Você está na página 1de 11

Chemistry for Sustainable Development  "&`' 81

Cobalt-Nickel Separation in Hydrometallurgy: a Review*

DOUGLAS S. FLETT

St. Barbara Consultancy Services, 17 Foster Close, Stevenage, Herts, SGI 4SA (UK)
E-mail: doug.flett@lineone.net

Abstract
The separation of cobalt from nickel in aqueous solution has always been a problem for hydrometallurgists.
Their adjacent positions in the transition metal series in the periodic table result in aqueous chemical behaviour
that is too similar for development of easy separation routes. Traditionally cobalt and nickel were separated by
processes based on selective oxidation and/or preci pitation of cobalt from either sulphate or chloride solution
and such processes are still in use today. However, the process of solvent extraction provides the high degree
of separation and yields demanded by industry nowadays.
While alkylamines are the extractants of choice for separation of cobalt from nickel from chloride liquors,
for weakly acidic sulphate liquors the alkyl phosphorous acids have found significant commercial application at
various locations around the world. Because of the high nickel to cobalt ratio encountered in liquors produced
in sulphate-based high nickel matte leach processes or those produced in the acid pressure leaching of nickel
laterites, very high separation factors (>1000) are required. That is why the dialkyl phosphinic acid CYANEX
272 has become the reagent of choice for such duties.
This paper reviews the chemistry of cobalt-nickel separation from aqueous solutions and comments on the
implications of this chemistry in hydrometallurgical applications. Descri ption of selected applications is given
and discussed.

INTRODUCTION a marked tendency to form a tetrahedral con-


figuration under more concentrated electrolyte
The separation of cobalt from nickel in conditions rather than the hexagon al configu-
aqueous solution has always been a problem in ration of the six-coordin ated species. These
hydrometallurgy. Their adjacent positions in the general differences help to provide the basis
transition metal series in the periodic table re- for the various separation processes currently
sults in aqueous chemical behaviour that is too used or proposed for cobalt-nickel separation
similar for development of easy separation routes. in hydrometallurgy.
However differences in chemical behaviour do Tradition ally cobalt and nickel were sepa-
exist. For example, although both cobalt and rated by processes based on selective oxidation
nickel preferentially exist as divalent hexahy- and/or preci pitation of cobalt from either sul-
drated ions in dilute aqueous solution, the rate
phate or chloride solution and such processes
of water exchange on the cobalt ion is very
are still in use today. Indeed, new, improved
much higher than for nickel. Thus complex ion
oxidants are available. However, it is certainly
formation often proceeds much more readily
to the process of solvent extraction that one
with divalent cobalt than with nickel. On the
looks nowadays to provide the high degree of
other hand, the trivalent cobalt ion is much
separation and yields demanded by today’s in-
less labile and forms in preference to nickel
dustry and there can be no doubt of the im-
even though the redox potentials for the Co2+/
pact that solvent extraction has had and in-
Co3+ and Ni2+/Ni3+ couples are nearly identi-
deed is increasingly having in commercial op-
cal. Cobalt also in the divalent state exhibits
erations both existing and under development.
*Materials from the 2nd Intern ation al Conference Thus, alkylamines are the extractants of choice
«Metallurgy of Nonferrous and Rare Metals», for separation of cobalt from nickel from chlo-
Krasnoyarsk, September 9–12, 2003. ride liquors such as arise in the Eramet process
82 DOUGLAS S. FLETT

in France and in the Chlorine Leach Process as elemental sulphur or iron or cobalt sulphides,
operated by Falconbridge in Norway [1]. etc., were used at pH values between 1–5 at
For weakly acidic sulphate liquors the alkyl temperatures above 80 °C.
phosphorous acids have found significant com- Cobalt and nickel can also be separated by
mercial application at various locations around oxidative preci pitation. Although the Eh-pH
the world. The first plant to use this class of diagrams are very similar [5], nickel is much
reagent was at the Rustenburg Base Metal Re- more difficult to oxidise and cobalt can be ox-
finery of Anglo Platinum Ltd at Rustenburg idised quite selectively in the presence of rel-
in South Africa [2, 3] which uses DEHPA for atively large amounts of nickel. Strong oxi-
separation and recovery of cobalt. Because of dants are needed in practice, such as chlorine,
the high nickel to cobalt ratio encountered in ammonium persulphate, Caro’s acid or ozone,
liquors such as those produced in sulphate-based as the redox potential for the cobalt oxidation
high nickel matte leach processes or in the li- reaction is +1.75.
quors produced in the pressure acid leaching Air under pressure can also be used: this
of nickel laterites, very high separation fac- formed the basis for the so-called cobaltic am-
tors (>1000) are required. Only one commercial mine process for cobalt-nickel separation, (see
reagent offers such separation factors and that ref. [1, 23]).
is why CYANEX 272 has become the reagent The use of chlorine for cobalt removal from
of choice for such duties [4]. nickel solutions is practised by INCO [5] and
Much less progress, on the other hand, has Falconbridge in Can ada [5] and by the Jin-
been made on the reverse problem of the re- chuan Group Ltd in Chin a. Careful control of
covery and removal of nickel from cobalt li- pH is needed to optimise the process, but in-
quors although solvent extraction and ion ex- evitably a compromise is necessary between
change now can compete with tradition al pre- cobalt yield and nickel contamin ation.
ci pitative processes. The use of peroxygen compounds for co-
This paper reviews the chemistry of co- balt separation was first reported at the beginn-
balt-nickel separation from aqueous solutions ing of the century. Thus ammonium persul-
and comments on the implications of this chem- phate was used by AMAX at Port Nickel in
istry in hydrometallurgical applications. De- Louisian a [5] to replace the well-known Outo-
scri ption of selected applications is given and kumpu process using electrolytically generated
discussed. nickelic hydroxide.
The use of electrolytically generated nickelic
hydroxide for cobalt removal from an impure
PRECIPITATIVE SEPARATION nickel electrolyte was developed in Finland by
Outokumpu Oy [5]. The Ni(OH)3 is produced by
Separation of cobalt and nickel by preci pi- electrolytic oxidation of a black Ni(OH)2 slurry
tative processes has been and still is carried in cells with iron rods as cathodes and nickel
out commercially by a number of processes. plates as anodes. The Ni(OH)3 slurry is then
Thus sulphide preci pitation has been used com- mixed with the impure nickel electrolyte in
mercially, for example in the origin al flow- two stages to preci pitate Co(OH)3. This preci pi-
sheet at Queensland Nickel in Australia [5], in tate contains more nickel than cobalt and can
order to completely remove cobalt from the be redissolved using H2SO4 and SO2 and fur-
nickel process liquor. A slightly different ver- ther processed to produce pure cobalt. This proc-
sion based on Sherritt Gordon technology was ess is still used at some locations round the
practised at the Marinduque Nickel refinery in world, for example at the Rustenburg Base
the Phili ppines [5]. Metal Refinery in South Africa.
Sulphide preci pitation can also be used to The use of Caro’s acid, H2SO 5, has also
preci pitate nickel from cobalt-rich liquors. This been of interest for cobalt-nickel separation.
De Merre process [5] was used by Metallurgie Caro’s acid is prepared by direct addition of
Hoboken Overpelt (now Umicore) in Belgium. strong sulphuric acid to hydrogen peroxide.
Reagents such as metallic iron or cobalt, plus Recent work on Caro’s acid [6] has been con-
COBALT-NICKEL SEPARATION IN HUDROMETALLURGY 83

cerned with cobalt/nickel separation from liq- mine sulphate and produce a crystalline Co(III)
uors produced in hydrometallurgical studies on hexammine sulphate an alysing 15 % Co with
recycling NiCd batteries. a Co : Ni ratio in the range 50 : 1 to 100 : 1.
The use of ozone for cobalt oxidation and A single stage of recrystallisation of the cobalt
removal has also been advocated for cobalt- salt in ammonium sulphate elimin ates the re-
nickel separation and recent work in Japan sidual nickel and upgrades the cobalt hexam-
has been reported [7], although no commercial mine salt to a Co : Ni ratio of 2000 : 1. In the
applications are known. The rate of reaction process copper, zinc, cadmium and essentially
can be slow, however the long induction peri- all other significant metal impurities, except
od for cobalt preci pitation can be significantly chromium and iron, are elimin ated to very
shortened by addition of preci pitate seed. Co- low levels [10].
balt/nickel ratios in the product are said to be
as high as 1000 : 1 with careful operation [8].
The best separation is reported as being achieved SEPARATION BY RESIN ION EXCHANGE

between pH 2.5 and 4.0.


A comparison of the separation of cobalt Ion exchange separation of cobalt and nickel
from nickel with Caro’s acid and ozone has is most readily accomplished from chloride so-
been reported recently by Dunn et al. [9]. A lution where advantage can be taken of the
pilot campaign employing both oxidants with tendency for cobalt to form complex chloro
a Ni : Co ratio of approximately 100 : 1 in anions, i. e. CoCl3–, CoCl4–, which nickel does
the feed was carried out. The residual cobalt not. These complexes are quite weak, howev-
content achieved in a single contact with er, and relatively high concentrations of chlo-
ozone under steady state conditions (~1 ppm) ride ion are needed to produce the CoCl4– spe-
was very similar to that achieved with Caro’s cies, which is tetrahedral. Other ions forming
acid (10–15 ppm) in the continuously operat- chloro-complexes will interfere, in particular
ing pilot plant. Longer retention times with ferric iron, copper, zinc, etc. No commercial
a single contactor were required for Caro’s use is made of such an ion exchange process
acid (7 h) compared with approximately 1.25 h and solvent extraction is preferred.
for ozone. Although no great degree of selectivity bet-
Corefco, the nickel and cobalt refining arm ween Ni2+ and Co2+ is achievable by ordin ary
of the Metals Combined Enterprise, operates cation exchange resins, chelating ion exchang-
the nickel refinery in Fort Saskatchewan, Al- ers can offer separation opportunities. In par-
berta, Canada which was constructed by Sher- ticular, the chelating resins origin ally pro-
ritt Gordon Mines in 1954. Redevelopment of duced by the Dow Chemical Company known
the operating processes has led to a new process as XFS4195, XFS4196 and XFS43084 can re-
flowsheet and a new cobalt-nickel separation move nickel selectively from cobalt [11–14].
process. The ammonia pressure leach however The resins are based on a macroporous poly-
was retained. Thus the feed to cobalt-nickel sep- styrene divinylbenzene matrix, on which
aration is a solution containing cobaltic and nick- weakly basic chelating functional groups based
elous hexammine. About 70 % of the cobalt is on picolylamine (2-aminomethyl pyridine)
preci pitated from this solution by sparging in have been attached. The XFS resin shows
anhydrous ammonia to saturate the solution with significant selectivity for nickel over cobalt.
ammonia while simultaneously cooling the so- Commercial application of the Dow resin
lution to below 35 °C results in preci pitation of XFS 4195 took place at INCO’s Port Col-
[Co(NH3)6]2(SO4)3 ⋅ 2Ni(NH3)6SO4 ⋅ (NH4)2SO4 ⋅ xH2O, borne cobalt refinery for nickel removal from
a crystalline complex salt of Co(III) hexammi- the cobalt electrolyte [5]. Traces of copper
ne sulphate, Ni(II) hexammine sulphate and present in the electrolyte are also removed
ammonium sulphate. with the nickel. The resin is also used com-
After filtration this salt is repulped with mercially in Zambia at the cobalt plants at
water to selectively redissolve nickel hexam- Chambishi and Nkan a [15].
84 DOUGLAS S. FLETT

SEPARATION BY SOLVENT EXTRACTION phosphoric acids are in the tens, for alkyl phos-
phonics the hundreds while for alkyl phosphinics
The separation of cobalt and nickel by sol- they are in the thousands. This remarkable vari-
vent extraction has been studied quite inten- ation in separation factor is due to a change in
sively over the last 25 years or so. A useful the n ature of the cobalt complex in the or-
review of the extractants available for cobalt- ganic phase, whereby with increasing tem-
nickel separation from laterite leach liquors has perature and cobalt concentration, the pink
been given by Ritcey [16] and the chemistry hydrated/solvated octahedral complex changes
of cobalt-nickel separation, including solvent into the blue anhydrous/unsolvated tetrahe-
extraction, has been discussed by Flett [5]. At dral polymeric species with a consequent in-
present many commercial plants are operat- crease in distribution coefficient. No such be-
ing, most of which use the dialkyl phosphinic haviour is shown by nickel which remains in
acid extractant, CYANEX 272 (Cytec Indust- the hydrated/solvated octahedral form
ries Inc.). With one exception, these plants re- throughout. Specific separation factor values
move cobalt selectively from nickel, in cont- will also depend on the degree of steric hin-
rast to the resin ion exchange developments drance caused by the degree and location of
discussed above. Solvent extraction, unlike the branching of the alkyl chains in the extract-
preci pitation processes briefly described earli- ant molecule.
er, does offer the opportunity of complete The selectivity series also undergoes chang-
separation with high yields and purity of the es within the series phosphoric, phosphonic and
separated metals. The two main methods for phosphinic acids as shown below:
solvent extraction of cobalt and nickel are DEHPA Fe3+>Zn>Ca>Cu>Mg>Co>Ni
solvent extraction by anion exchangers and PC88A Fe3+>Zn>Cu>Ca>Co>Mg> Ni
solvent extraction by acidic chelating extrac- CYANEX 272 Fe3+>Zn>Cu>Co>Mg>Ca> Ni
tants. The relative position of calcium in these
For cobalt-nickel separation by anion ex- series is worth noting: for DEHPA and PC88A
change the same situation exists as for resin it is extracted before cobalt but for CYANEX
anion exchangers with the most important li- 272 cobalt is preferred over calcium and mag-
gand in the aqueous phase being chloride. The nesium. This is a significant advantage. Unfor-
extracted anionic species has been shown to be tun ately none of these extractants can ex-
CoCl4–. This chemistry is used commercially in tract cobalt selectively from ferric iron. How-
solvent extraction plants at Falconbridge Nikkel- ever, unlike DEHPA which requires 6 M HCl
verk in Norway, by Eramet in France and was to stri p any co-extracted iron, CYANEX 272
used by Sumitomo at Niihama in Japan [17]. can be readily stri pped with relatively dilute
Excellent separation factors in excess of 4000 (150 g/l) H 2SO 4. Cobalt can be selectively
are achieved by this means. stri pped from any co-extracted iron or zinc
For cation exchangers only the alkyl phos- at a pH of 2.5.
phoric, phosphonic and phosphinic acids show The first cobalt SX plant from sulphate solu-
selectivity for cobalt over nickel. All the rest, tion was at Rustenburg Refiners in South Afri-
i. e. carboxylic acids, β-diketones, 8-hydroxy- ca. This plant operates on a cobalt cake pro-
quinolines and hydroxyoximes, show margin al duced by preci pitation of cobalt from the main
selectivity for Ni(II) over Co(II). nickel electrolyte with nickelic hydroxide (the
Separation of cobalt from nickel in weakly Outokumpu process). Dissolution of this cake
acid sulphate solutions had tradition ally been gives a solution containing 2 : 1 to 4 : 1 Co : Ni
difficult until it was realised [18, 19] that, with which was easily treated by solvent extraction
alkyl phosphorous acids, the separation factor with DEHPA at 50 °C to achieve a cobalt re-
was a complex function of temperature, co- covery of >95 % at a cobalt to nickel ratio of
balt concentration, diluent, modifier and acid >500 : 1 [2]. The flowsheet is shown in Fig. 1.
type. The separation factor increases in the se- For less favourable Co : Ni ratios such as would
ries phosphoric < phosphonic < phosphinic ac- arise from leaching of nickel ores, be they
ids. In summary, separation factors for alkyl laterites or sulphides, DEHPA would not be an
COBALT-NICKEL SEPARATION IN HUDROMETALLURGY 85

Fig. 1. Flowsheet for nicel and cobalt circuits at Rustenburg Base Metal Refinery.

adequate extractant. The effect of separation ration in some small plants as described by
factor is well exemplified for disparate Co/Ni Koppiker [21].
ratios in extraction isotherms shown in Fig 2. It The development in the 1980s of the di(2,
should be noted that, when the Rustenburg 4,4-trimethylpentyl)phosphinic acid, CYANEX
plant was installed, CYANEX 272 was not 272, by American Cyan amid, now Cytec In-
available. dustries Inc., opened the way for direct sol-
Ni ppon Mining used PC88A (2-ethyl hexyl vent extraction of cobalt from liquors contain-
ester of phosphonic acid) for Co removal and ing very disparate Co/Ni ratios. There are
recovery from the Co/Ni solution produced by thought to be at least 13 plants operating com-
leaching the sulphide cake from Queensland mercially using CYANEX 272. Approximately
Nickel [20]. High removal of Co was necessary 50 % of the Western World’s cobalt is pro-
to minimise degradation of the hydroxyoxime duced via a CYANEX 272 plant.
extractant used later in the flowsheet caused The CYANEX 272 cobalt solvent extraction
by oxidative extraction of Co(II). PC88A or Ion- plant at Harjavalta, now owned by OMG, treats
quest 801 are also used in India for Co/Ni sepa- a feed from leaching of the mattes produced
86 DOUGLAS S. FLETT

fication. Thus any iron, aluminium and chro-


mium present in the leach liquor are removed
hydrolytically in a two step preci pitation to
yield a liquor at pH 4.2–4.5. Cobalt together
with the manganese and zinc present in the
liquor is then extracted with CYANEX 272.
The nickel in the raffin ate is then extracted
and separated from magnesium with a carboxy-
lic acid, Versatic 10 [26–28]. Results of conti-
nuous mini plant trials [28] showed that ext-
raction with CYANEX 272 can achieve 97.5 %
cobalt recovery and >99 % removal of Mn
and Zn with very good separation of Co and
Ni with Co : Ni ratios in the stri p of >1000 : 1.
Fig. 2. Extraction of cobalt from a nickel matte leach New thio-based extractants, bisdithiophos-
liquor using different alkyl phosphorous acids. Organic phoramides [29], were developed by Zeneca
phase: 10 v/o alkyl phosphorous acid in Escaid 110, plus in the UK. Origin ally developed for zinc ext-
5 v/o TBP, 85 % conversion to Na form. Feed: Co, 0.22 g/l;
Ni, 89.6 g/l.
raction from sulphate media under the deve-
lopment title of DS5869, modifications to the
in the DON smelting process [22] containing molecule provided a further development re-
130 g/l nickel, 0.8–1.0 g/l cobalt with very agent design ated DS6001 specifically for co-
minor amounts of zinc, copper, lead, manga- balt/nickel separation. This reagent could sep-
nese, magnesium, calcium and iron. Cobalt is arate both cobalt and nickel from manganese
extracted in four countercurrent stages, the and magnesium. CYANEX 301 and 302 also
loaded organic scrubbed with dilute sulphuric separate cobalt from manganese and magne-
acid in five stages and cobalt is stri pped with sium and this attribute together with its strong
sulphuric acid in four stages to produce a raffina- pH function ality has led to the selection of
te containing 130 g/l Ni, 0.01 g/l Co and a co- Cyanex 301 as the reagent of choice for INCO’s
balt strip liquor containing 110 g/l Co, 0.02 g/l
Ni, together with coextracted copper, lead,
manganese and some calcium. Co-extracted zinc
and iron are not significantly stri pped with
the cobalt and these metal ions are removed in
a single stage with 200 g/l H2SO4. The mixer-
settlers used are the Outokumpu developed
Vertical Smooth Flow (VSF) mixer-settlers [23].
The continuous countercurrent operation is con-
trolled using the Outokumpu Courier X-ray
system for on-line an alysis of cobalt and nick-
el in both aqueous and organic phases. Organic
phase: 10 v/o alkyl phosphorus acid in Escaid
110, plus 5 v/o TBP, 85 % conversion to Na
form. Feed: Co, 0.22 g/l; Ni, 89.6 g/l.
CYANEX 272 has also been adopted as the
reagent of choice for various laterite acid pres-
sure leach projects in Australia. Thus the Murrin
Murrin project (Fig. 3) [24–26] uses solvent ex-
traction with CYANEX 272 for Co/Ni separation
from a mixed sulphide pressure leach liquor.
The Bulong project (Fig. 4) uses solvent ex-
traction directly on the leach liquor after puri- Fig. 3. Murrin Murrin purification flowsheet.
COBALT-NICKEL SEPARATION IN HUDROMETALLURGY 87

Fig. 4. Bulong nickel/cobalt purification flowsheet.

Goro project in New Caledonia [30, 31]. DS6001 150 as the model diluent showed that in the
also extracted both cobalt and nickel at much cobalt/DEHPA system the rate of oxidation
lower pH values than either CYANEX 272 and increased with increasing diluent aromaticity,
302. However, while Zeneca reported good sep- temperature and cobalt solvent loading. Phe-
aration of both cobalt and nickel from manga- nolic antioxidants such as BHT were shown to
nese [29], this was not found in work reported be effective in conferring diluent stability. Other
by Lakefield [32] which shows a long tail on the extractants were studied, n amely PC88A and
manganese extraction curve below the pH at CYANEX 272. Diluent oxidation with PC88A
which nickel ceases to be extracted. The cause was found to be faster than with DEHPA but
of this discrepancy is not known. This Zeneca significantly slower with Cyanex 272. Manga-
reagent has been withdrawn. nese was found to oxidise Solvesso 150 just as
The stability of the organic phase in cobalt fast as cobalt in the DEHPA system. A further
extraction became an issue in the Rustenburg study of cobalt catalysed diluent oxidation in
Refiners cobalt solvent extraction plant when the CYANEX 272 system has been carried out
it became clear that oxidative degradation of by Rickelton et al. [34]. In this work tetrade-
the diluent to a carboxylic acid was taking place cane was used as the model diluent. The mecha-
causing a significant reduction in separation nism of oxidative degradation was suggested
factor and increasingly poor phase break. The to be from the alkane to the hydroperoxide to
problem was shown to be due to cobalt cata- the alcohol to the aldehyde and fin ally to the
lysed oxidative degradation of the diluent. carboxylic acid. Adoption of BHT as the anti-
A study of this problem [33] using Solvesso oxidant for addition in the CYANEX 272 com-
88 DOUGLAS S. FLETT

Fig. 5. Flow sheet of the QNI Yabulu Plant.

mercial plants appears standard practice now nickel could be selectively separated from co-
at levels of 0.5–1.0 g/l [35]. balt. No commercialisation of this system has
The selective removal of nickel from co- taken place not least because there were prob-
balt has long been of interest. While, as noted lems with poor phase disengagement.
above, all extractants other than the phospho- From ammoniacal solutions however, pro-
rus acids and the dithiophosphoramides extract vided cobalt is in the Co(III) state, nickel can
nickel in preference to cobalt, the separation be successfully separated from cobalt with hy-
factors are not large. Thus other systems have droxyoximes as Co(III) is not extracted by these
been sought. As long ago as 1983 Grinstead and reagents. This has been successfully commer-
Tsang [36] showed that a mixture of an N-al- cialised by Queensland Nickel at their Yabulu
kylated bispicolylamine and dinonyl n aphtha- refinery in Queensland, Australia (Fig. 5). Rea-
lene sulphonic acid could extract both nickel gent screening showed that the best reagent
and cobalt selectively from ferric iron and that mixture was a modified LIX 84 in Escaid 110.
COBALT-NICKEL SEPARATION IN HUDROMETALLURGY 89

The main leach liquor is treated directly by cobalt ratio effect comes into play here just as
solvent extraction after a preboil to reduce in the aqueous phase work, although tests with
ammonia levels. Small amounts of Co(II) are organic phases loaded with cobalt only show
oxidatively extracted which cannot be stri pped no significant reaction under experimental con-
convention ally. Nickel stri pping was with ditions. Nickel reduction, on the other hand,
280 g/l NH3 to give a nickel concentration in was rapid and complete in 30 min. This inte-
the stri p of 80 g/l and <20 mg/l cobalt. Nick- resting approach has not been developed be-
el carbon ate is produced from the stri p liq- yond the laboratory.
uor which is worked up to nickel oxide or, if Nickel powder is also precipitated selectively
the calciner is operated under reducing con- by reduction of aqueous solutions containing
ditions, a product containing >97 % nickel nickel and cobalt ammines in concentrated
metal. Cobalt oxidation is accompanied by deg- ammonium sulphate solution at around 240 °C
radation of the hydroxyoxime to a ketone, with hydrogen gas at a total pressure of up to
but this can be reversed by re-oximation with 3103 kPa. When the concentration of nickel in
an aqueous ammoniacal solution of hydroxy- solution is lowered to around that of cobalt,
lamine sulphate [37]. the reaction is stopped and the solution dis-
This approach was also used at the origin al charged from the autoclave leaving nickel pow-
Cawse plant in Western Australia although the der inside [1]. This process, origin ally deve-
nickel here was acid stri pped and nickel re- loped by the Chemical Construction Corpora-
covered by electrowinning. This refinery was tion of America and further developed by
closed recently after takeover by OMG. The Sherritt Gordon in Can ada is used commer-
plant now produces mixed Co/Ni hydroxides cially, for example, at Fort Saskatchewan in
for shi pment to Harjavalta. Can ada, by Impala Platinum at Springs in
Separation of nickel from cobalt is also pos- South Africa and at Murrin Murrin in Western
sible with mixtures of carboxylic acids or alkyl Australia albeit here after Co/Ni separation by
phosphorous acids with pyridine carboxylic es- solvent extraction. Currently however, in the
ters [37, 38]. Good separation of Ni from Co is new Corefco plant at Fort Saskatchewan the
achieved with the former mixture, less good nickel reduction is carried out on a solution
with the latter which also has a major draw- after Ni/Co separation by selective precipita-
back in the very strong extraction of ferric tion of Co(III) hexammine [10].
iron which would require elimin ation before
nickel solvent extraction.
DISCUSSION AND CONCLUSIONS

SEPARATION BY PRESSURE HYDROGEN REDUCTION Because of the in ability of preci pitation


processes to produce high quality cobalt pro-
Separation of nickel and cobalt is possible ducts directly it is small wonder that solvent
by direct hydrogen reduction of nickel + co- extraction has attracted so much attention over
balt loaded DEHPA solutions [39]. Just as it is the years, offering, as it does, a one step ap-
possible to recover nickel selectively from aque- proach to achieving a very high degree of sepa-
ous solutions by direct hydrogen reduction in ration of cobalt from nickel with high yields
the presence of cobalt, so nickel can be selec- of both metals with low levels of contamin a-
tively reduced in the presence of cobalt from tion of each metal in the respective cobalt and
a metal-loaded DEHPA phase in an autoclave nickel streams. The first breakthrough in this
at 140 °C and an initial pressure of 120 atm. It respect was the chloride-based processes oper-
is reported that a solution containing 24 g/l Ni ated by Falconbridge and Eramet.
and 1.2 g/l Co could be reduced to produce a Application of solvent extraction for cobalt-
nickel powder containing less than 0.15 % Co nickel separation from weakly acidic sulphate
(the limit of the an alytical method used) and solutions really did not take off until the de-
a fin al organic phase containing 3.5 g/l Ni and velopment of CYANEX 272. This reagent has
1.2 g/l Co. However it is clear that the nickel to transformed the Co/Ni separation process in
90 DOUGLAS S. FLETT

weakly acidic sulphate solutions, particularly cobalt from manganese as well as the alkaline
for high Ni : Co ratio liquors. However there is earth elements which is certainly of interest
no standard set of operating conditions as the in the treatment of the leach liquors arising
objectives set for such solvent extraction plants from the pressure acid leach process for nickel
varies from plant to plant. For example, at laterites. However CYANEX 302, which does
Bulong, it is necessary to ensure minimum co- separate cobalt from nickel as well as, if not
balt in the nickel liquor going forward to Ver- better than, CYANEX 272, irreversibly de-
satic acid extraction and subsequently elect- composes to CYANEX 272 and elemental sul-
rowinning, whereas, at Murrin Murrin some phur in the presence of even minor amounts
cobalt in the nickel going forward to pressure of ferric iron. CYANEX 301, on the other hand,
hydrogen reduction is tolerable because of the decomposes in two stages, the first of which is
degree of selectivity found for nickel in this reversible. That this reagent has been chosen
reduction step. Rather it is the requirement for for Inco’s Goro project in New Caledonia, stems
minimal nickel in the cobalt liquor proceeding not from its ability to separate Co from Ni but
to cobalt pressure hydrogen reduction that is rather from its ability to bulk extract both co-
the main requirement. balt and nickel selectively from Mn, Ca and
On the other hand, separation of nickel Mg from the acid leach liquor produced in the
selectively from cobalt remains elusive except pressure leach process after removal of iron.
for the chelating resin ion exchange process for Both reagents also extract copper in a redox
removal of small amounts of nickel from rela- process which also causes degradation of these
tively rich cobalt streams as at Inco’s Port Col- reagents and so copper must be eliminated prior
borne operations in Can ada, at the cobalt re- to cobalt solvent extraction. This will be done
fineries at Nkan a and Chambishi in Zambia by use of a chelating ion exchange resin in the
and at the QNI SX plant in Australia. Little Goro project [31]. Such decompositions and the
research effort appears to be on-going in this oxidation of cobalt on extraction with di(2-
area currently and there appears little incen- ethyl hexyl)dithiophosphoric acid (DTPA) have
tive for such work. Adoption of solvent ext- been studied by V. I. Kuz’min et al. [42, 43]
raction to provide the interface between the who have shown that the irreversible decom-
purified cobalt liquors arising in the cobalt re- position of DTPA occurs by its direct water
fineries in Zambia and the Democratic Repub- hydrolysis and decomposition of the disulphide
lic of Congo and cobalt electrowinning, as en- which is formed as a result of the reversible
visaged by Burks [40], should avoid the need to redox reaction with cations of transition me-
use this expensive resin ion exchange process tals such as Cu, Co and Fe. The results of this
and developments here are keenly awaited. work are of direct relevance to the decompo-
Molecular recognition technology (MRT) has sition issue of CYANEX 301.
been promoted as being of great potential for While it has been successfully demonstrated
selective recovery of cobalt and extensive that the decomposition of CYANEX 301 can be
trials have been carried out using a skid-mounted contained and, indeed, reversed [31], the same
unit in Australia [41] and on the Zambian Cop- cannot be said for CYANEX 302. What would be
per Belt, for example. While it is believed that useful here would be the development of some
technically the trials were successful, the eco- means of retarding the rate of decomposition
nomics were unfavourable, not least because of CYANEX 302 in order to render it accepta-
of the very high replacement rate requirement ble in terms of solvent loss in operating condi-
for the very expensive MRT material. tions. Unfortunately no such development work
Thus it is concluded that it is unlikely that appears to be on-going at this time.
any radical, new methods for the separation
of cobalt from nickel are likely to emerge in
REFERENCES
the immediate future. The two thio an alogues
of CYANEX 272, n amely CYANEX 301 (the
1 A. R. Burkin, in A. R. Burkin (Ed.), Critical Reports in
dithio an alogue) and CYANEX 302 (the mono- Applied Chemistry, vol. 17, John Wiley & Sons, 1987,
thio an alogue) do have the ability to separate p. 98.
COBALT-NICKEL SEPARATION IN HUDROMETALLURGY 91

2 D. De J. Clemente, B. I. Dewar and J. Hill, Paper pre- 24 G. Motteram, M. Ryan, R. Berezowsky and R. Raud-
sented at CIM 10th Annual Hydrometallurgical Meet- sepp, Nickel/Cobalt Pressure Leaching and Hydro-
ing, Edmonton, Alberta, Can ada, 1980. metallurgy Forum, Alta Metallurgical Services, Aus-
3 Z. Hofirek and P. J. Nofal, Hydrometallurgy, 39 (1995) 91. tralia, 1996.
4 W. A. Rickelton and D. Nucciarone, in W. C. Cooper 25 G. Motteram, M. Ryan and R. Weizenbach, in W. C. Coo-
and I. Mihaylov (Eds.), Nickel–Cobalt’97, vol. 1: Hy- per and I. Mihaylov (Eds.), Nickel–Cobalt’97, vol. 1:
drometallurgy and Refining of Nickel and Cobalt, Hydrometallurgy and Refining of Nickel and Cobalt,
The Metallurgical Society of CIM, 1997, p. 275. The Metallurgical Society of CIM, 1997, p. 391.
5 D. S. Flett, in A. R.Burkin (Ed.), Critical Reports in 26 K. C. Sole and P. Cole, Ion Exchange and Solvent
Applied Chemistry, vol. 17, John Wiley & Sons, 1987, Extraction, 15 (2001) 143.
p. 76 and references therein. 27. G. M. Ritcey, N. L. Hayward and T. Salinovich, Aus-
6 P. J. Wyborn and C. F. McDon agh, Minerals, Metals tralian provision al patent, PN0441, 9 January, 1995.
and the Environment II, IMM, London, 1996, p. 421. 28 K. Soldenhoff, N. Hayward and D. Wilkins, in B. Mish-
7 T. Nishimura and Y. Umetsu, Hydrometallurgy, ra (Ed.), EPD Congress 1998, The Minerals, Metals
30 (1992) 483. & Materials Soc., 1998, p. 153.
8 J. R. Boldt and P. Queneau, The Winning of Nickel, 29 J. R. Lawson, and M. A. Cox, in W. C. Cooper and
Methuen, London, 1967, p. 372. I. Mihaylov (Eds.), Nickel–Cobalt’97, vol. 1: Hydromet-
9 G. M. Dunn, H. W. Schubert and H. E. Holliday, in allurgy and Refining of Nickel and Cobalt, The Met-
W. C. Cooper and I. Mihaylov (Eds.), Nickel–Cobalt’97, allurgical Society of CIM, 1997, p. 319.
vol. 1: Hydrometallurgy and Refining of Nickel and 30 I. O. Mihaylov, E. Krause, S. W. Laundry and
Cobalt, The Metallurgical Society of CIM, 1997, p. 195. C. V. Luong, US Pat. 5 378 262, Jan. 3, 1995.
10 D. G. E. Kerfoot and P. D. Cordingley, in W. C. Cooper 31 I. Mihaylov, E. Krause, D. F. Colton et al., CIM Bull.,
and I. Mihaylov (Eds.), Ibid., p. 355. 93 (2000) 124.
11 K. C. Jones and R. M. Wheaton, US Pat. 3 998 924, 32 C. J. Ferron, R. Moln ar and R. G. Williamson, in
Dec. 21, 1976. W. C. Cooper and I. Mihaylov (Eds.), Nickel–Cobalt’97,
12 R. R. Grinstead and W. A. Nasutavicus, US Pat. 4 031 038, vol. 1: Hydrometallurgy and Refining of Nickel and
June 21, 1977. Cobalt, The Metallurgical Society of CIM, 1997,
13 R. R. Grinstead and W. A. Nasutavicus, US Pat. 4 098 867, p. 427.
July 4, 1978. 33 D. S. Flett and D. W. West, Proc. Int. Solvent Extraction
14 R. R. Grinstead and A. L. Tsang, in: Int. Solvent Ext- Conf. ISEC’86, Preprints, vol. II, 1986, p. 3.
raction Conf., Denver, Colorado, 1983, American Inst. 34 W. A. Rickelton, A. J. Robertson and J. H. Hillhouse,
of Chem. Engrs., New York, 1983, p. 230. Solvent Extraction and Ion Exchange, 9 (1991) 73.
15 A. S. Rao, R. Mingango, J. Nkhoma and H. P. Singh, in 35 S. F. Burks, in Nickel/Cobalt Pressure Leaching and
R. G. Reddy and R. N. Weizenbach (Eds.), The Paul Hydrometallurgy Forum, Alta Metallurgical Servic-
Queneau Int. Symp. Extractive Metallurgy of Copper, es, Australia, 1996.
Nickel and Cobalt, vol. 1: Fundamental Aspects, The 36 I. G. Skepper and J. E. Fittock, in D. C. Shallcross,
Minerals, Metals & Materials Soc., 1993, p. 853. R. Paimin and L. M. Prvcic (Eds.), Value Adding
16 G. M. Ritcey, in: Nickel’96, The Australasian Institute through Solvent Extraction, Proc. ISEC’96, vol. 1, The
of Mining and Metallurgy, 1996, p. 251. University of Melbourne, 1996, p. 777.
17 M. Anthony and D. S. Flett, Minerals Industry In- 37 J. S. Preston and A. C. du Preez, Solvent Extraction
tern., January (1997) 26. and Ion Exchange, 13 (1995) 465.
18 D. S. Flett and D. W. West, in: Complex Metallur- 38 J. S. Preston and A. C. du Preez, J. Chem. Technol.
gy’78, Institution of Mining and Metallurgy, Lon- Biotechnol., 66 (1996) 86.
don, 1978, p. 49. 39 A. R. Burkin and J. E. A. Burgess, Het Ingeniersblad,
19 J. S. Preston, Hydrometallurgy, 9 (1982) 115. 41 (1972) 459.
20 M. Ando, M. Takahashi and T. Ogata, in: Hydrome- 40 S. F. Burks, in Nickel/Cobalt Pressure Leaching and
tallurgy, Research, Development and Plant Practice, Hydrometallurgy Forum, Alta Metallurgical Servic-
Symposium, Atlanta, AIME, New York, 1983, p. 463. es, Australia, 1996.
21 K. S. Koppiker, Cobalt News, 2 (2001) 9. 41 S. R. Izatt, R .L. Bruening, N. E. Izatt et al., in: Alta
22 I. V. Kojo, T. Mäkinen and P. Hanniala, in C. Díaz, 2000 SX/IX-1, Conf. Tech. Proc., Day 1, Session 4,
I. Holubec and C. G. Tan (Eds.), Nickel–Cobalt’97, vol. Alta Metallurgical Services, Blackburn South, Victo-
III: Pyrometallurgical Operations, the Environment and ria, Australia, 2000.
Vessel Integrity in Nonferrous Smelting and Convert- 42 V. I. Kuz’min, T. A. Klimkin a and O. A. Logutenko,
ing, The Metallurgical Soc. CIM, 1997, p. 25. Russian J. Inorg. Chemistry, 43 (1998) 798.
23 B. Nyman, A. Aaltonen, S.-E. Hultholm and K. Kar- 43 V. I. Kuz’min, O. A. Logutenko and V. N. Kuz’min a,
pale, Hydrometallurgy, 29 (1992) 461. Ibid., 44 (1998) 1155.