effect of tartrate ion on extraction behavior of ni and co via d2ehpa in sulfate media

Minerals Engineering 69 (2014) 177–184

Contents lists available at ScienceDirect

Minerals Engineering

journal homepage: www.elsevier .com/locate /mineng

Effect of tartrate ion on extraction behavior of Ni and Co via D2EHPAin sulfate media

http://dx.doi.org/10.1016/j.mineng.2014.08.0080892-6875/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +98 21 6454 2971; fax: +98 21 6640 5846.E-mail addresses: dahafa@aut.ac.ir, davoudhaghshenas@gmail.com

(D.H. Fatmehsari).

Hamed Nadimi, Amirmostafa Amirjani, Davoud Haghshenas Fatmehsari ⇑, Sadegh Firoozi,Amirreza AzadmehrDepartment of Mining and Metallurgical Engineering, Amirkabir University of Technology, P.O. Box 15875-4413, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history:Received 20 March 2014Accepted 14 August 2014Available online 16 September 2014

Keywords:NickelCobaltD2EHPASolvent extractionTartrate ionResponse surface methodology

Development of a solvent extraction process with a single solvent for separating Co and No has theadvantages of succinct process and recyclability of components. Main problem with the known specificorganic extractants are they are expensive and have limited availability. Previous research on Co/Ni sol-vent extraction systems have employed two main strategies; modification of organic phase and adjustingthe conditions of aqueous phase. The aim of present work was to study the effect of tartrate ion, a car-boxylate ligand, in the separation conditions of Co and Ni via D2EHPA. Preliminary experiments wereconducted to examine the effect of tartrate ion on the extraction curves of Co and Ni. Fourier TransformInfrared Spectroscopy (FT-IR) was employed to elucidate whether metal–organic complexes containingCo/Ni and tartrate were formed. For the estimation of the stoichiometric coefficients, slope analysismethod was applied. Response surface methodology (RSM) with a central composite design (CCD) wasused to quantify the effect of pH, tartrate concentration and temperature on the extraction process ofCo and Ni and also to develop second order polynomial models for optimization purposes.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction successfully separated Cu, Ni, Co and Zn from chloride solutions

Separation of Co and Ni has always been problematic due totheir similar physicochemical properties (Xing et al., 2012;Darvishi et al., 2005). Solvent extraction has been shown to be anefficient method for separation and recovery of metal ions, whichhas been the focus of much research. This technique has the advan-tages of being a succinct process, low cost and recyclability of com-ponents (Keshavarz Alamdari et al., 2012; Xing et al., 2012). In thecase of Co and Ni, Cyanex organic extractants are the most effectivefor the separation purposes (Flett, 2004; Tsakiridis and Agatzini,2004); however, researchers are trying to improve solvent extrac-tion based methods with the aim of partially/entirely avoiding theuse of Cyanex extractants due to their relatively high cost as wellas their limited availability (Okewole et al., 2012; Zhu et al.,2012; Cheng, 2006; Darvishi et al., 2005).

One common strategy is the use of a combination of organicextractants resulting in the synergistic effect for better Co/Ni sep-aration efficiency. For instance, Cheng (2006) proposed a synergis-tic SX system, LIX63 with Versatic 10 acid, for the separation andrecovery of Ni and Co from leach solutions. Zhu et al. (2012)

employing a synergistic solvent extraction system consisting ofLIX63, Versatic 10 and TBP. Also, it has been reported the separa-tion of Ni from Co is achieved by using 1-octyl-2-(20-pyridyl) imid-azole (OPIM) together with dinonylnaphthalene sulfonic acid(DNNSA) as the synergist (Okewole et al., 2012). The applicationof new extractant is another applicable strategy. Xing et al.(2012) have introduced a new synthesized phosphinic acid, di-decylphosphinic acid (DDPA) extractant for separating Co fromNi. Quaternary ammonium salt, namely Aliquat 336 chloride, hasbeen studied for the extraction and separation of Co and Ni fromsulfate solutions (Nayl, 2010). The difference in the extractionbehaviors of Co and Ni was attributed to the poor extractabilityof the Ni complex compared with that of Co and their differentrates of extractions.

The above-mentioned studies show that the main approach forthe Co/Ni separation is focused on the organic phase modifications.However, another important approach for making differencebetween the adsorption behaviors of Co or Ni towards a givenorganic extractant can be found in the conditioning the aqueousphase. Acetate ion, which belongs to carboxylate groups, is a ligandwith lone pair of electrons, which the ion can share with metallicions to form a complex. Carboxylate ligands contain CH3COO� inwhich the negative charge is shared equally by the two oxygenatoms. In this regard, Gu et al. (1986) introduced acetate anion

178 H. Nadimi et al. / Minerals Engineering 69 (2014) 177–184

ligands into the aqueous phase containing Co, Cu and Ni. Theyshowed that the presence of acetate ions leads to the formation ofligand–metal complexes, which improves the separation efficiencyin the liquid membrane system composed of D2EHPA. Later, Van deVoorde et al. (2005) studied the effect of sodium acetate on therecovery of Cu, Ni, Co, Mg and Fe using different organic phases.They reported a higher extraction percentage of both Ni and Co inthe presence of acetate ions, which was attributed to the bufferingnature of sodium acetate. On the other hand, Ren et al. (2007) statedthat the distribution coefficient of Cu, in the ‘‘Cu/sodium acetate–D2EHPA/kerosene’’ system, significantly changes due to the differ-ent extraction mechanisms of various ligand–metal complexes.Recently, Guezzen and Didi (2012) have investigated Zn extractionfrom acetate solution with D2EHPA and found that Zn–acetatecomplexes (ZnCH3COO+) are absorbed to the organic phase.

A survey of previous literature provides limited clues on theeffect of carboxylate ligands on the separation of Co and Ni via sol-vent extraction. Additionally, an answer to the question of how theligand affects the mechanism of a metallic ion extraction cannot beclearly found from the references. This might be due to the factthat in previous research, the main focus was on the bufferingeffect of the carboxylate ligands (Begum et al., 2012; Van deVoorde et al., 2005). Another important point is the effect of perti-nent factors in a solvent extraction system, which can be correctlyidentified by employing an efficient methodology. Factorial designof experiments (DOE) can simultaneously consider several factorsand also, give a suitable statistical model to evaluate the changein the factors and their responses (Montgomery, 2006).

Tartrate dianion, another well-known carboxylate ligand, canalso form complexes with metals by virtue of their carboxylicand alcoholic groups. As well, tartrate salts such as sodium tartratecan act as a buffering agent to prevent a rapid change in pH. Theaim of this study was to evaluate the effect of tartrate ion on theextraction behavior of both Ni and Co by D2EHPA diluted in kero-sene. Initial experiments examined the change in tartrate ion con-centration at different equilibrium pH to hinder the buffering effectof sodium tartrate and to ascertain the role of tartrate ligand in theextraction of Co and Ni. Fourier Transform Infrared Spectroscopy(FT-IR) was utilized to examine the metal–organic complexes con-taining Co/Ni and tartrate. The slope analysis method wasemployed to estimate the stoichiometric coefficients of theextractant for both Co and Ni. Finally, response surface methodol-ogy (RSM) was employed to study the effect of important factorslike pH, tartrate concentration and temperature, in the solventextraction process of Co and Ni. A central composite design(CCD) was chosen for developing a second order polynomial model,which was used to optimize the process.

Table 1Central composite design arrangement.

Number pH Tartrate ion concentration (M) T (�C)

1 2.5 0.1 202 2.5 0.1 603 2.5 0.1 204 2.5 0.4 605 5 0.1 206 5 0.1 607 5 0.4 208 5 0.4 609 3.75 0.25 20

10 3.75 0.25 6011 3.75 0.1 40

2. Materials and methods

2.1. Materials

Commercial bis-2-ethylhexyl phosphoric acid (D2EHPA), as theorganic extractant was purchased from Sandong Chemical,Chengdu China. Kerosene, as the diluent, and sodium tartrate(C4H4Na2O6) were obtained from Tehran Refinery Company (Iran)and Merck Chemicals (Germany), respectively. Aqueous solutionsof 5 g/L Co and Ni were prepared by dissolving metal sulfates indistilled water; and NH4OH and H2SO4 were used for pH adjust-ment to the desired pH.

12 3.75 0.4 4013 2.5 0.25 4014 5 0.25 4015 3.75 0.25 4018 3.75 0.25 4017 3.75 0.25 40

2.2. Experiments

Batch experiments were conducted in a flask containing equalvolumes (40 mL; Vorg/Vaq = 1) of aqueous and organic phases. Initial

concentrations of both Co and Ni in the aqueous phase were 5 g/L.The mixture was mechanically agitated for 60 min to assure equi-librium conditions (Darvishi et al., 2005). Then, the agitated sam-ples were separated in a flask. After the separation, the aqueousphase was analyzed to determine the metallic ion concentrationin solution.

2.3. Analysis

The concentrations of Ni and Co in all aqueous solutions weredetermined by atomic absorption spectrometry (AAS). Metalliccontents of the organic phase were calculated through mass bal-ance calculations. Fourier Transform Infrared Spectroscopy (FT-IR) was carried out with a Unicam FT-IR Spectrometer (Mattson1000 model) using NaCl Windows.

2.4. Design of experiments

A central composite design (CCD) was adopted to study threefactors (pH, tartrate ion concentration and temperature) at threelevels. Seventeen experimental runs were generated by the princi-ple of RSM using MINITAB Release 15. The levels employed for thedifferent factors, according to CCD design are listed in Table 1. Thequadratic polynomial regression model (Eq. (1)) was chosen fordeveloping the statistical model between the response variable interms of the three independent variables:

Y ¼ b0 þX3

i¼1

biXi þX3

i¼1

biiX2i þ

X2

i¼1

X3

j¼iþ1

bijXiXj ð1Þ

In Eq. (1) Y is the response variable (Co and Ni extraction per-centages), b0, bi, bii, and bij are the coefficients of the intercept, lin-ear, quadratic and interaction terms, respectively, and Xi and Xj

represent the three independent variables (pH, tartrate concentra-tion and temperature). The experiments were carried out with tworeplicates and conducted in a randomized order to avoid system-atic bias.

The statistical significance of the full quadratic models pre-dicted was evaluated by the analysis of variance (ANOVA). Unlessotherwise stated, the significance level employed in the analysiswas 90%. Finally, the model was used to assess both the optimumregion of the level of the factors which results in maximum orfairly high separation factor. All the analysis was carried out usingMINITAB Release 15.

Fig. 2. Percentages of Ni and Co extracted by D2EHPA (20 vol.%) in the presence oftartrate ion (0.3 and 0.4 M) at 25 �C.

H. Nadimi et al. / Minerals Engineering 69 (2014) 177–184 179

3. Results and discussion

3.1. Effect of tartrate ion on the extraction behavior of Co and Ni

In order to examine the effect of sodium tartrate on the extrac-tion percentages of Co and Ni via D2EHPA (20 vol.%) diluted in ker-osene, preliminary experiments were conducted by the addition ofsodium tartrate into the aqueous phase at different concentrations(0, 0.2, 0.3 and 0.4 M). The equilibrium pH of the system wasadjusted by the use of H2SO4 and NH4OH.

Fig. 1 shows the extraction percentages of Co and Ni versus pHat different concentrations of tartrate ion (0 and 0.2 M) underambient conditions. According to the results presented in Fig. 1,the addition of sodium tartrate leads to a shift in the extractioncurves of Co and Ni towards higher pH; in the case of Ni, this shiftis more significant compared to Co.

The effect of tartrate ion at higher concentrations (0.3 and0.4 M) on the extraction of Co and Ni are depicted in Fig. 2. Itshould be pointed out that, high concentration of tartrate ion andpH are favorable conditions for Co/Ni precipitation being problem-atic phenomenon at pH higher than 5.5.

Based on the extraction curves shown in Fig. 2, the higher con-centration of tartrate ion leads to continual shift in the extractioncurves of Co and Ni towards higher pH compared to the curve cor-responding to 0.2 M tartrate ion concentration (Fig. 1). The Coextraction curves for the solutions containing 0.3 M and 0.4 M tar-trate ion were virtually identical. However, the extraction curvesfor Ni were practically identical below pH 4.7 under the sameconditions.

The other important point is the difference in the extractionbehavior of Co and Ni by D2EHPA, which is enhanced in the pres-ence of tartrate ion. For example, in the presence of tartrate ionwith the concentration of 0.2 M the value of DpH0.5

Ni–Co (the differ-ence in pH values corresponding to 50% Ni extraction and 50% Coextraction) reached �0.3 being approximately three times higherthan that obtained in the absence of tartrate ion (�0.1). However,when the concentration of tartrate ion in the aqueous solution is0.4 M the extraction curve of Ni does not reach 50% even atpH = 5.5 due to the formation of precipitates under theseconditions. Thus, it was decided to consider the difference in pHvalues corresponding different level of extraction lower than 50(DpHNi–Co). The results of these calculations are listed in Table 2.

Table 2 reveals that the highest DpHNi–Co has a value of �1when the concentration of tartrate ion is 0.4 M at the extractionpercentage of 30%. Furthermore, the value DpHNi–Co significantly

Fig. 1. Percentage of Ni and Co extracted by D2EHPA (20 vol.%) in the absence andpresence of tartrate ion (0.2 M) at 25 �C.

increases when tartrate ion concentration is higher than 0.2 M. Acomparison between the results reported by Darvishi et al.(2005) and those obtained in the present study shows that theaddition of tartrate ion (0.3 M) results in a DpH0.5

Ni–Co value similarto what obtained by using a mixture of D2EHPA and Cyanex 302under the same conditions.

Another factor when introducing sodium tartrate into the aque-ous phase is to see whether Na ion is co-extracted with Co and Ni.Hano et al. (1992) reported that that organophosphorus extract-ants such as D2EHPA and MEHPA have a higher affinity towarddivalent cations in comparison with monovalent cations. Also,there are several works in which the adsorption of Co or Ni hasbeen carried out employing sodium salt of D2EHPA, (Mohapatraet al., 2007; Reddy and Bhaskara Sarma, 2001; Devi et al., 2000)which supports this higher affinity. Thus, it seems that the pres-ence of Na ion in the aqueous phase has no significant effect onthe Co and Ni extractions by D2EHPA.

In Conclusion, based on the above discussion, sodium tartrateaddition into aqueous phase can improve the separation of Coand Ni by D2EHPA, which is due to the role of tartrate ion in com-plex formation with Co and Ni.

3.2. FT-IR spectroscopy analysis

In order to evaluate the role of tartrate ion in the extractionreactions of Co and Ni, a set of FT-IR measurements was carriedout. The bonding of metallic ions with the organophosphorousextractants can be revealed by comparing the FT-IR spectra ofthe extractant diluted in the kerosene with those obtained for Coand Ni loaded extractant in the presence and absence of tartrateion.

The characteristic vibrational bands of D2EHPA correspondingto P@O, PAOAC or PAOAH and OH appear at 1230, 1034 and1690 cm�1, respectively. The bands related to the CAH stretchare also at 2923 and 2861 cm�1. The CAH deformation vibrationband has the frequencies of 1469 and 1384 cm�1 which is due tothe presence of more than one CH3 group on a carbon atom(Darvishi et al., 2005).

In Fig. 3 the FT-IR spectrum of the organic phase (D2EHPA20 vol.% diluted in kerosene) is compared with the same organicphase which kept in contact with an aqueous phase containing0.4 M tartrate ion. As it can be observed, there is no significant dif-ference between the spectra presented in Fig. 3. This means thattartrate ion has no effect on the spectra corresponding to D2EHPA20 vol.% diluted in kerosene under these conditions.

Table 2The values of DpHNi–Co at different extraction percentages; T = 25 �C.

Extraction percentage Tartrate ion concentration = 0.3 M Tartrate ion concentration = 0.4 M

pHNi pHCo DpHNi–Co pHNi pHCo DpHNi–Co

10 3.50 2.75 0.75 3.70 2.80 0.9020 4.20 3.35 0.85 4.25 3.35 0.9030 4.60 3.70 0.90 4.80 3.80 1.0040 4.95 4.15 0.80 – 4.15 –50 5.40 4.50 0.90 – 4.55 –

Fig. 3. A comparison between the FT-IR spectra of D2EHPA (20 vol.%) with the sameorganic phase which kept in contact with an aqueous phase containing 0.4 Mtartrate ion.

Fig. 4. A comparison between the FT-IR spectra of D2EHPA (20 vol.%) loaded with(a) Co and Co–tartrate and (b) Ni and Ni–tartrate; T = 25 �C, Co and Ni extractionpercentage = �45 corresponding to the pH values of 4.3 and 3.1 (for cobalt) and 5and 3.2 (for nickel) in the presence and absence of 0.3 M tartrate, respectively.

180 H. Nadimi et al. / Minerals Engineering 69 (2014) 177–184

The FT-IR spectra corresponding to Ni and Co loaded D2EHPAare depicted in Fig. 4 in the presence and absence of tartrate ionin the aqueous phase. After metal binding the strong band at1034 cm�1, assigned to the OH, becomes weak indicating the coor-dination of D2EHPA with the metallic ions via deprotonationmechanisms. Moreover, there is a shift of the P@O from 1230 to1203 cm�1 signifying the coordination of D2EHPA with the metal-lic ions (Sainz-Diaz et al., 1996). The same variations can beobserved for the spectra obtained in the presence of tartrate ion;however, there are some additional differences. The appearanceof a slight band at �1620 cm�1 (which is ascribed to C@O banddue to the absorption of tartrate ion) together with the changesin CAH stretch and deformation vibration bands indicates the co-extraction of tartrate ion during metallic ions adsorption byD2EHPA.

With the help of the assignment made above, it is most likelythat the Co/Ni–tartrate cationic complexes are absorbed when tar-trate ion is present in the aqueous phase.

3.3. Extraction reaction stoichiometry

According to the results and evidences presented in previoussections, it is most likely that the tartrate ion participates in theextraction reaction of both Co and Ni by D2EHPA. The metal–tar-trate complex formed in the aqueous phase can be considered byfollowing reaction:

M2þaq þ a tar2�

aq ¢ ½MðtarÞa�2�2aaq ð2Þ

where Maq2+, taraq

2� and [M(tar)a]aq2�2a are the metallic cations,

C4H4O62� (tartrate anion) and metal–tartrate complex present in

the aqueous phase, respectively. Also, a is the stoichiometric coeffi-cient of tartrate ion. Since D2EHPA can only adsorb cationic species,

the extractable metal–tartrate complex should possess positivecharge. This brings a boundary condition for [M(tar)a]aq

2�2a; i.e. thevalue of a must be lower than one. Accordingly, the extraction reac-tion of the metal–tartrate complex is proposed as:

½MðtarÞa�2�2aaq þ bRHorg ¢ ½MðtarÞa � RbHbþ2a�2�org þ ð2� 2aÞHþaq

ð3Þ

where RHorg and [M(tar)a�RbHb+2a�2]org stand for D2EHPA and themetal–organic complex, respectively; b is the stoichiometric coeffi-cient of D2EHPA. The equilibrium concentration of D2EHPA(RHorg,aq) is:

RHorg;aq ¼ RHorg;0 � b½MðtarÞa � RbHbþ2a�2�org ð4Þ

Fig. 5. The variation of f(D) = logD � blog (0.6 � b[M(tar)a�RbHb+2a–2]org) versus pHfor different set of a and b; D2EHPA = 20 vol.%; (a) Co, (b) Ni; T = 25 �C, tartrateconcentration = 0.3 M.

Table 3Co and Ni extraction percentages as ‘‘mean of three replicates’’ at each of the 17combination of factor levels according to Table 1.

Number Co extraction percentage Ni extraction percentage

1 7 62 17 353 8 44 14 295 83 686 87 817 52 348 66 399 28 18

10 32 4411 61 5412 33 3713 3 3614 73 6915 34 4318 36 4517 32 46

H. Nadimi et al. / Minerals Engineering 69 (2014) 177–184 181

in which the initial concentration of D2EHPA (RHorg,0) is 0.6 M.Regarding the distribution coefficient (D) of the metallic ion and

by taking logarithm of the equilibrium constant of Eq. (3), the fol-lowing equation is achieved:

f ðDÞ ¼ log D� b log 0:6� b½MðtarÞa � RbHbþ2a�2�org

� �

¼ log Kþ ð2� 2aÞpH ð5Þ

Another boundary condition is that the argument of logarithmshould be higher than zero; i.e. b[M(tar)a�RbHb+2a�2]org must belower than 0.6 M. The maximum extractable Ni and Co, when theextraction percentage is 100%, is around 0.085 M. This means thatthe value of b should be lower than 7. The above mentionedboundary conditions can be summarized as a < 1 and b < 7.

It should be noted that despite considering all the boundaryconditions, estimation of the values of a and b is very complicatedregarding the possible numerous simultaneous equilibria betweenthe species. The values of a and b were examined by plotting‘‘f(D) = logD � blog (0.6 � b[M(tar)a.RbHb+2a�2]org)’’ versus equilib-rium pH employing a set of a and b. Then, the slope of each linewas compared with the value of 2–2a corresponding to theemployed set; if there was an adequate agreement between thesetwo values, that set of a and b would be selected.

It should be cleared that no proper set of a and b was found overthe whole range of pH (2–5) for both Co and Ni due to differentsimultaneous equilibria and possibly more than one metallic spe-cies being extracted. However, as it can be observed in Fig. 5 undera very limited range of pH (�3< and <�5), the appropriate set of aand b could be proposed as (0.5, �5) for Co and Ni. It is worthy ofnote that the proposed values of a and b are only applicable whenmetallic ion concentration in the organic phase does not exceed0.085 M (Darvishi et al., 2005).

3.4. Experimental design for RSM

3.4.1. Model fittingTable 3 lists the values of Co and Ni extraction percentages at

each of the 17 combination of factors presented in Table 1. The val-ues given are the mean of two independent experiments.

Regarding the values of the regression coefficient given inTable 4, the linear and quadratic terms were significant; this indi-cated that a second order polynomial model was necessary to rep-resent the data. All the second order terms of the independentparameters, apart from tartrate ion concentration in Ni and pH inCo extraction percentages equations, were significant togetherwith all the linear terms. The statistical analysis of the interactionterms showed that, at 10% significance level, there was a significantinteraction between pH and each of the other two factors, namelytemperature and tartrate ion concentration, in the case of Ni.While, there was only a significant interactive term between pHand tartrate ion concentration in the case of Co. Based on the esti-mated values of the regression coefficients (Table 4) two polyno-mial equations which fitted greater than 96% of the variation inthe data were calculated as (coded form):

%ENi ¼ 45:22þ 9:8T� 10:1½tar� þ 18:1pH� 14:79T2

þ 6:71pH2 � 4:5T:pH� 8:5pH � ½tar� ð6Þ

%ECo ¼ 35:62þ 3:8T� 8:1½tar� þ 31:2pH� 5:68T2

þ 11:82½tar�2 � 6:25pH � ½tar� ð7Þ

Also, the un-coded form of Eqs. (6) and (7) are:

%ENi ¼ �84:7þ 4:2T� 127:5½tar� þ 0:5pH� 0:04T2

þ 4:4pH2 � 0:2T � pH� 45:3pH � ½tar� ð8Þ

Table 4Values of regression coefficients calculated for the Co and Ni extraction percentages.

Independent factor Co Ni

Regression coefficient Standard error P value Regression coefficient Standard error P value

Un-coded Coded Un-coded Coded

Constant �57.93 35.29 2.051 0.000 �84.74 45.27 1.114 0.000LinearT 1.34 3.8 1.516 0.041 4.23 9.8 0.823 0.000[tar] �188.61 �8.1 1.516 0.001 �127.52 �10.1 0.823 0.000pH 24.58 31.2 1.516 0.000 0.46 18.1 0.823 0.000

QuadraticT�T �0.016 �6.27 2.928 0.070 �0.037 �14.72 1.591 0.000[tar]�[tar] 499.22 11.23 2.928 0.006 �9.7 �0.22 1.591 0.895pH�pH 1.11 1.73 2.928 0.573 4.34 6.78 1.591 0.004

InteractiveT�[tar] 0.25 0.75 1.694 0.671 �0.5 �1.5 0.921 0.147T�pH 0.01 0.25 1.694 0.887 �0.18 �4.5 0.921 0.002[tar]�pH �33.33 �6.25 1.694 0.008 �45.33 �8.5 0.921 0.000

Table 5ANOVA Table.

Co Ni

df SS P value df SS P value

Total 16 11445.1 16 6728.24Regression 9 11284.3 0.000 9 6680.76 0.000Residual error 7 160.8 7 47.47Lack of fit (model error) 5 152.8 0.120 5 42.80 0.228Pure error (replicate error) R2 2 8.0 2 4.67

Abbreviations: df = degree of freedom; SS = sum of squares.

182 H. Nadimi et al. / Minerals Engineering 69 (2014) 177–184

%ECo ¼ �57:9þ 1:4T� 188:6½tar� þ 24:6pH� 0:02T2

þ 499:2½tar�2 � 33:3pH � ½tar� ð9Þ

Regarding the ANOVA table (Table 5), the low values of P deter-mined for the regression (P < 0.001), as well as the fact that the lackof fit of the model was not significant (P > 0.1), revealed the suit-ability of the model (Table 5).

Fig. 6. The coded values of linear terms of the Co and Ni extraction percentagepolynomial equations.

3.4.2. The effect of pertinent factorsIn Fig. 6, the coded values of linear terms of the Co and Ni

extraction percentages polynomial equations are compared. Bothtemperature and pH have a positive effect on the Co and Ni extrac-tion, while tartrate ion concentration negatively affected theextraction percentages of Co and Ni. The reason for positive effectof pH is simply attributed to the acidic nature of D2EHPA as anextractant; i.e. an increase in pH value leads to the higher extrac-tion percentages of both Co and Ni. Moreover, it is very likely thatthe extraction reactions of Co and Ni are endothermic, which canbe accounted for the positive effect of temperature. On the otherhand, as it was shown in Section 4.1, the addition of tartrate ioninto the aqueous phase resulted in a shift towards higher pH val-ues; this meant that under certain conditions of pH and tempera-ture, an increase in concentration of tartrate ion decreased theextraction percentages of Co and Ni.

Fig. 6 also shows that pH is the most significant factor in the Co/Niextraction compared to temperature and tartrate ion concentration.Furthermore, the effect of pH on the extraction of Co is considerablyhigher than that of Ni; this is in agreement with the resultspresented in Fig. 2.

Surface plots give more information regarding the interactiveeffects of these factors on the response. Fig. 7 illustrates the surfaceplots corresponding to the interactions between pH and tartrateion concentration (in the case of Co) and pH and temperature/

tartrate ion concentration (in the case of Ni). Examination of thesurface plots presented in Fig. 7a reveals that, the variation of Coextraction versus tartrate ion concentration was insignificant com-pared to the changing pH. In the case of Ni (Fig. 7b), the extractionpercentage of Ni remained low even with increasing tartrate ionconcentration. But at high pH the extraction of Ni was significantlyhigher and increased with decreasing tartrate ion concentration.

According to the results presented in Fig. 7c, the extraction per-centage of Ni increased to an optimum temperature (around 40 �C)in the studied pH range; after this optimum temperature Ni extrac-tion percentage decreases by an increase in temperature.

Fig. 7. Surface plots for Co and Ni extraction percentages with respect to tartrateconcentration and pH (a and b) and with respect to temperature and pH (c); otherfactors fixed at their middle values.

H. Nadimi et al. / Minerals Engineering 69 (2014) 177–184 183

3.4.3. Optimization of factors in the Co/Ni separation processIt should be noted that the optimization of the factors (pH, tar-

trate ion concentration and temperature) for achieving the highest/lowest separation factor was carried out using the proposed qua-dratic polynomial equations (Eqs. (6) and (7)). Defining the separa-tion factor as:

SFCo=Ni ¼ DCo=DNi ð10Þ

and considering the relation between the distribution coefficient(D) and extraction percentage (E) for the equal volumes of organicand aqueous phases:

D ¼ E=ð100� EÞ ð11Þ

the following equation can be presented as:

SFCo=Ni ¼ DCo=DNi ¼ ENið100� ECoÞ=ECoð100� ENiÞ ð12Þ

By substituting Eqs. (6) and (7) into Eq. (10), the lowest separa-tion factor value is 0.03 under the following conditions of pH � 2.7,[tar] � 0.3 and T � 60 �C. To confirm this prediction, and the appli-cability of the proposed model, confirmation runs were carried outin duplicate. The 90% confidence interval for the separation factorunder optimized conditions was obtained as 0.032, which indicatesthe suitability of the regression models.

4. Conclusion

In the present study the effect of tartrate ion on the separationof Ni and Co by D2EHPA in the sulfate solutions was evaluated withthe following results:

– The addition of tartrate ion into the sulfate solutions containingCo and Ni resulted in a shift in the extraction curves of Co and Nitowards higher pH and in the case of Ni, this shift was more sig-nificant compared to Co. In the presence of tartrate ion with theconcentration of 0.3 M DpH0.5

Ni–Co reached the highest value of�0.9 which is close to that obtained by using a mixture ofD2EHPA and Cyanex 302 under the same conditions.

– The comparative FT-IR measurements in the presence andabsence of tartrate ion revealed that the additional variationsin C–H bands can be attributed to the contribution of tartrateion to the Co and Ni extraction by D2EHPA.

– The extraction reaction of the metal–tartrate complex was pro-posed as ½MðtarÞa�

2�2aaq þ bRHorg ¢ ½MðtarÞa � RbHbþ2a�2�orgþ

ð2� 2aÞHþaq in which the values of a and b were found to be sig-nificantly dependent on the chemical conditions of the system.

– CCD coupled with RSM experiments was used to develop twosecond order polynomial models for extraction percentages ofCo and Ni versus pH, tartrate concentration and temperatureas (coded form):

%ENi ¼ 45:22þ 9:8T� 10:1½tar� þ 18:1pH� 14:79T2 þ 6:71pH2

� 4:5T:pH� 8:5pH:½tar�

%ECo ¼ 35:62þ 3:8T� 8:1½tar� þ 31:2pH� 5:68T2 þ 11:82½tar�2

� 6:25pH:½tar�

The positive effect of temperature on the Co and Ni extractionpercentage was attributed to the endothermic nature of the extrac-tion reactions. Optimization of the levels of pH, tartrate concentra-tion and temperature for achieving the lowest separation factorshowed that, when the values of pH, tartrate concentration andtemperature were set at 2.7, 0.3 and 60 �C respectively, the SFCo/Ni

(DCo/DNi) was 0.03.

References

Begum, N., Bari, F., Jamaludin, S.B., Hussin, K., 2012. Solvent extraction of copper,nickel and zinc by Cyanex 272. Int. J. Phys. Sci. 7, 2905–2910.

Cheng, C.Y., 2006. Solvent extraction of nickel and cobalt with synergistic systemsconsisting of carboxylic acid and aliphatic hydroxyoxime. Hydrometallurgy 84,109–117.

Darvishi, D., Haghshenas, D.F., Keshavarz Alamdari, E., Sadrnezhaad, S.K., Halali, M.,2005. Synergistic effect of Cyanex 272 and Cyanex 302 on separation of cobaltand nickel by D2EHPA. Hydrometallurgy 77, 227–238.

Devi, N.B., Nathsarma, K.C., Chakravortty, V., 2000. Separation of divalentmanganese and cobalt ions from sulphate solutions using sodium salts ofD2EHPA, PC 88A and Cyanex 272. Hydrometallurgy 54, 117–131.

Flett, D.S., 2004. Cobalt–Nickel separation in hydrometallurgy: a review. Chem.Sustain. Develop. 12, 81–91.

Gu, Z.M., Wasan, D.T., Li, N.N., 1986. Ligand accelerated liquid membrane extractionof metal ions. J. Membr. Sci. 26, 129–142.

184 H. Nadimi et al. / Minerals Engineering 69 (2014) 177–184

Guezzen, B., Didi, M.A., 2012. Removal of Zn(II) from aqueous acetate solution usingDi (2-Ethylhexyl) Phosphoric Acid & Tributylphosphate. Int. J. Chem. 4 (3), 32–41.

Hano, T., Matsumoto, M., Ohtake, T., Egashira, N., Hori, F., 1992. Recovery of lithiumfrom geothermal water by solvent extraction technique. Solvent Extr. Ion Exch.10 (2), 195–206.

Keshavarz Alamdari, E., Darvishi, D., Haghshenas, D.F., Yousefi, N., Sadrnezhaad, S.K.,2012. Separation of Re and Mo from roasting-dust leach-liquor using solventextraction technique by TBP. Sep. Purif. Technol. 86, 143–148.

Mohapatra, D., Hong-In, K., Nam, C.W., Park, K.H., 2007. Liquid–liquid extraction ofaluminium(III) from mixed sulphate solutions using sodium salts of Cyanex 272and D2EHPA. Sep. Purif. Technol. 56, 311–318.

Montgomery, D.C., 2006. Design and Analysis of Experiments, sixtth ed. John Wiley& Sons, New York.

Nayl, A.A., 2010. Extraction and separation of Co(II) and Ni(II) from acidic sulfatesolutions using Aliquat 336. J. Hazard. Mater. 173, 223–230.

Okewole, A.I., Magwa, N.P., Tshentu, Z.R., 2012. The separation of nickel(II) frombase metal ions using 1-octyl-2-(20-pyridyl)imidazole as extractant in a highlyacidic sulfate medium. Hydrometallurgy 121–124, 81–89.

Reddy, B.R., Bhaskara Sarma, P.V.R., 2001. Transfer of nickel from sodium sulphatesolutions to the spent electrolyte through solvent extraction and stripping.Hydrometallurgy 60, 123–128.

Ren, Z., Zhang, W., Meng, H., Liu, Y., Dai, Y., 2007. Extraction equilibria of copper (II)with D2EHPA in kerosene from aqueous solutions in acetate buffer media. J.Chem. Eng. Data 52, 438–441.

Sainz-Diaz, C.I., Klocker, H., Marr, R., Bart, H.J., 1996. New approach in the modellingof the extraction equilibrium of zinc with bis-(2-ethylhexyl) phosphoric acid.Hydrometallurgy 42, 1–11.

Tsakiridis, P.E., Agatzini, S.L., 2004. Process for recovery of cobalt and nickel inpresence of magnesium from sulfate solutions by Cyanex 272 and Cyanex 302.Miner. Eng. 17 (7–8), 913–923.

Van de Voorde, I., Pinoy, L., Courtijn, E., Verpoort, F., 2005. Influence of acetate ionsand the role of the diluents on the extraction of copper (II), nickel (II), cobalt (II),magnesium (II) and iron (II, III) with different types of extractants.Hydrometallurgy 78, 92–106.

Xing, P., Wang, C., Ju, Z., Li, D., Yin, F., Chen, Y., Xu, S., Yang, Y., 2012. Cobaltseparation from nickel in sulfate aqueous solution by a new extractant: Di-decylphosphinic acid (DDPA). Hydrometallurgy 113–114, 86–90.

Zhu, Z., Zhang, W., Pranolo, Y., Cheng, C.Y., 2012. Separation and recovery of copper,nickel, cobalt and zinc in chloride solutions by synergistic solvent extraction.Hydrometallurgy 127–128, 1–7.