- Email: [email protected]
The development of a resin-in-pulp process for the recovery of nickel and cobalt from laterite leach slurries
Int. J. Miner. Process. 72 (2003) 407 – 415 www.elsevier.com/locate/ijminpro
The development of a resin-in-pulp process for the recovery of nickel and cobalt from laterite leach slurries Michael J. Nicol *, Zaimawati Zainol Parker Center, Murdoch University, South Street, Murdoch, WA 1650, Western Australia Received 21 October 2002; received in revised form 12 June 2003; accepted 1 July 2003
Abstract Up to 10% of the soluble nickel and cobalt can report to the tailings of a pressure acid leach (PAL) process for the recovery of nickel from lateritic ores as a result of the difficulties associated with the poor settling characteristics of the autoclave discharge. This paper describes the results of laboratory and pilot-plant investigations aimed at establishing the potential for the application of a resin-in-pulp (RIP) process to improve the overall nickel and cobalt recoveries from such tailings. Small-scale batch tests in the laboratory established the technical feasibility and this was followed by several campaigns using a continuous six-stage miniplant which demonstrated that the process could achieve the desired results and which enabled a pilot plant to be designed. The integrated pilot plant operated on site at a local plant over a period of several months. The paper will discuss the results from this testwork, which has culminated in a preliminary process design for a full-scale plant. In addition, the more fundamental aspects associated with the adsorption and elution processes have been investigated and process models developed for these operations. D 2003 Elsevier B.V. All rights reserved. Keywords: nickel; recovery; ion exchange; resin-in-pulp; laterite
1. Introduction The pressure acid leach (PAL) process has been applied in three recent new plants which recover nickel and cobalt from lateritic ores in Western Australia. This process was first developed and implemented at Moa Bay in Cuba and there are several new plants at the feasibility, planning and construction phases in Australia and elsewhere (Kuck, 2001).
* Corresponding author. Fax: +61-8-9360-6343. E-mail address: [email protected] (M.J. Nicol). 0301-7516/$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0301-7516(03)00115-7
The PAL process involves the leaching of the ore with sulfuric acid at temperatures in the range 250 – 270 jC followed by a number of stages of countercurrent decantation (CCD) in which the pregnant liquor is separated from the unleached residue and passed as overflow to neutralisation and metal recovery while the underflow (typically 35% solids) is removed as residue to the tailing dam. Due to washing inefficiencies with the fine laterite leach residue, an economically significant fraction of soluble nickel and cobalt passes unrecovered to the tails. Ion exchange processes are generally more suitable for treating dilute solutions of metal ions such as those
408
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
in the tailings. The major advantage of ion exchange is the ability to recover metals from slurries directly in a resin-in-pulp (RIP) process. Cation chelating resins such as those with the iminodiacetate functional group have the advantage of operating in the low pH region with high capacity and high selectivity for the metals of interest and they are also relatively easily eluted with dilute acid (Rohm and Haas, 1998). This paper summarises the results of an extended investigation into the application of an RIP process for the recovery of nickel and cobalt from the tailings at a local PAL plant. The study has involved small-scale batch experiments in the laboratory followed by the operation of a continuous miniplant in the laboratory and has culminated in several campaigns involving a continuous pilot plant on site.
2. Experimental 2.1. Laboratory testwork The experiments were carried out using a commercial weak acid iminodiacetate resin IRC 748 supplied by Rohm and Haas. The resin obtained in the sodium form was converted to the proton form by treatment with sulfuric acid and washed to pH 5 with distilled water. The theoretical capacity was measured by an acid – base titration procedure and a value of 1.132 eq/ l of resin (wet-settled volume) was obtained. The resin was screened and the fraction between 0.6 and 1 mm used in the testwork. The PAL pulp sample supplied by Anaconda was autoclave discharge and contained 6 g/l of nickel. The CCD circuit was simulated by washing with water in several stages to reduce the nickel concentration to approximately 0.5 g/l. The oversize solids were removed from the pulp by screening at 325 Am in order to allow the separation of slurry and loaded resin by wet screening prior to stripping. The pulp was then adjusted to 35% solids and was neutralised to pH 4 with lime and oxidised at 60 jC for 2 h in order to precipitate iron. The adsorption behaviour of nickel and cobalt and impurity ions were examined using a batch equilibration technique. A measured amount of the slurry was allowed to equilibrate with a known amount of resin. The change of pH due to the protons released during
the adsorption was controlled by the addition of lime. After equilibration, resin and slurry were separated by wet screening. The loaded resin was then eluted with 1 M sulfuric acid. The concentration of the metals remaining in the slurry and the eluate from elution were analysed by AAS or ICP. 2.2. Miniplant operation The pulp sample supplied by Anaconda Nickel Limited was seventh stage CCD underflow and contained approximately 0.4 g/l of nickel at 35% solid (f 5%>250 Am). The pulp was neutralised to pH 4 with calcrete (supplied by Anaconda) and oxidised at 60 jC to precipitate iron by bubbling with air. The oversized particles were removed from the pulp by prescreening at 250 Am using a Syntron vibrating screen and transferred to a 100-l mechanically agitated feed tank. A custom-built resin-in-pulp miniplant with five 2-l contactors was used. Each contactor was equipped with a 425-Am peripheral screen to retain the resin beads while permitting the slurry to flow through the plant by gravity. The flow sheet was very similar to that of the pilot plant shown in Fig. 1. The feed to the miniplant was drawn from the feed tank using a peristaltic pump at a fixed flow rate of 4 l/h. Each of the stages contained a certain volume of resin designed to yield optimum loading of nickel onto the resin from the first stage and a suitably low nickel concentration in the barren slurry exiting the last stage. The resin was transferred countercurrent to the pulp by pumping the pulp using small peristaltic pumps from each stage. The gap on the pump head was adjusted in such a way as to permit the desired flow of pulp while at the same time preventing breakage of the resin. The backmixing flow was fixed at 0.6 l/h to obtain the optimum contact time between resin and pulp. In each stage, vigorous agitation was provided to ensure efficient contact between the resin beads and the pulp, and to produce a rapid shear velocity across the screen to avoid blockage by the resin beads. The miniplant was operated in a countercurrent mode with the feed pulp being pumped into stage 1 in the circuit, which contains resin with the highest nickel loading, while the barren pulp emerges from the last stage, which contains resin with the lowest nickel loading.
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
409
Fig. 1. Flow sheet of the feed preparation and adsorption section of the RIP pilot plant.
The loaded resin from the first stage was screened from the pulp, wash and eluted with sulphuric acid. Fresh resin was added periodically to the last stage. Elution of the loaded resin was carried out in a small glass column with a bed volume (BV) of 100 ml. The eluant was supplied to the column by using a peristaltic pump at 2 BV/h. Fractions consisting of 10 ml of the eluate were obtained using a fraction collector and were analysed by ICP-OES. 2.3. Pilot plant operation 2.3.1. Feed preparation The pulp sample was a CCD 7 underflow and contained approximately 0.2 g/l of nickel at 20% solids. The pulp was pumped to an agitated 10 m3 fibre-glass mixing tank and neutralised to pH 4 with 50 l of a calcrete slurry. The pulp was oxidised at room temperature to precipitate iron by sparging with air for 2 days. In addition, 2.5 l of hydrogen peroxide were added to accelerate the precipitation of iron. The hydrostatic head in the neutralisation tank allowed flow of the pulp over a Kason vibrating screen on which it was screened at 180 Am to remove oversized particles. The underflow was fed directly to the 150l mechanically agitated feed tank that was sited under the vibrating screen. The flow sheet is shown schematically in Fig. 1.
2.3.2. Adsorption A custom-built resin-in-pulp pilot plant with five 40-l contactors was used during the operation of the pilot plant. Each contactor was equipped with a 425Am cylindrical peripheral screen to retain the resin beads while permitting the slurry to flow through the plant by gravity. The feed to the pilot plant was drawn from a feed tank using a peristaltic pump at a fixed flow rate of 80 l/h. The resin was transferred countercurrent to the pulp by using airlift devices, which were attached to each tank. The regulation of air and the backmixing flow was controlled by a rotameter installed on each tank. The backmixing flow was fixed at 12 l/h to obtain an optimum residence time for the resin. In each stage, mechanical agitation was provided to ensure efficient contact between the resin beads and the pulp, and to produce a rapid shear velocity across the screen to avoid blockage by the resin beads. Fresh resin was added manually at fixed times to the last tank. 2.3.3. Elution The loaded resin was screened from the pulp by a laboratory Syntron vibrating screen with the underflow pulp flowing to waste. The loaded resin was then washed with water and transferred periodically to a 20-l elution column as shown in Fig. 2.
410
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
Fig. 4. Kinetics of adsorption of nickel from neutralised pulp at pH 4. Treatment of the data in terms of a first-order approach to equilibrium is also shown. Fig. 2. Flow sheet of the elution circuit of the RIP pilot plant.
After loading, the resin in the column was washed at 30 l/h with process water from the plant for approximately 30 min to remove as much entrained pulp particles as possible. Sulphuric acid (1 M) was then pumped upflow through the column at 18 l/h and the eluate collected in a 200-l plastic tank. The elution process takes approximately 3 h. Samples of eluate were taken periodically and analysed by ICP-OES. On completion of the elution, the column was washed again with process water until the solution exiting the column had a pH of at least 3. The volume of the eluate was measured and a sample was taken to be analysed. The regenerated resin was drained from the column and recycled to the adsorption circuit.
In all cases, the metal loading on the resin has been calculated as mass of metal per unit volume of wetsettled resin as is the convention in the resin industry.
3. Results and discussion 3.1. Batch laboratory testwork Small-scale batch tests were conducted in the laboratory to assess the technical feasibility of the recovery of nickel and cobalt from a typical CCD tail pulp from a local operation. The screened pulp was contacted batchwise with fresh resin in four stages with the resin separated after each contact for 4 h at a pH value of 4, eluted and the eluate analysed for nickel. Kinetic tests showed that equilibrium was essentially achieved after 4 h. The results are shown as an adsorption isotherm in Fig. 3. It is apparent that a nickel loading of about 20 g/l of resin can be achieved by adsorption from the pulp. The relatively steep isotherm at low solution concenTable 1 Analysis of the solution phase in the feed pulp
Fig. 3. Equilibrium loading of nickel from pulp at pH 4.
Species
Concentration (g/l)
Nickel Cobalt Calcium Copper Iron Magnesium Manganese
0.355 0.03 0.56 0.001 0.27 9.7 0.6
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
411
Fig. 7. Elution of loaded resin with 1 M sulfuric acid. Nickel (x), iron (E), magnesium (n). Fig. 5. Solution loading profiles after 5 (- - - - -) and 10 (_____) h.
trations permits the efficient extraction of nickel. The deportment of the other metal ions in this system will be discussed below. A Freundlich isotherm was found to give a suitable fit to the adsorption isotherm data. The kinetics of the adsorption of nickel during a batch experiment under the same conditions are shown in Fig. 4. This and other data could be fitted well to a simple rate equation based on a first-order approach to equilibrium as a result of rate-determining mass transport in the pulp phase as shown by the linear logarithmic plot in Fig. 4. These results were used to develop a preliminary simulation for a multistage RIP process using software which has been used with success in the simulation of carbon-in-pulp processes. The results of the
simulation were then used to define the operating parameters for the continuous RIP miniplant testwork. 3.2. Miniplant testwork Two experimental runs were conducted with two commercial resins under the same conditions and parameters to confirm the predicted loading performance and to establish the elution characteristics for each type of resin. Both runs were designed for operation over 10 h, with a mean retention time of slurry in each adsorption tank of 30 min and a resin concentration of 30 g/l of pulp in each stage. The composition of the CCD tails used as a feed to the miniplant is given in Table 1. The first run was carried out using Amberlite IRC 748 resin. The solution loading profile was obtained by sampling the solution in each stage after 5 and 10 h and the results shown in Fig. 5. A sample of the loaded resin was taken every 2 h to establish the approach to steady state which was achieved after about 6 h. It is apparent from Fig. 5 that the desired nickel concentration in the barren (0.03 g/l) had been achieved in the five stages. The distorted profile after 10 h was found to be due to an uneven distribution of the resin with a higher concentration in the last three stages. Despite this, the Table 2 Recovery of metal ions (percentage of feed to RIP)
Fig. 6. Solution (E) and resin (n) loading profiles after 10 h. The dotted lines are the corresponding profiles predicted by the simulation.
(Head – tail)/head Resin eluate/head
Ca
Co
Cu
Fe
Mg
Mn
Ni
11.8 3.2
61.2 51.2
82.0 164.6
8.1 17.2
15.1 0.5
19.7 4.7
89.4 89.3
412
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
Table 3 Main operating parameters for pilot plant run 1 Pulp flow rate Backmixing flow rate Resin flow rate Resin concentration Initial resin load Fresh resin addition
80 l/h 12 l/h 0.96 l/h 30 g/l 3.2 l/tank 480 ml every 30 min
performance in terms of metal recovered was not substantially affected. This phenomenon has been commonly observed in carbon-in-pulp plants and illustrates the robustness of the multistage approach. Fig. 6 illustrates the fact that the solution and resin loading after 10 h follow the profile as predicted in the RIP simulation based on the batch adsorption data. The loaded resin was eluted in a column at a flow rate of 2 bed volumes (BV) per hour using a 1-M sulphuric acid eluant. The results showed that 2 BV of eluant is sufficient to strip all the metals loaded. Typical elution profiles for Mg, Ni and Fe are shown in Fig. 7. The sequence of elution follows the expected trend based on the published selectivity coefficients for the adsorption of the metal ions on this type of resin. The recovery of the various metals based on both the loss from the pulp and as recovered in the elate is given in Table 2. It is apparent that the recovery of nickel achieved was less than 1% below the target value of 90% while more than 50% cobalt was been recovered together
with small amounts of the other metal ions. The poor accountability for some of the metal ions can be attributed to the difficulties in the assay of the low concentrations in the pulp. A second miniplant run was carried out using a sample of Lewatit TP207 resin under essentially similar conditions. Batch tests with this resin had shown that the kinetics of loading were inferior to that of the IRC 748 resin used in the first run. This was borne out in practice in that, despite a higher resin flow rate (and consequent lower loading of nickel on the resin), the recovery of nickel was less than 70%. For this reason, it was decided to use the IRC 748 resin in the pilot plant. The results achieved in the miniplant were very promising and a decision was made to progress to an integrated pilot plant at a scale some 20 that of the miniplant. 3.3. Pilot plant testwork The pilot plant was assembled as described above and transported to site where it was located close to the final CCD thickener from which the feed pulp was drawn. Several campaigns were run over a period of about 2 months. The operating parameters used during the first run were based on the RIP simulation which was slightly modified to take into account the results obtained during the miniplant runs. All the operating parameters, which are summarised in Table 3, were based on
Fig. 8. Results obtained during the second pilot-plant run. (___) Ni extraction, (.....) Co extraction, (E) Ni loading on the resin.
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
413
Fig. 9. Results obtained during the third pilot plant run. (____) Nickel recovery, (.....) cobalt recovery.
a target of 90% nickel and 60% cobalt extraction. The pulp was screened at 180 Am and the + 600 Am resin fraction used in the testwork. Unreacted coarse calcrete particles constituted the main component of the + 180 Am fraction of the pulp, which was less than 5% of the solids in the feed. Due to several operational problems associated with the preparation of the feed pulp and the control of the air to the interstage airlifts, the first run achieved an average extraction of 55% for nickel and 15% for cobalt. Although the elution of the resin was successful in recovering the nickel and cobalt, it was noticed that the iron, and particularly, the chromium content of the eluted resin increased during the run. A laboratory investigation showed that an in-
crease in the concentration of the acid in the eluant and an increase in the temperature of elution to 50 jC resulted in essentially complete removal of these metals from the resin. With hindsight, this could have been expected given the slow ligand exchange kinetics for iron(III) and, particularly, chromium(III). As a result of experience gained in the first run, a second run was conducted with similar operating parameters and significantly improved results were obtained as shown by the data in Fig. 8. It is apparent that both the recovery of both nickel and cobalt appeared to decrease with time of operation and, although the extraction of nickel exceeded the target of 90% for the first 60 h, it dropped to about 80% soon thereafter while the target of 70% for
Fig. 10. Steady-state concentrations of nickel (___) and cobalt (.....) in each stage during operation of the pilot plant.
414
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
Table 4 Analysis of solids in the feed and tails
Average feed Average tail Feed/tail
Al (%)
Ca (%)
Cr (%)
Fe (%)
Mg (%)
Si (%)
Co (%)
Mn (%)
Ni (%)
Zn (%)
2.23 2.21 1.01
4.36 4.48 0.97
0.423 0.394 1.07
16.93 16.91 1
1.69 1.56 1.09
15.31 15.49 0.99
0.0097 0.0062 1.57
0.131 0.117 1.12
0.155 0.115 1.36
0.0061 0.0045 1.33
cobalt was not achieved except in the first 12 h. The loading of nickel on the resin was acceptable. The drop in the nickel recovery after 60 h is believed to be associated with changes in the pulp density in the neutralisation tank. This was caused by the settling of solids as a result of insufficient agitation after the pulp level dropped below the level of the agitator in the tank. This resulted in a greater feed rate of solution (and hence dissolved nickel) to the RIP circuit. Elution with warm 1.5-M acid proved to be very effective in removing residual iron and chromium from the resin. Removal of residual acid from the resin after elution by washing with water was found to be a slow process and it was decided that a limited wash with water followed by dilute ammonia would reduce the elution/wash time. A third run was therefore undertaken after installing additional agitation in the neutralisation tank, eliminating the heating of the eluant and introducing an ammonia wash cycle. The results obtained during the next run are shown in Fig. 9. The improvement in both nickel and cobalt recovery is obvious and the target values were easily achieved once steady state had been achieved. The solution profiles at steady state are shown in Fig. 10. It is interesting to note that the cobalt profile does not follow the expected exponential trend and this can be attributed to the fact that it is adsorbed less strongly than nickel and is therefore ‘‘squeezed off’’ the resin by the high concentration of loaded nickel in the first two stages. An important observation was made as a result of a careful analysis of the solids in the feed and tail from the adsorption process. The average values of samples taken at regular intervals throughout the run are shown in Table 4. Thus, for the metals in the first section, the metal content does not change during adsorption. However, for Co, Ni, Zn and possibly Mn, the ratio of tail to feed suggests that additional quantities of these metal
ions were recovered from the solid phase in the RIP adsorption process. It is suspected that this additional recovery is associated with metal ions adsorbed on the fine hematite produced in the pressure leach process. It is well known that nickel and, particularly, cobalt ions are strongly adsorbed onto the surfaces of iron oxides. This ‘‘extra dissolution’’ due to desorption of metal ions as a result of the low concentrations in the later stages of a RIP process is an additional benefit which is lost in a conventional solid/liquid separation step. Typical profiles of the composition of the eluate are shown in Fig. 11. The results for Ni and Co confirmed the laboratory testwork in that about 2 BV is required for efficient elution. The slow elution for chromium is apparent in this data. An analysis of the resin after a number of loading/ elution/regeneration cycles is compared with that of fresh resin in Table 5. With the exception of Ca, Fe and Mg, the metal ion content of the resin does not increase noticeably during the process of adsorption and elution. In the case of the former metal ions, the concentrations appeared to reach a steady-state value similar to those in Table 5.
Fig. 11. Typical concentration profiles for (___) Ni, (.....) Co and (_n_) Cr during the elution of loaded resin with 1.5 M sulfuric acid.
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
415
Table 5 Analyses of recycled and fresh resin samples
Fresh resin Regenerated resin
Al (%)
Ca (ppm)
Co (ppm)
Cr (ppm)
Cu (ppm)
Fe (ppm)
Mg (ppm)
Mn (ppm)
Ni (%)
Zn (ppm)
0.33 0.16
440 2280
12 28
17 17
5.5 15
400 1000
480 2420
5.6 10.3
0.02 0.01
43 <2
3.4. Other testwork A number of additional aspects of the process were investigated both in the laboratory and on a pilot-plant scale. In particular, extended testwork was carried out on a single stage pump-cell contactor of volume 1.4 m3 in order to establish the screening duty at various pulp flow rates and resin sizes and concentrations in the pulp. The extent of resin degradation was also assessed as a part of this testwork. On the basis of the results obtained during the various phases of the project, preliminary designs of a full-scale plant were made using the simulation software, which was shown to accurately predict the performance of the pilot plant from the batch laboratory data. Both conventional continuous multistage contactors as used in the CIP process and pump cell units operated in a carousel system were modeled. The possibility of incorporating the RIP process after one of the earlier stages of CCD was also considered and economic comparisons made of the various options.
site of an Australian operation. The most appropriate resin has been selected and the important operating parameters have been established. The results obtained in the final pilot plant run exceeded expectations in terms of metal recovered. Problems associated with preparation of the pulp, elution of the resin and control of the adsorption process have been resolved. A useful modelling and simulation package has been shown to accurately predict the performance of both the laboratory and pilot-scale adsorption processes and can be used for the design of a fullscale plant.
Acknowledgements The contributions made by Serge Lallenec, Les Stewart and Spencer Smith from Anaconda Nickel to the success of this project are gratefully acknowledged. Thanks are also due to the A J Parker Cooperative Research Center for Hydrometallurgy for financial support for ZZ.
4. Conclusions The technical feasibility of a resin-in-pulp process for the efficient recovery of nickel and cobalt from tailings of a pressure acid leach process has been conclusively demonstrated in an extended study involving laboratory and pilot-plant investigations at the
References Kuck, P.H., 2001. Mineral Commodity Summaries. U.S. Geological Survey, U.S. Department of the Interior, Washington, DC, p. 113. Rohm and Haas, 1998. Product Data Sheet Amberlite IRC748, PDS 0575A, Philadelphia, USA.
The development of a resin-in-pulp process for the recovery of nickel and cobalt from laterite leach slurries Michael J. Nicol *, Zaimawati Zainol Parker Center, Murdoch University, South Street, Murdoch, WA 1650, Western Australia Received 21 October 2002; received in revised form 12 June 2003; accepted 1 July 2003
Abstract Up to 10% of the soluble nickel and cobalt can report to the tailings of a pressure acid leach (PAL) process for the recovery of nickel from lateritic ores as a result of the difficulties associated with the poor settling characteristics of the autoclave discharge. This paper describes the results of laboratory and pilot-plant investigations aimed at establishing the potential for the application of a resin-in-pulp (RIP) process to improve the overall nickel and cobalt recoveries from such tailings. Small-scale batch tests in the laboratory established the technical feasibility and this was followed by several campaigns using a continuous six-stage miniplant which demonstrated that the process could achieve the desired results and which enabled a pilot plant to be designed. The integrated pilot plant operated on site at a local plant over a period of several months. The paper will discuss the results from this testwork, which has culminated in a preliminary process design for a full-scale plant. In addition, the more fundamental aspects associated with the adsorption and elution processes have been investigated and process models developed for these operations. D 2003 Elsevier B.V. All rights reserved. Keywords: nickel; recovery; ion exchange; resin-in-pulp; laterite
1. Introduction The pressure acid leach (PAL) process has been applied in three recent new plants which recover nickel and cobalt from lateritic ores in Western Australia. This process was first developed and implemented at Moa Bay in Cuba and there are several new plants at the feasibility, planning and construction phases in Australia and elsewhere (Kuck, 2001).
* Corresponding author. Fax: +61-8-9360-6343. E-mail address: [email protected] (M.J. Nicol). 0301-7516/$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0301-7516(03)00115-7
The PAL process involves the leaching of the ore with sulfuric acid at temperatures in the range 250 – 270 jC followed by a number of stages of countercurrent decantation (CCD) in which the pregnant liquor is separated from the unleached residue and passed as overflow to neutralisation and metal recovery while the underflow (typically 35% solids) is removed as residue to the tailing dam. Due to washing inefficiencies with the fine laterite leach residue, an economically significant fraction of soluble nickel and cobalt passes unrecovered to the tails. Ion exchange processes are generally more suitable for treating dilute solutions of metal ions such as those
408
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
in the tailings. The major advantage of ion exchange is the ability to recover metals from slurries directly in a resin-in-pulp (RIP) process. Cation chelating resins such as those with the iminodiacetate functional group have the advantage of operating in the low pH region with high capacity and high selectivity for the metals of interest and they are also relatively easily eluted with dilute acid (Rohm and Haas, 1998). This paper summarises the results of an extended investigation into the application of an RIP process for the recovery of nickel and cobalt from the tailings at a local PAL plant. The study has involved small-scale batch experiments in the laboratory followed by the operation of a continuous miniplant in the laboratory and has culminated in several campaigns involving a continuous pilot plant on site.
2. Experimental 2.1. Laboratory testwork The experiments were carried out using a commercial weak acid iminodiacetate resin IRC 748 supplied by Rohm and Haas. The resin obtained in the sodium form was converted to the proton form by treatment with sulfuric acid and washed to pH 5 with distilled water. The theoretical capacity was measured by an acid – base titration procedure and a value of 1.132 eq/ l of resin (wet-settled volume) was obtained. The resin was screened and the fraction between 0.6 and 1 mm used in the testwork. The PAL pulp sample supplied by Anaconda was autoclave discharge and contained 6 g/l of nickel. The CCD circuit was simulated by washing with water in several stages to reduce the nickel concentration to approximately 0.5 g/l. The oversize solids were removed from the pulp by screening at 325 Am in order to allow the separation of slurry and loaded resin by wet screening prior to stripping. The pulp was then adjusted to 35% solids and was neutralised to pH 4 with lime and oxidised at 60 jC for 2 h in order to precipitate iron. The adsorption behaviour of nickel and cobalt and impurity ions were examined using a batch equilibration technique. A measured amount of the slurry was allowed to equilibrate with a known amount of resin. The change of pH due to the protons released during
the adsorption was controlled by the addition of lime. After equilibration, resin and slurry were separated by wet screening. The loaded resin was then eluted with 1 M sulfuric acid. The concentration of the metals remaining in the slurry and the eluate from elution were analysed by AAS or ICP. 2.2. Miniplant operation The pulp sample supplied by Anaconda Nickel Limited was seventh stage CCD underflow and contained approximately 0.4 g/l of nickel at 35% solid (f 5%>250 Am). The pulp was neutralised to pH 4 with calcrete (supplied by Anaconda) and oxidised at 60 jC to precipitate iron by bubbling with air. The oversized particles were removed from the pulp by prescreening at 250 Am using a Syntron vibrating screen and transferred to a 100-l mechanically agitated feed tank. A custom-built resin-in-pulp miniplant with five 2-l contactors was used. Each contactor was equipped with a 425-Am peripheral screen to retain the resin beads while permitting the slurry to flow through the plant by gravity. The flow sheet was very similar to that of the pilot plant shown in Fig. 1. The feed to the miniplant was drawn from the feed tank using a peristaltic pump at a fixed flow rate of 4 l/h. Each of the stages contained a certain volume of resin designed to yield optimum loading of nickel onto the resin from the first stage and a suitably low nickel concentration in the barren slurry exiting the last stage. The resin was transferred countercurrent to the pulp by pumping the pulp using small peristaltic pumps from each stage. The gap on the pump head was adjusted in such a way as to permit the desired flow of pulp while at the same time preventing breakage of the resin. The backmixing flow was fixed at 0.6 l/h to obtain the optimum contact time between resin and pulp. In each stage, vigorous agitation was provided to ensure efficient contact between the resin beads and the pulp, and to produce a rapid shear velocity across the screen to avoid blockage by the resin beads. The miniplant was operated in a countercurrent mode with the feed pulp being pumped into stage 1 in the circuit, which contains resin with the highest nickel loading, while the barren pulp emerges from the last stage, which contains resin with the lowest nickel loading.
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
409
Fig. 1. Flow sheet of the feed preparation and adsorption section of the RIP pilot plant.
The loaded resin from the first stage was screened from the pulp, wash and eluted with sulphuric acid. Fresh resin was added periodically to the last stage. Elution of the loaded resin was carried out in a small glass column with a bed volume (BV) of 100 ml. The eluant was supplied to the column by using a peristaltic pump at 2 BV/h. Fractions consisting of 10 ml of the eluate were obtained using a fraction collector and were analysed by ICP-OES. 2.3. Pilot plant operation 2.3.1. Feed preparation The pulp sample was a CCD 7 underflow and contained approximately 0.2 g/l of nickel at 20% solids. The pulp was pumped to an agitated 10 m3 fibre-glass mixing tank and neutralised to pH 4 with 50 l of a calcrete slurry. The pulp was oxidised at room temperature to precipitate iron by sparging with air for 2 days. In addition, 2.5 l of hydrogen peroxide were added to accelerate the precipitation of iron. The hydrostatic head in the neutralisation tank allowed flow of the pulp over a Kason vibrating screen on which it was screened at 180 Am to remove oversized particles. The underflow was fed directly to the 150l mechanically agitated feed tank that was sited under the vibrating screen. The flow sheet is shown schematically in Fig. 1.
2.3.2. Adsorption A custom-built resin-in-pulp pilot plant with five 40-l contactors was used during the operation of the pilot plant. Each contactor was equipped with a 425Am cylindrical peripheral screen to retain the resin beads while permitting the slurry to flow through the plant by gravity. The feed to the pilot plant was drawn from a feed tank using a peristaltic pump at a fixed flow rate of 80 l/h. The resin was transferred countercurrent to the pulp by using airlift devices, which were attached to each tank. The regulation of air and the backmixing flow was controlled by a rotameter installed on each tank. The backmixing flow was fixed at 12 l/h to obtain an optimum residence time for the resin. In each stage, mechanical agitation was provided to ensure efficient contact between the resin beads and the pulp, and to produce a rapid shear velocity across the screen to avoid blockage by the resin beads. Fresh resin was added manually at fixed times to the last tank. 2.3.3. Elution The loaded resin was screened from the pulp by a laboratory Syntron vibrating screen with the underflow pulp flowing to waste. The loaded resin was then washed with water and transferred periodically to a 20-l elution column as shown in Fig. 2.
410
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
Fig. 4. Kinetics of adsorption of nickel from neutralised pulp at pH 4. Treatment of the data in terms of a first-order approach to equilibrium is also shown. Fig. 2. Flow sheet of the elution circuit of the RIP pilot plant.
After loading, the resin in the column was washed at 30 l/h with process water from the plant for approximately 30 min to remove as much entrained pulp particles as possible. Sulphuric acid (1 M) was then pumped upflow through the column at 18 l/h and the eluate collected in a 200-l plastic tank. The elution process takes approximately 3 h. Samples of eluate were taken periodically and analysed by ICP-OES. On completion of the elution, the column was washed again with process water until the solution exiting the column had a pH of at least 3. The volume of the eluate was measured and a sample was taken to be analysed. The regenerated resin was drained from the column and recycled to the adsorption circuit.
In all cases, the metal loading on the resin has been calculated as mass of metal per unit volume of wetsettled resin as is the convention in the resin industry.
3. Results and discussion 3.1. Batch laboratory testwork Small-scale batch tests were conducted in the laboratory to assess the technical feasibility of the recovery of nickel and cobalt from a typical CCD tail pulp from a local operation. The screened pulp was contacted batchwise with fresh resin in four stages with the resin separated after each contact for 4 h at a pH value of 4, eluted and the eluate analysed for nickel. Kinetic tests showed that equilibrium was essentially achieved after 4 h. The results are shown as an adsorption isotherm in Fig. 3. It is apparent that a nickel loading of about 20 g/l of resin can be achieved by adsorption from the pulp. The relatively steep isotherm at low solution concenTable 1 Analysis of the solution phase in the feed pulp
Fig. 3. Equilibrium loading of nickel from pulp at pH 4.
Species
Concentration (g/l)
Nickel Cobalt Calcium Copper Iron Magnesium Manganese
0.355 0.03 0.56 0.001 0.27 9.7 0.6
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
411
Fig. 7. Elution of loaded resin with 1 M sulfuric acid. Nickel (x), iron (E), magnesium (n). Fig. 5. Solution loading profiles after 5 (- - - - -) and 10 (_____) h.
trations permits the efficient extraction of nickel. The deportment of the other metal ions in this system will be discussed below. A Freundlich isotherm was found to give a suitable fit to the adsorption isotherm data. The kinetics of the adsorption of nickel during a batch experiment under the same conditions are shown in Fig. 4. This and other data could be fitted well to a simple rate equation based on a first-order approach to equilibrium as a result of rate-determining mass transport in the pulp phase as shown by the linear logarithmic plot in Fig. 4. These results were used to develop a preliminary simulation for a multistage RIP process using software which has been used with success in the simulation of carbon-in-pulp processes. The results of the
simulation were then used to define the operating parameters for the continuous RIP miniplant testwork. 3.2. Miniplant testwork Two experimental runs were conducted with two commercial resins under the same conditions and parameters to confirm the predicted loading performance and to establish the elution characteristics for each type of resin. Both runs were designed for operation over 10 h, with a mean retention time of slurry in each adsorption tank of 30 min and a resin concentration of 30 g/l of pulp in each stage. The composition of the CCD tails used as a feed to the miniplant is given in Table 1. The first run was carried out using Amberlite IRC 748 resin. The solution loading profile was obtained by sampling the solution in each stage after 5 and 10 h and the results shown in Fig. 5. A sample of the loaded resin was taken every 2 h to establish the approach to steady state which was achieved after about 6 h. It is apparent from Fig. 5 that the desired nickel concentration in the barren (0.03 g/l) had been achieved in the five stages. The distorted profile after 10 h was found to be due to an uneven distribution of the resin with a higher concentration in the last three stages. Despite this, the Table 2 Recovery of metal ions (percentage of feed to RIP)
Fig. 6. Solution (E) and resin (n) loading profiles after 10 h. The dotted lines are the corresponding profiles predicted by the simulation.
(Head – tail)/head Resin eluate/head
Ca
Co
Cu
Fe
Mg
Mn
Ni
11.8 3.2
61.2 51.2
82.0 164.6
8.1 17.2
15.1 0.5
19.7 4.7
89.4 89.3
412
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
Table 3 Main operating parameters for pilot plant run 1 Pulp flow rate Backmixing flow rate Resin flow rate Resin concentration Initial resin load Fresh resin addition
80 l/h 12 l/h 0.96 l/h 30 g/l 3.2 l/tank 480 ml every 30 min
performance in terms of metal recovered was not substantially affected. This phenomenon has been commonly observed in carbon-in-pulp plants and illustrates the robustness of the multistage approach. Fig. 6 illustrates the fact that the solution and resin loading after 10 h follow the profile as predicted in the RIP simulation based on the batch adsorption data. The loaded resin was eluted in a column at a flow rate of 2 bed volumes (BV) per hour using a 1-M sulphuric acid eluant. The results showed that 2 BV of eluant is sufficient to strip all the metals loaded. Typical elution profiles for Mg, Ni and Fe are shown in Fig. 7. The sequence of elution follows the expected trend based on the published selectivity coefficients for the adsorption of the metal ions on this type of resin. The recovery of the various metals based on both the loss from the pulp and as recovered in the elate is given in Table 2. It is apparent that the recovery of nickel achieved was less than 1% below the target value of 90% while more than 50% cobalt was been recovered together
with small amounts of the other metal ions. The poor accountability for some of the metal ions can be attributed to the difficulties in the assay of the low concentrations in the pulp. A second miniplant run was carried out using a sample of Lewatit TP207 resin under essentially similar conditions. Batch tests with this resin had shown that the kinetics of loading were inferior to that of the IRC 748 resin used in the first run. This was borne out in practice in that, despite a higher resin flow rate (and consequent lower loading of nickel on the resin), the recovery of nickel was less than 70%. For this reason, it was decided to use the IRC 748 resin in the pilot plant. The results achieved in the miniplant were very promising and a decision was made to progress to an integrated pilot plant at a scale some 20 that of the miniplant. 3.3. Pilot plant testwork The pilot plant was assembled as described above and transported to site where it was located close to the final CCD thickener from which the feed pulp was drawn. Several campaigns were run over a period of about 2 months. The operating parameters used during the first run were based on the RIP simulation which was slightly modified to take into account the results obtained during the miniplant runs. All the operating parameters, which are summarised in Table 3, were based on
Fig. 8. Results obtained during the second pilot-plant run. (___) Ni extraction, (.....) Co extraction, (E) Ni loading on the resin.
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
413
Fig. 9. Results obtained during the third pilot plant run. (____) Nickel recovery, (.....) cobalt recovery.
a target of 90% nickel and 60% cobalt extraction. The pulp was screened at 180 Am and the + 600 Am resin fraction used in the testwork. Unreacted coarse calcrete particles constituted the main component of the + 180 Am fraction of the pulp, which was less than 5% of the solids in the feed. Due to several operational problems associated with the preparation of the feed pulp and the control of the air to the interstage airlifts, the first run achieved an average extraction of 55% for nickel and 15% for cobalt. Although the elution of the resin was successful in recovering the nickel and cobalt, it was noticed that the iron, and particularly, the chromium content of the eluted resin increased during the run. A laboratory investigation showed that an in-
crease in the concentration of the acid in the eluant and an increase in the temperature of elution to 50 jC resulted in essentially complete removal of these metals from the resin. With hindsight, this could have been expected given the slow ligand exchange kinetics for iron(III) and, particularly, chromium(III). As a result of experience gained in the first run, a second run was conducted with similar operating parameters and significantly improved results were obtained as shown by the data in Fig. 8. It is apparent that both the recovery of both nickel and cobalt appeared to decrease with time of operation and, although the extraction of nickel exceeded the target of 90% for the first 60 h, it dropped to about 80% soon thereafter while the target of 70% for
Fig. 10. Steady-state concentrations of nickel (___) and cobalt (.....) in each stage during operation of the pilot plant.
414
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
Table 4 Analysis of solids in the feed and tails
Average feed Average tail Feed/tail
Al (%)
Ca (%)
Cr (%)
Fe (%)
Mg (%)
Si (%)
Co (%)
Mn (%)
Ni (%)
Zn (%)
2.23 2.21 1.01
4.36 4.48 0.97
0.423 0.394 1.07
16.93 16.91 1
1.69 1.56 1.09
15.31 15.49 0.99
0.0097 0.0062 1.57
0.131 0.117 1.12
0.155 0.115 1.36
0.0061 0.0045 1.33
cobalt was not achieved except in the first 12 h. The loading of nickel on the resin was acceptable. The drop in the nickel recovery after 60 h is believed to be associated with changes in the pulp density in the neutralisation tank. This was caused by the settling of solids as a result of insufficient agitation after the pulp level dropped below the level of the agitator in the tank. This resulted in a greater feed rate of solution (and hence dissolved nickel) to the RIP circuit. Elution with warm 1.5-M acid proved to be very effective in removing residual iron and chromium from the resin. Removal of residual acid from the resin after elution by washing with water was found to be a slow process and it was decided that a limited wash with water followed by dilute ammonia would reduce the elution/wash time. A third run was therefore undertaken after installing additional agitation in the neutralisation tank, eliminating the heating of the eluant and introducing an ammonia wash cycle. The results obtained during the next run are shown in Fig. 9. The improvement in both nickel and cobalt recovery is obvious and the target values were easily achieved once steady state had been achieved. The solution profiles at steady state are shown in Fig. 10. It is interesting to note that the cobalt profile does not follow the expected exponential trend and this can be attributed to the fact that it is adsorbed less strongly than nickel and is therefore ‘‘squeezed off’’ the resin by the high concentration of loaded nickel in the first two stages. An important observation was made as a result of a careful analysis of the solids in the feed and tail from the adsorption process. The average values of samples taken at regular intervals throughout the run are shown in Table 4. Thus, for the metals in the first section, the metal content does not change during adsorption. However, for Co, Ni, Zn and possibly Mn, the ratio of tail to feed suggests that additional quantities of these metal
ions were recovered from the solid phase in the RIP adsorption process. It is suspected that this additional recovery is associated with metal ions adsorbed on the fine hematite produced in the pressure leach process. It is well known that nickel and, particularly, cobalt ions are strongly adsorbed onto the surfaces of iron oxides. This ‘‘extra dissolution’’ due to desorption of metal ions as a result of the low concentrations in the later stages of a RIP process is an additional benefit which is lost in a conventional solid/liquid separation step. Typical profiles of the composition of the eluate are shown in Fig. 11. The results for Ni and Co confirmed the laboratory testwork in that about 2 BV is required for efficient elution. The slow elution for chromium is apparent in this data. An analysis of the resin after a number of loading/ elution/regeneration cycles is compared with that of fresh resin in Table 5. With the exception of Ca, Fe and Mg, the metal ion content of the resin does not increase noticeably during the process of adsorption and elution. In the case of the former metal ions, the concentrations appeared to reach a steady-state value similar to those in Table 5.
Fig. 11. Typical concentration profiles for (___) Ni, (.....) Co and (_n_) Cr during the elution of loaded resin with 1.5 M sulfuric acid.
M.J. Nicol, Z. Zainol / Int. J. Miner. Process. 72 (2003) 407–415
415
Table 5 Analyses of recycled and fresh resin samples
Fresh resin Regenerated resin
Al (%)
Ca (ppm)
Co (ppm)
Cr (ppm)
Cu (ppm)
Fe (ppm)
Mg (ppm)
Mn (ppm)
Ni (%)
Zn (ppm)
0.33 0.16
440 2280
12 28
17 17
5.5 15
400 1000
480 2420
5.6 10.3
0.02 0.01
43 <2
3.4. Other testwork A number of additional aspects of the process were investigated both in the laboratory and on a pilot-plant scale. In particular, extended testwork was carried out on a single stage pump-cell contactor of volume 1.4 m3 in order to establish the screening duty at various pulp flow rates and resin sizes and concentrations in the pulp. The extent of resin degradation was also assessed as a part of this testwork. On the basis of the results obtained during the various phases of the project, preliminary designs of a full-scale plant were made using the simulation software, which was shown to accurately predict the performance of the pilot plant from the batch laboratory data. Both conventional continuous multistage contactors as used in the CIP process and pump cell units operated in a carousel system were modeled. The possibility of incorporating the RIP process after one of the earlier stages of CCD was also considered and economic comparisons made of the various options.
site of an Australian operation. The most appropriate resin has been selected and the important operating parameters have been established. The results obtained in the final pilot plant run exceeded expectations in terms of metal recovered. Problems associated with preparation of the pulp, elution of the resin and control of the adsorption process have been resolved. A useful modelling and simulation package has been shown to accurately predict the performance of both the laboratory and pilot-scale adsorption processes and can be used for the design of a fullscale plant.
Acknowledgements The contributions made by Serge Lallenec, Les Stewart and Spencer Smith from Anaconda Nickel to the success of this project are gratefully acknowledged. Thanks are also due to the A J Parker Cooperative Research Center for Hydrometallurgy for financial support for ZZ.
4. Conclusions The technical feasibility of a resin-in-pulp process for the efficient recovery of nickel and cobalt from tailings of a pressure acid leach process has been conclusively demonstrated in an extended study involving laboratory and pilot-plant investigations at the
References Kuck, P.H., 2001. Mineral Commodity Summaries. U.S. Geological Survey, U.S. Department of the Interior, Washington, DC, p. 113. Rohm and Haas, 1998. Product Data Sheet Amberlite IRC748, PDS 0575A, Philadelphia, USA.