading is very sensitive to the liquor pH, with maximum loading taking place between about pH 7.5 and 9.5 with the 300-g/l-(NH4)2SO4 solutions. The loading of nickel on the solvents was nearly the same in grams per liter per volume-percent LIX 64N for the 12- and 25- vol-pct solvents. The data indicate that 0.24 and 0.06 g/l Ni per volume-percent LIX 64N was loaded at pH 9.4 and 10.5, respectively. Therefore, loading the solvent at the pH of the leach liquor as produced (pH 10.5) is impractical. However, if the pH could be lowered, this solvent extraction system might be practical.
In order to lower the pH, part of the ammonia was stripped from the leach liquor by heating and air sparging of the liquor. The ammonia removed from the leach liquor was absorbed into the final raffinate and recycled to leaching with minimal losses. Typical pregnant liquor leach contains about 85 g/l NH4OH (pH -10.5). The ammonium hydroxide concentration is lowered to between 17.5 and 25 g/l (pH 9.2 to 9.7 depending on the metal content of the liquor) by ammonia stripping. In this pH range, LIX 64N is an efficient extractant of both nickel and copper from 300-g/l-(NH4)2SO4 liquor.
Effect of Ammonium Sulfate Concentration
The effect of ammonium sulfate concentration in the leach liquor on the maximum loading capacity of nickel on LIX 64N was briefly investigated. This series was carried out at 20° C with 12-vol-pct-LIX 64N solvent. Synthetic solutions containing about 1.2 g/l Ni, either 25 or 100 g/l NH4OH, and from 150 to 300 g/l (NH4)2SO4 were used. With one exception, the data show reduced nickel loading with increasing ammonium sulfate concentrations (table 2), at constant ammonium hydroxide levels. It is interesting to note that a relatively high nickel-loading capacity (0.18 g/l Ni per vol-pct LIX 64N) was obtained at a high NH4OH concentration when the (NH4)2SO4 content of the liquor was reduced to 150 g/l.
Sulfate and Carbonate Liquor Comparisons
Hydrometallurgical commercial processes that extract nickel from laterite ores (Nicaro, Cuba; Greenvale, Australia; and Marinduque’s Philippine operation) use carbonate leach liquors. Consequently, much of the work reported regarding nickel solvent extraction has dealt with ammoniacal
ammonium carbonate solutions. These studies have shown relatively high nickel loadings on LIX 64N solvents from solutions containing appreciable ammonia concentrations (that is, high pH), whereas the
present study has seen reduced nickel loadings at moderate ammonium hydroxide levels. Therefore, a comparison was made (fig. 3) between the maximum nickel loading capacities on 12-vol—pct—LIX 64N solvent from ammonium sulfate and ammonium carbonate solutions at various pH levels. For convenience, comparisons were made on the basis of pH rather than ammonium hydroxide levels.
Synthetic solutions containing about 1.2 g/l Ni were used in these 20° C tests. The carbonate solutions
contained about 69 g/l CO2 and from 65 to 100 g/l total NH3. Merigold and coworkers applied solvent extraction techniques to liquors containing 37 to 73 g/l CO2 and 40 to 100 g/l NH3. The sulfate solutions contained about 300 g/l (NH4)2SO4 and from 0 to 100 g/l NH4OH. It is apparent from figure 3 that nickel loading from these synthetic solutions falls off at a lower pH (-9.7) in the sulfate system than in the carbonate system (pH-10.0).
Preliminary loading tests with partially loaded and fully loaded solvents had shown reduced ammonia loading as the solvents were fully loaded with nickel. Literature had described the crowding effects of copper-nickel and nickel-zinc with carbonate liquors. Therefore, a series of shake-out tests was performed to determine the preferential loading of ions on a LIX 64N solvent from ammoniacal ammonium sulfate liquor.
Leach liquor was partially stripped of NH3; then copper and zinc were added. The prepared liquor contained 1.3 g/l Ni, 0.14 g/l Co, 0.16 g/l Cu, 0.15 g/l Zn, 18.5 g/l NH4OH, and 300 g/l (NH4)2SO4 (at
pH-9.3). Copper and zinc were added to decrease the number of contacts necessary before any preferential loading on 12-vol-pct-LIX 64N solvent would be apparent. Equal volumes of solvent and
liquor were contacted to equilibrium at 20° C. After phase separation both phases were sampled; the remaining solvent phase was then contacted again with prepared leach liquor (O/A = 1). The procedure was repeated six times. This testing approximates the effects that a crowd stage would have in the solvent extraction circuit.
The data in figure 4 show that the loading preference of the 12-vol-pct- LIX 64N solvent from the prepared liquor was Cu > Ni >>Zn, NH3, with no Co (III) extracted. Because of the relatively high nickel content of the leach liquor, the nickel loading on the solvent rapidly increased to -2.4 g/l. As the copper loading on the solvent increased, a point was reached where the nickel loading started to decrease, indicating that copper was crowding nickel from the solvent. The ammonia loading on the solvent was maximum after the first contact (-0.33 g/l). It then decreased and after the fifth contact remained constant at -0.15 g/l. Zinc was only slightly extracted by the solvent; it built up to -0.07 g/l after the second contact. The zinc loading then decreased to a constant 0.02 g/l. Thus, both ammonia and zinc were crowded to some extent by high nickel and/or copper loading on the solvent. Essentially no cobalt (Co III) was loaded in any of the contacts (<0.001 g/l).
Loading of ammonia on the solvent necessitates the scrubbing of the solvent before nickel stripping. Otherwise, (NH4)2SO4 would be generated in the electrolyte, causing precipitation of the double salt, NiSO4 · (NH4)2SO4 6H2O. In the scrubbing operation, a dilute acid solution (-1 g/l H2SO4) is used to remove and neutralize the ammonia from the solvent. To minimize the loss of ammonia and therefore the subsequent generation of ammonium sulfate, it is important to minimize the loading of ammonia on the solvent. This may be done through the use of a crowd stage.
It is known that LIX 64N is capable of extracting metal ions other than nickel and copper; therefore, methods for controlling possible impurities in the leach liquor such as Mn, Mg, and Fe were considered. Manganese should be removed from the leach liquor before nickel-copper extraction; otherwise it can be coextracted by the LIX 64N solvent. Conversely, magnesium is only very slightly extracted by LIX 64N. Both manganese and magnesium are removed from the leach liquor prior to nickel-copper extraction by precipitation with NH4H2PO4 (fig. 1). In the controlled-oxidation leach, iron initially dissolves as ferrous ions; then at a higher oxidation potential, the dissolved iron is oxidized and precipitated as ferric hydroxide. Typical leach liquors contain <0.001 g/l Fe.
To further describe the extraction of nickel from typical leach liquors produced by this method, two extraction isotherms were constructed (fig. 5). Equilibrium curve A represents 25-vol-pct-LIX 64N solvent contacting liquor containing 6.3 g/l Ni, 25 g/l NH4OH (pH 9.35), and 280 g/l (NH4)2SO4. Equilibrium curve B represents contacts between 12-vol-pct-LIX 64N solvent and liquor containing 1.74 g/l Ni, 18.3 g/l NH4OH (pH 9.45), and 281 g/l (NH4)2SO4. Both equilibrium curves are basically the same shape. Low nickel concentrations (<0.01 g/l) in the raffinate will result from the use of multiple extraction stages.
Kinetics of Extraction
Extraction Versus Agitation
A series of tests was made to define the kinetics of nickel extraction with the LIX 64N solvent. The batch tests were made in the previously described apparatus and manner. Temperature and residence time were the variables investigated.
To determine the optimum agitation rate to be used in these batch kinetic tests, nickel extraction versus agitator speed was investigated first. Synthetic liquor containing about 1.2 g/l Ni, 20 g/l NH4OH, and 300 g/l (NH4)2SO4 was contacted with 12-vol-pct-LIX 64N solvent for 2 minutes at various agitation rates. Fresh solutions were used for each contact. Agitator speeds ranging from 1,000 to 2,100 rev/min (tip speeds of 100 to 209 m/min or 327 to 687 ft/min) were tested. Results (fig. 6) show that the nickel concentration reached a constant value in the raffinate with about 1,700 rev/min and higher agitator speeds. Therefore, all subsequent batch kinetic testing was made with agitator speeds of about 1,700 rev/min (tip speed of 169 m/min or 556 ft/min).
Extraction Versus Time and Temperature
The kinetics of nickel extraction from laterite liquor containing about 1.24 g/l Ni, 17.5 g/l NH4OH, and 266 g/l (NH4)2SO4 by 12-vol-pct-LIX 64N solvent is affected by temperature, as is shown in figure 7. The 20° C curve indicates that a residence time of at least 6 minutes is necessary (in batch test) to approach equilibrium. At 40° C about 4 minutes was necessary to approach equilibrium. These tests were conducted at an O/A = 1.
Tests were also run with liquor containing about 6.6 g/l Ni, 280 g/l (NH4)2SO4 and either 20 or 25 g/l
NH4OH. In these tests, 25-vol-pct-LIX 64N solvent was contacted with the liquor at O/A ratios of either 1 or 2. It appears that a residence time of about 4 minutes or longer is necessary at 40° C to approach equilibrium (fig. 8) with an O/A ratio of 2. In a continuous-flow circuit, mixer short-circuiting will probably require additional residence time so that high stage efficiencies may be realized.
Effect of Ammonium Sulfate Concentration
The results from the kinetics of nickel extraction tests show lower extraction rates than expected. In the Merigold and Sudderth report, 2 minutes’ retention time was used with success in the ammonium carbonate system. The high ammonium sulfate concentration in the ammoniacal liquor from the laterite leach could have caused the lower rates. A series of tests was conducted to investigate this hypothesis.
Synthetic liquors containing about 1.2 g/l Ni, 20 g/l NH4OH, and 50 to 300 g/l (NH4)2SO4 were contacted with 12-vol-pct-LIX 64N solvent (O/A = 1) in batch kinetic tests. Results of these tests are shown in figure 9. The results show that high ammonium sulfate concentration in the liquor (250 g/l and greater) definitely shows nickel extraction. The percentages on the 300-g/l (NH4)2SO4 curve represent the percent approach to the 10-minute raffinate nickel concentration at that point.
For comparison, synthetic ammoniacal ammonium carbonate solutions (range from 30 to 83 g/l NH3 and 23 to 92 g/l CO2) were tested in a like manner. This testing produced raffinates with essentially all the same values of nickel. Thus, the extraction rates with carbonate solutions (within this test range) were very high, and the carbonate concentration had a negligible effect on the kinetics of nickel extraction. The average value of the raffinates for the carbonate testing is also shown in figure 9 (-0.002 g/l Ni).
Effect of Residence Time in Extraction Mixer
The laboratory-size mixer-settler continuous circuit was operated in a series of runs with 25 vol-pct solvent and leach liquor (6.3 g/l Ni, 25 g/l
NH4OH, and 280 g/l (NH4)2SO4 to obtain data concerning the effect of mixer retention time on nickel extraction, on nickel loading on the solvent, and on the number of stages required. The mixer turbines were operated at an average speed of 1,600 rev/min resulting in a tip speed of -192 m/min (630 ft/min).
In the first test, a solvent flow rate of 150 ml/min and a liquor flow rate of 126 ml/min were used. These flows resulted in an average residence time of 2.2 minutes in each of the extraction-stage mixers. Recycle flows within these stages were not required during this testing. The extraction circuit profile for this test is shown in figure 10 (profile B). The raffinate contained about 0.04 g/l Ni after three stages when a residence time of 2.2 minutes was used. As shown in figure 10, the first stage is displaced from the equilibrium curve (A) . This indicates that equilibrium was not reached in that stage. The solvent was subsequently loaded to about 6.35 g/l Ni in a crowd stage.
The second test used solvent and liquor flow rates of 75 ml/min and 58.5 ml/min, respectively. This resulted in an average residence time of 4.6 minutes in each of the extraction mixers. This circuit profile is also shown in figure 10 (profile C). A raffinate containing only about 0.01 g/l Ni was produced with two stages. The third stage was unnecessary when a residence time of 4.6 minutes was used. In this test the solvent was loaded to about 6.15 g/l Ni in a crowd stage.
The third test used solvent and liquor flow rates of 49 ml/min and 39 ml/min, respectively. This resulted in about 7.0 minutes of residence time in each extraction mixer. The extraction circuit profile is also shown in figure 10 (profile D). A raffinate containing about 0.05 g/l Ni was produced with two stages (>99-pct extraction); again the third extraction stage was unnecessary. Increasing the residence time to 7.0 minutes shifted the first extraction stage closer to the equilibrium curve. The solvent was subsequently loaded to about 6.34 g/l Ni in a crowd stage.
In similar series, leach liquor (1.4 g/l Ni, 18.0 g/l NH4OH, and 284 g/l (NH4)2SO4 was contacted with 12-vol-pct-LIX 64N solvent in the continuous circuit. In the first test a solvent flow rate of 88 ml/min and a liquor flow rate of 125 ml/min were used. Organic recycle flows within the extraction stages (0.52 ml/min) were used to give an overall O/A ratio of 1.1. These flows resulted in an average residence time of 2.3 minutes in each of the extraction stage mixers. The extraction circuit profile for this test is shown in figure 11 (profile B). The raffinate from the second extraction stage still contained about 0.08 g/l Ni, so a third stage was necessary to reduce the nickel content of the raffinate still further (final raffinate was 0.01 g/l Ni). In this test the solvent was subsequently loaded to about 2.37 g/l Ni in a crowd stage.
The solvent, liquor, and organic recycle flow rates were then reduced to about 40 ml/min, 55 ml/min, and 21 ml/min, respectively. This resulted in an average residence time of 5.3 minutes in each of the extraction mixers. This circuit profile is also shown in figure 11 (profile C). These data show a closer approach to the equilibrium curve (A). The raffinate produced from
the second extraction stage contained only about 0.04 g/l Ni; thus, two extraction stages were adequate. In this test the solvent was subsequently loaded to about 2.46 g/l Ni in a crowd stage.
These continuous circuit data (figs. 10 and 11) support the findings of the batch kinetic studies in that (1) the kinetics of nickel extraction from laterite liquor (high ammonium sulfate levels) is relatively slow, and (2) an extraction mixer residence time of about 4.5 minutes or longer resulted in a close approach to equilibrium, and thus a minimum number of stages could be used.
The ability of nickel to strip from the LIX 64N solvent at relatively weak acid concentrations, while copper stripping must be accomplished at relatively high acid concentrations, serves as the separation route in this study. The stripping characteristics of nickel and copper from LIX 64N were mentioned by Merigold and Sudderth, and then further described in the S.E.C. Nickel Process and in Kennecott’s F.I.X. process.
The nickel stripping and electrowinning sections of the process must be compatible. Nickel electrowinning experiments in a nondiaphragm cell have indicated that efficient operation and good deposits can be produced under a variety of conditions. Some of the optimum conditions were (1) cell temperature (50° to 65° C), (2) feed electrolyte nickel concentration (85 to 95 g/l), and (3) feed electrolyte sulfuric acid concentration (0.5 to 0.75 g/l optimum, but up to 2 g/l with some loss of efficiency). A separate report is being prepared covering the nickel electrowinning investigation. Nickel-stripping tests were therefore conducted to produce pregnant nickel electrolyte (feed to electrowinning) with 0.5 to 2 g/l H2SO4. The desired range of nickel buildup during stripping was about 5 to 6.7 g/l Ni. Therefore, depending on the desired nickel buildup during stripping, spent nickel electrolyte solutions (feed to nickel stripping) contained about 9.6 to 12.1 g/l H2SO4.
Any nickel remaining on the solvent after the nickel-stripping stages was stripped in the copper-strip stage. For this reason an objective of this study was to strip nickel from the solvent as completely as was practicable (-0.005 g/l Ni). Eventual nickel buildup in the copper electrolyte would be controlled with a bleed stream.
A strong acid solution (-165 g/l H2SO4) was used to strip copper. The copper concentration in this electrolyte was about 25 g/l. The equilibrium copper concentration in 12- and 25-vol-pct-LIX 64N solvents with the above- mentioned electrolyte was about 0.1 and 0.2 g/l, respectively.
A typical nickel-stripping isotherm is shown in figure 12. In this case the equilibrium curve was produced from contacts between 25-vol-pct- LIX 64N solvent containing about 6.33 g/l Ni and spent nickel electrolyte containing about 89.5 g/l Ni and 12.1 g/l H2SO4. Inspection of the equilibrium curve indicates that only one to two theoretical stages are required, and that almost complete nickel stripping from the solvent is possible (residual nickel -0.005 g/l).
Effect of Residual Acid on Nickel Stripping
The nickel-stripping section of the laboratory-size mixer-settler continuous circuit (previously described) was operated in a series of runs to determine the effect of residual acid in the pregnant nickel electrolyte on nickel stripping. Residual acid is the acid left in the nickel electrolyte after contact with the nickel- loaded solvent. In this testing, both 25-vol-pct-LIX 64N solvent (containing -6.1 g/l Ni) and 12-vol-pct- LIX 64N solvent (containing -2.5 g/l Ni) were used. The solvents were contacted with spent nickel electrolytes containing about 95 g/l Ni with 12.1 g/l H2SO4 and 95 g/l Ni with 9.6 g/l H2SO4, respectively. Flow rates were set to give an average in each stage, and also to yield pregnant nickel electrolytes of either pH 2.1 (-1.9 g/l H2SO4) or pH 2.4 (-0.9 g/l H2SO4), depending upon the test.
Reducing the pH of the produced pregnant nickel electrolyte (that is, leaving more residual acid) for both the 25-vol-pct-LIX 64N and 12-vol-pct- LIX 64N solvents resulted in more complete stripping per stage (table 3). However, because of the desired low residual nickel content of the nickel-stripped solvent (-0.005 g/l) , the data for both solvents show that the pH of the pregnant nickel electrolyte (within the range of these tests) did not affect the required number of stages. The 25-vol-pct-LIX 64N solvent required five stages to reach 0.004 and 0.006 g/l Ni, respectively, when pregnant nickel electrolytes at pH 2.1 and 2.45 were produced. Similarly, the 12-vol- pct-LIX 64N solvent required four stages to reach <0.001 g/l Ni with both electrolytes.
Kinetics of Stripping
Stripping Versus Time and Temperature
A series of tests was made to define the kinetics of nickel stripping from typical loaded solvents produced by this method. These batch kinetic tests were conducted in a manner and in an apparatus similar to the previously described kinetic testing. The rate of stripping nickel from 12-vol-pct-LIX 64N solvent is rather slow (fig. 13). These tests were conducted with loaded solvent containing about 2.27 g/l Ni. Spent nickel electrolyte containing about 85 g/l Ni and 8.1 g/l H2SO4 was used. As in nickel extraction, temperature has a large effect on the rate of nickel stripping. Of the curves shown, the 40° C curve is the only one to even approach a constant value (equilibrium) at the 10-minute point.
Effect of Residence Time in Stripping Mixer
The nickel and copper stripping section of the laboratory mixer-settler continuous circuit was operated with 12-vol-pct-LIX 64N solvent containing about 2.5 g/l Ni and with spent nickel electrolyte containing about 95 g/l Ni and 9.63 g/l H2SO4. These tests were designed to measure the effect of mixer retention time on nickel stripping, and the number of stages required. The mixer turbines were operated at an average speed of 1,600 rev/min resulting in a tip speed of about 192 m/min (630 ft/min).
In the first test a solvent flow rate of 93 ml/min and spent nickel electrolyte flow of 43 ml/min were used. Aqueous recycle flows within the nickel strip stages (-41 ml/min) were used to give an overall O/A ratio of 1.1. These flows resulted in an average residence time of 3.5 minutes in each of the nickel strip stage mixers. The circuit profile for this test is shown in figure 14 (profile B). The pregnant nickel electrolyte produced in this test contained about 100 g/l Ni at pH 2.55 (-0.6 g/l H2SO4) . The first four nickel strip stages lowered the nickel loading on the solvent to about 0.02 g/l, and the final fifth stage reduced it to <0.001 g/l. Although the equilibrium curve (A) indicates that only one or two theoretical stages are necessary for nickel stripping, the low stripping rate requires that more stages be used at this residence time.
The solvent, spent nickel electrolyte, and aqueous recycle flow rates were then reduced to 46 ml/min, 21.5 ml/min, and 20 ml/min, respectively. This resulted in an average residence time of 7.1 minutes in each of the nickel strip stage mixers. This circuit profile is also shown in figure 14 (profile C). The pregnant nickel electrolyte produced in this test contained about 100 g/l Ni at pH 2.4 (0.9 g/l H2SO4). The stripped solvent from the fourth nickel strip stage contained <0.001 g/l Ni; thus the fifth strip stage was unnecessary in this test. It also should be noted that the nickel strip stages (profile C) were shifted closer to the equilibrium curve (A) with the conditions of this test.
25-vol-pct-LIX 64N Solvent
The laboratory-size mixer-settler continuous circuit was operated using 25-vol-pct-LIX 64N solvent and leach liquor that had been cycled through four leaches and thus was built up to about 5.65 g/l Ni. The ammonium hydroxide content was lowered to about 23 g/l (liquor pH -9.3) by stripping ammonia, and zinc was added (resulting in -0.6 g/l) so that its disposition in the circuit could more easily be determined. The optimum parameters from the previous tests were applied in this run.
The circuit (fig. 15) consisted of three extraction stages, one crowd stage, two pH-controlled scrub stages, five nickel strip stages, and one copper strip stage. The operating temperature was maintained at an average of -40° C, and the average mixer turbine speed was 1,600 rev/min (tip speed of 192 m/min or 630 ft/min). Flow rates were set to give an average mixer residence time of 6 to 7 minutes in each extraction stage; average for the run was -6.2 minutes. In the nickel-stripping section the flow rates resulted in -7.8 minutes of average mixer residence time for each stage. The circuit was operated in two sections, with each section being operated for about 10 hours.
The extraction through pH-controlled scrub stages was operated first, with the loaded solvent being collected in a surge tank. The nickel and copper strip stages were then operated. The circuit was operated using the 25-vol-pct-LIX 64N solvent for at least one complete cycle prior to this run, and therefore copper was at equilibrium.
Results of the circuit operation are shown in table 4. Nickel in the liquor was reduced from 5.65 g/l to only 0.006 g/l in the final raffinate (>99 pct extraction). Essentially all of the copper was extracted from the liquor, but none of the cobalt and only a small amount of the zinc; the loaded solvent contained -0.03 g/l Zn.
The crowd stage reduced the ammonia loaded on the solvent by maximizing the loading of nickel on the solvent. The loaded solvent from the crowd stage contained about 6.13 g/l Ni, 0.22 g/l Cu, 0.21 g/l NH3, and only traces of Co and Zn. Zinc crowding was not evident in this test because only a trace (0.03 g/l) was loaded onto the solvent in the extraction stages.
The ammonia concentration on the solvent was lowered to <0.03 g/l in the pH-controlled scrub stages. The feed scrub solution had a pH of about 3.5 (at 40° C) and contained about 1.0 g/l H2SO4. The resultant scrub solution was at about pH 7.8 and had picked up a small amount of nickel, which had increased from about 0.17 to 0.22 g/1.
The nickel strip circuit, using spent nickel electrolyte containing about 90 g/l Ni and 11.9 g/l H2SO4, stripped essentially all of the nickel from the loaded solvent (6.10 to 0.006 g/l) while leaving the copper loaded on the solvent (0.21 g/l). The resultant pregnant nickel electrolyte contained about 96.7 g/l Ni and 0.68 g/l H2SO4 (pH -2.45). Only traces of Cu, Co, and Zn were detected in the pregnant nickel electrolyte. The buildup of ammonium sulfate through the transfer of ammonia on the loaded solvent to the nickel electrolyte remained constant at about 0.8 g/l during this run.
The equilibrium copper level on the solvent with copper electrolyte (-24 g/l Cu and 164 g/l H2SO4) is about 0.2 g/l; therefore, no copper was stripped from the solvent in this run. A mass balance based on the nickel loading on the stripped solvent indicated that the transfer of nickel to the copper electrolyte should have been -0.002 g/l Ni rather than the indicated 0.08 g/l. The higher transfer is undoubtedly caused by the entrainment of nickel electrolyte in the advancing stripped solvent. The aqueous entrainment was calculated to be -820 ppm.
12-vol-pct-LIX 64N Solvent
The laboratory-slze mixer-settler continuous circuit was also operated with 12-vol-pct-LIX 64N solvent and single-pass leach liquor containing about 1.74 g/l Ni. The ammonium hydroxide level was lowered to about 18.5 g/l (liquor pH – 9.45), and zinc was again added to the liquor (resulting in – 0.21 g/l). The previous circuit (fig. 15) was used in this run. The operating temperature was maintained at an average of about 40° C, and the average mixer turbine speed was 1,730 rev/min (tip speed of -207 m/min or -680 ft/min) The circuit was again operated in two sections. The extraction through pH- controlled scrub stages was operated for about 7 hours. The nickel and copper strip stages were then operated for about 11 hours. Because fresh solvent was used for the 12-vol-pct test, the solvent was contacted with the copper electrolyte prior to the run to establish the equilibrium copper level in the solvent. The selected flow rates resulted in an average extraction mixer residence time of 5.3 minutes per stage and an average nickel strip mixer residence time of 7.1 minutes per stage.
Results of this run are shown in table 5. Nickel extraction was again very good (>99 pct); the raffinate contained only 0.002 g/l Ni. Reference to the extraction raffinate and loaded solvent shows that most of the copper was again extracted with the nickel, and again very little if any cobalt or zinc was extracted. Although three extraction stages were used, only two would have been necessary because the raffinate from the second stage contained only 0.03 g/l Ni.
The crowd stage reduced the ammonia loading on the solvent from about 0.15 to 0.10 g/l while increasing the loading of nickel on the solvent from about 2.06 to 2.50 g/l. The Cu, Co, and Zn concentrations on the solvent remained constant in this operation at about 0.11, 0.02, and 0.01 g/l, respectively.
The ammonia concentration on the solvent was then lowered to <0.03 g/l in the pH-controlled scrub stages. The feed scrub solution had a pH of about 3.6 (at 40° C) and contained about 0.61 g/l H2SO4. The resultant scrub solution pH was about 7.5. No Ni, Cu, Co, or Zn was stripped from the solvent in the scrubbing operation on this run.
The nickel strip circuit using spent nickel eletrolyte containing about 95 g/l Ni and 9.63 g/l H2SO4 stripped essentially all of the nickel from the loaded solvent (2.50 to <0.001 g/l) while leaving the copper loaded on the solvent (-0.10 g/l). The resultant pregnant nickel electrolyte contained about 100.4 g/l Ni and 0.91 g/l H2SO4 (pH – 2.4). Reference to the spent and pregnant nickel electrolytes shows that no buildup of Cu, Co, Zn, or (NH4)2SO4 was experienced in this run. Although five nickel strip stages were used, only four were necessary because the stripped solvent from the fourth stage contained <0.001 g/l Ni.
Loading of copper on the solvent was again too low to see any appreciable change in the copper concentration of either the solvent or the copper electrolyte. A mass balance based on the nickel loading on the solvent indicated that transfer of nickel to the copper electrolyte should have been <0.001 g/l Ni rather than the indicated 0.07 g/l. As mentioned before, the higher transfer is undoubtedly caused by entrained nickel electrolyte in the advancing stripped solvent (calculated to be -655 ppm).
Survey of Extractants
The evaluation of LIX 64N for the extraction of nickel and copper from liquor produced by this method has indicated that the extractant has two undesirable features: (1) In the concentrated ammonium sulfate liquors, the nickel-loading capacity on the solvent rapidly decreases with increasing pH (i.e., increasing ammonium hydroxide concentration), and (2) some ammonia is dissolved in and probably extracted by the solvent.
These features necessitate the following respective corrective steps:
- Partial ammonia stripping from the leach liquor prior to nickel-copper solvent extraction is necessary to more fully utilize the extractant, and
- the loaded solvent must be scrubbed to remove ammonia prior to nickel stripping. Otherwise, the double salt NiSO4·(NH4)2SO4·6H2O would be generated and precipitate in the nickel electrolyte. Since several similar extractants were developed or recommended during this investigation, a series of preliminary tests was performed to determine if a more favorable extractant was available.
Maximum-nickel-loading-capacity tests were run on the following extractants: Acorga P5100, XI 54, LIX 34, LIX 70, LIX 605, and LIX 64N (for comparison). Leach liquor containing about 1.2 g/l Ni, 0.02 g/l Cu, 0.14 g/l Co, 266 g/l (NH4)2SO4, and 17.5 or 100 g/l NH4OH was used in these 20° C tests. The solvents were obtained by mixing 12 vol-pct of the extractants (as they came from the manufacturers) and 88 vol-pct Kermac 470B. The solvents were fully loaded with nickel at the two ammonium hydroxide concentrations, and both nickel and ammonia loadings were determined. The loaded solvents were stripped with 50 g/l H2SO4 (O/A = 1).
Results of the tests are shown in table 6. XI 54 did not load nickel at either of the ammonium hydroxide concentrations tested. LIX 70 loaded nickel well at the low ammonium hydroxide level and had a very low ammonia loading. Unfortunately, at the higher ammonium hydroxide level, the LIX 70 solvent formed an emulsion; also, it was difficult to strip nickel from this solvent. Acorga P5100, LIX 34, and LIX 605 all had relatively high nickel loading capacities at both of the ammonium hydroxide levels tested. It is important to note that in these tests none of the extractants loaded cobalt (Co III).
Acorga P5100 had a high nickel-loading capacity (2.85 g/l) from the 100- g/l-NH4OH liquor; thus ammonia stripping from the liquor would not be necessary. It also appears that nickel will strip easily from this extractant. However, two apparent disadvantages of this extractant were (1) very high ammonia loading (with the 100-g/l-NH4OH liquor, 2.55 g/l NH3 was loaded), and (2) high entrainment and phase separation problems during stripping. The latter problem might be solved by using a phase modifier in the solvent.
LIX 34 had a high nickel-loading capacity (1.65 g/l) from the 100-g/l- NH4OH liquor. It appears that nickel will also strip easily from this extractant. Of the extractants tested that loaded an appreciable amount of nickel from the 100-g/l-NH4OH liquors, LIX 34 had the lowest ammonia loading (0.95 g/l). However, this is still an appreciable ammonia loading. Another undesirable feature of this extractant was that upon standing overnight, a massive precipitate formed in the loaded solvent. This problem might be solved by one or more of the following: (1) Add a phase modifier to the solvent, (2) use a more aromatic diluent, or (3) reduce the extractant concentration in the solvent.
LIX 605 had a high nickel-loading capacity (2.94 g/l) from the 100-g/l- NH4OH liquor. Unfortunately, it also had a high ammonia loading (1.96 g/l). This extractant showed very poor phase separation and a high emulsion characteristic during stripping. A phase modifier may be required if this extractant were used.
The preliminary survey had shown that none of the extractants tested showed a clear superiority, although several extractants appeared to have potential for nickel extraction from ammoniacal ammonium sulfate liquor. Because it appears that none of the extractants were clearly superior to LIX 64N, and because LIX 64N has been proven to be stable through many years of commercial operation, it will be used in subsequent larger scale continuous-circuit testing.
SX of Ni & Cu in Laterite Ammoniacal Leach Liquors
A solvent extraction system using LIX 64N solvents efficiently extracted nickel and copper from ammoniacal ammonium sulfate leach liquors and afforded their separation into pure electrolytes. Tests with 25- and 12-vol-pct-LIX 64N solvents in a laboratory-size mixer-settler continuous circuit extracted >99 pct of the nickel, substantial copper, only trace zinc, and essentially no cobalt from leach liquors produced by the method being developed. Extraction studies have shown that high ammonium hydroxide levels in the leach liquor (pH ≥ 9.5) reduce nickel loading on the solvents. For this reason, the ammonia content of the leach liquors was lowered by stripping some of the ammonia prior to the nickel-copper solvent extraction system. High ammonium sulfate levels in the leach liquors (≥ 250 g/l) had a definite slowing effect on nickel extraction. Thus, two or three nickel extraction stages were used with mixer residence times of 5 to 6 minutes per stage.
It was found that some ammonia from the leach liquors loaded onto the LIX 64N solvents. Some of this ammonia was crowded from the LIX 64N solvents by maximizing the loading of nickel. The residual ammonia was removed from the loaded solvents with a pH-controlled scrub.
Nickel was selectively stripped from the loaded 25- and 12-vol-pct- LIX 64N solvents with spent electrolytes containing 90.0 g/l Ni and 11.9 g/l H2SO4, and 95.0 g/l Ni and 9.63 g/l H2SO4, respectively. Nickel-stripped LIX 64N solvents were produced that contained 0.006 g/l Ni in five stages and <0.001 g/l Ni in four stages, respectively. Pregnant Ni electrolytes at pH 2.4 to 2.45 were produced with only traces of Cu, Co, and Zn being detected. Copper loaded on the LIX 64N solvents was controlled in a copper strip stage using electrolyte containing 25 g/l Cu and 160 g/l H2SO4.
A limited survey of some nickel and/or copper extractants (Acorga P5100, XI 54, LIX 34, LIX 70, and LIX 605) was made. The survey showed that none of these extractants were clearly superior to LIX 64N in this application. Therefore, LIX 64N will be used in subsequent larger scale continuous-circuit testing.