Model for Metal Extraction with Basic Extractants: a Coordination Chemistry Approach

HAL Id: hal-02271223

https://hal.archives-ouvertes.fr/hal-02271223

Submitted on 26 Aug 2019

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Model for Metal Extraction with Basic Extractants: a Coordination Chemistry Approach

Rayco Lommelen, Tom Hoogerstraete, Bieke Onghena, Isabelle Billard, Koen Binnemans

To cite this version:

Rayco Lommelen, Tom Hoogerstraete, Bieke Onghena, Isabelle Billard, Koen Binnemans. Model for

Metal Extraction with Basic Extractants: a Coordination Chemistry Approach. Inorganic Chemistry,

American Chemical Society, In press, �10.1021/acs.inorgchem.9b01782�. �hal-02271223�

Model for Metal Extraction with Basic

1

Extractants: a Coordination Chemistry Approach

2

Rayco Lommelen,

Tom Vander Hoogerstraete,

Bieke Onghena,

Isabelle Billard

and Koen 3

Binnemans

4

a

KU Leuven, Department of Chemistry, Celestijnenlaan 200F, P.O. box 2404, B

3001 5

Leuven, (Belgium) 6

b

Université Grenoble Alpes, CNRS, Grenoble INP, LEPMI, 38000 Grenoble (France) 7

8

ABSTRACT 1

2

Solvent extraction is a technique very often used for metal separation on industrial scale.

3

The extractants for solvent extraction can be subdivided into three categories: acidic, neutral 4

and basic extractants. The metal extraction mechanism of basic extractants is typically 5

described as an anion exchange process, but this mechanism does not correctly explain all 6

observations. This paper introduces a novel model for the extraction of metals by basic 7

extractants supported by experimental data on methyltrioctylammonium chloride and Aliquat 8

336 chloride systems. The hypothesis is that the metal species least stabilized in the aqueous 9

phase by hydration (i.e. the metal species with the lowest charge density) is extracted more 10

efficiently than the more stabilized species (i.e. species with higher charge densities). Once 11

transferred to the organic phase, the extracted species undergoes further Lewis acid-base 12

adduct formation reactions with the chloride anions available in the organic phase to form 13

negatively charged chloro complexes, which can associate with the organic cations. Salting- 14

out agents have an influence on the extraction, most likely by decreasing the concentration of 15

free water molecules, which destabilizes the metal complex in the aqueous phase. The 16

evidence provided includes: (1) the link between extraction and transition metal speciation;

17

(2) the trend in extraction efficiency as a function of different salting-out agents, and (3) the 18

behavior of HCl in the extraction system. The proposed extraction model explains better the 19

experimental observations and allows predicting the optimal conditions for metal extractions 20

and separations a priori, by selecting the most suitable salting-out agent and its concentration.

21 22 23

KEYWORDS: Aliquat 336; Anion exchangers; Transition metals; Solvent extraction;

24

PCA-MCR-ALS.

25

1. INTRODUCTION 1

2

Solvent extraction is a technique very often used for metal separation on industrial scale, 3

because it can process large volumes in a controllable manner.

In solvent extraction, a 4

metal-containing aqueous phase is contacted with an immiscible organic phase (solvent) 5

containing an extractant, a diluent and sometimes a modifier. The extractant is an organic 6

ligand designed to selectively coordinate to the target metal ions. The diluent is used to 7

increase the solubility of the metal complex in the organic phase and to reduce the viscosity, 8

but is in some cases omitted. The modifier is used to change some important physical 9

properties of the organic phase, for instance to prevent crud formation and to avoid third- 10

phase formation.

During the extraction, metal separation is achieved based on the difference 11

in affinity of the metal ions for the selected extractant. The extractants can be divided into 12

three main classes: (1) cation exchangers or acidic extractants (e.g. alkyl phosphorus acids or 13

carboxylic acids),

(2) solvating extractants or neutral extractants (e.g. ketones or 14

organophosphorus esters),

and (3) anion exchangers or basic extractants (e.g. protonated 15

amine or quaternary ammonium salts).

16

The extraction mechanism of metals by basic extractants is typically described as an anion 17

exchange process in which a negatively charged metal complex present in the aqueous phase 18

is exchanged for anions in the organic phase:

19 20

(1) 21

22

In equation (1), M represents the metal, X a metal-coordinating anion (e.g. Cl

, NO

, 23

SCN

, …), Q is a cation (e.g. protonated amine, quaternary ammonium ion,…) and the 24

horizontal bars indicate species in the organic phase.

25

The extraction of metals by basic extractants is assumed to be facilitated by the formation 1

of the anionic

complex in the aqueous phase (equation 2).

A sufficiently high 2

concentration of the metal-coordinating anion X

is thus required:

3 4

with – and (2) 5

6

Although many examples on the extraction of metals by basic extractants have been 7

reported in the literature, the extraction mechanism has not been fully elucidated yet, due to 8

the difficulties in describing all the equilibria involved in the extraction process.

9

Several interactions occur between the amine in the organic phase and the metal–

10

coordinating anion X

in the aqueous phase. First, non-protonated amines extract acids via an 11

acid-base reaction in the form of , with A being an amine. Secondly, the extractants 12

might form oligomers in the nonpolar organic phase.

Thirdly, the extractant is involved in 13

the extraction of the metal, and fourthly, all these equilibria depend on the nature of the 14

diluent. Moreover, the driving force behind extraction is the addition of salt to the aqueous 15

phase. This salt addition does not only change the metal speciation in the aqueous phase, but 16

it also changes the ionic strength and thus the activity of the species in the aqueous phase.

17

Hence, it is very difficult to determine stability constants and derive appropriate equations for 18

the extraction of metals by basic extractants. Moreover, additional equilibria occur at high 19

salt concentrations related to the association of the salting-out cation with the anion, reducing 20

the “free” anion concentration in the aqueous phase, (e.g. the formation of CaCl

by 21

coordination of Cl

to Ca

. Finally, various anionic metal species are present in the organic 22

phase as well.

23

In this paper, an alternative model for the extraction of metals by basic extractants is 24

presented. This model better explains the experimental observations described in the

25

literature and therefore it is probably closer to the true extraction mechanism. Although the 1

term extraction mechanism is often referring to how a metal is transferred from the aqueous 2

into the organic phase, the current approach rather focuses on how the extraction equilibrium 3

state is reached and influenced by the aqueous and organic phase. After introduction of the 4

new model, it is experimentally tested by studying the influence of salting–out agents on the 5

distribution ratios of Cu(II), Co(II) and Zn(II), related to the ionic strength, speciation and 6

hydration of the metal complexes in the aqueous phase, and the speciation in the organic 7

phase. Also, the behavior of HCl in solvent extractions using basic extractants is considered 8

to investigate the supposed competition between HCl and metal extraction reported in the 9

literature.

10

The difficulty in explaining all observations from metal extraction with basic extractants 11

has attracted attention in recent literature. Uchikoshi et al. investigated the aqueous 12

coordination chemistry of Cu(II) and Co(II) and tried to relate their findings to the absorption 13

of Cu(II) and Co(II) on anion-exchange resins. They concluded that an anion-exchange 14

mechanism for adsorption on resins was correct, although significant adsorption was already 15

found when only very small concentrations of anionic species were present.

Also, no 16

explanation for the decreasing adsorption efficiency at high HCl concentrations could be 17

given. Onghena et al.

and Vander Hoogerstraete et al.

investigated the speciation of 18

trivalent lanthanide ions in solvent extraction systems and concluded that the speciation in the 19

aqueous and organic speciation is different. Deferm et al. studied the relation between solvent 20

extraction and speciation, but now of In(III).

They already made a link between the aqueous 21

In(III) species and the extraction efficiency of In(III). Also, the extraction mechanism of 22

Fe(III) with Cyphos IL 101 has gotten some attention.

The authors proposed that Fe(III) can 23

also be extracted as FeCl

via an ion association mechanism together with the extraction of 24

FeCl

via an anion exchange mechanism.

25

2. EXPERIMENTAL SECTION 1

2

2.1. Chemicals 3

An aqueous solution of hydrogen bis(trifluoromethylsulfonyl)imide (80%) was purchased 4

from Iolitec (Heilbronn, Germany). HNO

(65 wt%), NH

Cl (99.99%), NaCl (99.99%), LiCl 5

(99.9%), NaOH (0.1 M), HCl (~37 wt%) and toluene were purchased from VWR (Leuven, 6

Belgium). CuO (99.99%), CuCl

(99.995%), MgCl

·2H

O (>99%), CaCl

·2H

O (>99%), 7

AlCl

·6H

O (>99%) and Aliquat® 336 were purchased from Sigma-Aldrich (Overijse, 8

Belgium). The cesium, copper, cobalt, scandium, indium and zinc aqueous standards (1000 9

mg L

in 35% HNO

), Lanthanum organic standard (1000 µg g

in Standard matrix oil 55- 10

65 mPas), CoCl

·6H

O (>98%), ZnCl

(>98%) and KCl (>99.5%) were obtained from Chem 11

Lab (Zedelgem, Belgium). Methyltrioctylammonium chloride (98%) was purchased from 12

J&K Scientific (Lommel, Belgium). CsCl (>99.9%) was obtained from Carl Roth (Karlsruhe, 13

Germany). Ethanol (absolute, >99.8%) was purchased from Fisher Scientific (Merelbeke, 14

Belgium). Water was always of ultrapure quality, deionized to a resistivity of >18.2 MΩ cm 15

with a Sartorius Arium Pro ultrapure water system. All chemicals were used as received, 16

without any further purification.

17 18

2.2. Metal extraction and quantification 19

Metal extractions (Cu(II), Co(II) and Zn(II)) were performed with 1 mL of aqueous phase 20

and 1 mL of organic phase in glass vials with a volume of 4 mL. The metal concentration in 21

the aqueous phase was kept constant by adding a fixed volume of a metal stock solution in 22

water to an aliquot of a highly concentrated acid or salt solution (HCl, LiCl, NaCl, KCl, 23

CsCl, MgCl

, CaCl

, AlCl

or NH

Cl), diluted with a certain volume of ultrapure water to a 24

total volume of 10 mL. The final concentrations of Cu(II), Co(II) and Zn(II) in the aqueous

25

samples was 0.5 g L

, 1.0 g L

and 2.0 g L

, respectively. A higher concentration of Zn(II) 1

was chosen as it is extracted much more efficiently, making it harder to measure accurately 2

the Zn(II) concentration in the aqueous phase after extraction. The salt solutions were 3

prepared by working very close to the solubility limit. The exact salt concentrations were 4

calculated based on the densities measured after preparing the solutions. In this way, 5

weighing errors due to the uptake of water by the hygroscopic salts were avoided.

The 6

concentrations of the salt solutions of AlCl

were calculated based on the weights because of 7

the unavailability of density–concentration data. HCl (0.5 mL, 37 wt%) was added to 25 mL 8

of the AlCl

salt solution to avoid hydrolysis of Al(III) and 0.0225 mol L

HCl was added to 9

all Zn(II) aqueous phases to avoid hydrolysis of Zn(II) during extraction.

10

The organic phase was made by diluting methyltrioctylammonium chloride (TOMAC) 11

(dried overnight on a Schlenk line) in toluene to a concentration of 0.2 M. The extractions 12

were performed for 30 min at room temperature at 2000 rpm and phase separation was 13

accomplished by centrifugation for 2 min at 5000 rpm.

14

The metal concentration in the aqueous phase before and after extraction was measured 15

using ICP-OES and the aqueous HCl concentration was corrected for the loss of HCl to the 16

organic phase, based on HCl extraction experiments. Distribution ratios (D) were calculated 17

with the following formula:

18 19

(3)

20 21

where

and

are the equilibrium metal concentrations in the aqueous and 22

organic phase after extraction, respectively. The concentration of metal in the organic phase 23

was calculated via the mass balance. In case of an equal volume of organic and aqueous 24

phase, equation (3) can be rewritten as:

25

1

(4)

2 3

with

being the initial metal concentration in the aqueous phase. The experimental 4

error was calculated based on triplicate measurements and was less than 5%. Error bars on 5

graphs were omitted for the sake of legibility.

6 7

2.3. HCl extraction and quantification 8

Extractions of HCl were performed on a slightly larger scale, i.e. 5 mL of aqueous and 5 9

mL of organic phase in centrifuge tubes of 15 mL. The aqueous phase consisted of HCl of 10

which the concentration was determined via its density. The organic phase was either water- 11

saturated Aliquat 336, water-saturated TOMAC or 0.24 mol L

Aliquat 336 in toluene (0.2 12

mol L

quaternary compound; commercial Aliquat 336 has 80 wt% quaternary 13

compounds).

All organic phases were presaturated with water to avoid large volume 14

changes. Extractions were performed using a Burrel wrist action shaker at room temperature 15

for 1 hour at 450 rpm. Afterwards the HCl concentration in the aqueous phase was 16

determined via its density and corrections were made for volume changes, determined 17

visually using a graduated cylinder. The organic water content was determined using Karl 18

Fischer titration.

19 20

2.4. Instrumentation and analysis methods 21

UV/VIS absorption spectra of the aqueous phases were measured with an Agilent Cary 22

6000i spectrophotometer and Cary WinUV software. Metal ion concentrations were 23

determined by inductively coupled plasma optical emission spectroscopy (ICP

OES), with a 24

Perkin Elmer Avio 500 spectrometer equipped with an axial/radial dual plasma view, a

25

GemCone High Solids nebulizer, a baffled cyclonic spray chamber and a demountable quartz 1

torch with a 2.0 mm internal diameter alumina injector. Samples, calibration solutions and 2

quality controls solutions were diluted with HNO

(2 vol%). All ICP-OES spectra were 3

measured in triplicate. Calibration curves were made using a solution of 0.1, 1 and 10 mg L

4

of the corresponding metal from a standard solution. Quality checks were performed with 5 5

mg L

metal in different concentrations of the salt, equal to the matrix concentrations after 6

dilution. In(III) or Sc(III) were added and only applied as internal standards if the quality 7

checks failed because of matrix effects.

8

Densities of the acid and salt solutions were measured with an Anton Paar DMA 4500M 9

densitometer. Nemus Life Thermo Shakers TMS–200 were used for the extraction 10

experiments. A Heraeus Labofuge 200 centrifuge was used to accelerate phase separation.

11

The water content in organic phases was measured using a Mettler-Toledo V30S volumetric 12

Karl Fischer titrator. The HCl in the organic phase (2 mL) was neutralized prior to titration 13

using triethylamine (10 mL) in dry methanol (10 mL) to avoid interference in the 14

measurements.

15 16

2.5. Synthesis 17

Copper bis(trifluoromethylsulfonyl)imide (Cu(Tf

N)

) was synthesized by mixing Cu(II) 18

oxide CuO (1.5 g, 18.8 mmol) with hydrogen bis(trifluoromethylsulfonyl)imide (HTf

N, 5.53 19

g, 15.7 mmol) in water for 4 h at 80 °C in a sealed vial. Afterwards, excess of CuO was 20

removed by filtration. The obtained metal solution was measured by ICP–OES and used as 21

metal stock solution for the UV/VIS measurements of the aqueous phase from 0 to 2 mol L

22

HCl. 50 mL of the stock solution was removed and water was evaporated by a rotary 23

evaporator. Afterwards, the obtained Cu(Tf

N)

salt was dissolved in 50 mL of ~37 wt% HCl

24

(12 mol L

) and used as metal stock solution for the UV/VIS measurements of the aqueous 1

phase between 2 and 12 mol L

HCl.

2 3

2.6. Calculations 4

The UV/VIS absorption spectra were analyzed following the Multivariate Curve 5

Resolution-Alternative Least Square (MCR-ALS) technique using toolboxes working in 6

Matlab R2018b software.

Chemical constraints were fixed to a minimum: (1) the total 7

amount of the metal was normalized (2) the independent UV/VIS absorption spectra do not 8

have negative absorbance values and (3) the concentrations of the species cannot be negative.

9

Principal component analysis (PCA) (using the same toolbox in Matlab) was performed prior 10

to the MCR-ALS study to determine how many different species were present in the solutions 11

and the results were compared with the literature.

12

The results of the MCR-ALS analysis are displayed in the form of the UV/VIS absorption 13

spectra of the independent species present in all samples containing that metal ion and of the 14

mole fraction of all species in function of the acid or salt concentration. The mole fraction of 15

a metal species (M

) is expressed as follows:

16 17

(5)

18 19

with n being the maximum amount of chlorides coordinated to M

. 20

Note that this mathematical treatment of the data is not based on any chemical model.

21

Therefore, ascribing a UV/VIS absorption spectrum to a certain metal complex is done by 22

considering the change in the concentration percentages as a function of the HCl 23

concentration and/or by comparison with previously published UV/VIS absorption spectra.

24

25

3. RESULTS AND DISCUSSION 1

2

3.1. Background 3

There are some generally accepted principles in the extraction of metals by basic 4

extractants.

First, the extraction of metals by basic extractants is assumed to occur via the 5

exchange of an anionic metal complex in the aqueous phase by one or more negatively 6

charged anions in the organic phase (equation 1). Secondly, the distribution ratio is higher 7

when the counter-ion of the salting-out agent is a metal cation instead of a proton. For 8

instance, the distribution ratio for extraction of metal ions from an aqueous solution 9

containing 8 mol L

LiCl is higher than from 8 mol L

HCl (Figure 1, curve 1 and 2).

10

Thirdly, the extraction of metal ions from HCl media typically shows a maximum in the 11

graph of the distribution ratio as a function of the HCl concentration at intermediate HCl 12

concentrations, resulting in a “bell-shaped” curve (Figure 1, curve 2). This phenomenon has 13

been attributed to the presence of the HCl

species, which competes with the anionic metal 14

complex for the extractant.

Fourthly, metal ions forming strong chloro complexes have 15

maxima in the distribution ratios at lower HCl concentration compared to metal ions that 16

form weaker chloro complexes (Figure 1, curve 3).

17

1

Figure 1. Typical shape of curves showing the distribution ratios of a metal ion as function of the 2

salting-out concentration with (1) LiCl as salting–out agent, (2) HCl as salting–out agent or (3) HCl 3

as salting–out agent in case the metal forms strong chloride complexes. It is assumed that there is 4

only one species in the organic phase.

5

Historically, the term anion exchange originates from salt metathesis reactions, in which 6

the anion initially present in the organic phase is exchanged by another anion initially present 7

in the aqueous phase. For instance, anion exchange resins are based on this phenomenon, 8

which are solids supports that trap ions from a solution.

The exchange between anions can 9

be predicted by the Hofmeister series for anions (equation 6).

Hydrophilic ions, in 10

general anions with a high charge density, will preferentially distribute to the aqueous phase 11

(left side of the series), whereas hydrophobic anions, which generally have a low charge 12

density, will preferentially distribute to the organic phase (right side of the series). Another 13

factor influencing the position of an anion in the Hofmeister series are the intramolecular 14

interactions between the anion and the aqueous or organic phase. The tendency of the anion 15

to be transferred from the aqueous to the organic phase is:

16 17

Citrate

< SO

< HPO

< F

< Cl

< Br

< I

< NO

< ClO

< SCN

(6)

18

1

The term anion exchange has been taken over by practitioners in the field of solvent 2

extraction. However, this term does not completely fit with experimental observations, and 3

this for several reasons. For instance, lanthanides (Ln) are extracted from nitrate media as 4

complexes by basic extractants, but it has been shown that, even at very high 5

nitrate concentrations, only hydrated

species exist in the aqueous phase and no 6

species.

The situation is even more striking in case of a split-anion 7

extraction where lanthanides are extracted from aqueous chloride media to a nitrate organic 8

phase in the form of

complexes.

Also in chloride media, only positively 9

charged lanthanide complexes exist and efficient extraction to basic extractants is observed at 10

high LiCl concentrations.

The extraction of In(III) is a second example for which an anion 11

exchange mechanism does not fit the observations. The anion exchange mechanism cannot 12

explain the change in speciation: In(III) is present in the aqueous phase as the octahedral 13

complex (x ≤ 3), while it is present in the organic phase as the tetrahedral 14

complex.

The experimental data of many reported metal extraction studies from 15

HCl media do not support the anion exchange hypothesis: a decrease in distribution ratio as a 16

function of the HCl concentration (Figure 1, curve 2 and 3) is not in accordance with Le 17

Châtelier's principle. The combination of equation (1) and (2) demonstrates that any increase 18

in chloride concentration should shift the equilibrium to the right, thus increasing the 19

extraction. The decrease in distribution ratio has often been attributed to the coextraction of 20

HCl as the anion species HCl

into the organic phase, competing with the metal 21

extraction.

However, no direct evidence for the competition between metal and HCl 22

extraction can be found in the literature. Furthermore, a decreasing distribution ratio at high 23

chloride concentrations to Aliquat 336 is observed for the extraction of Zn(II) from LiCl 24

media and decreasing distribution ratios of Zn(II) are also discovered for different chloride

25

salt solution – anion exchange resin systems.

In this case, no competition between LiCl 1

and Zn(II) extraction occurs as LiCl is not extracted. Also, the decreasing distribution ratios 2

cannot be explained by a changing speciation in the organic phase or because of a changing 3

composition of the organic phase, as the composition of the organic phase is almost constant 4

over the whole LiCl range.

5

6

3.2. New extraction model 7

The analysis of former extraction studies and the difficulty of explaining the whole 8

extraction process with currently accepted theories show that another model is required to 9

describe the extraction of metals by basic extractants. An alternative extraction model for the 10

extraction of metals with basic extractants is proposed here and experimentally tested as 11

described in the next sections. The new model is explained for a divalent metal ion (M

) and 12

its five different chloro complexes in the aqueous phase ([MCl

]

with 0 ≤ x ≤ 4), but can be 13

applied to metal ions with another charge and a different aqueous speciation, as well (Figure 14

2).

15

At low chloride concentrations, M

is hydrated in its first, second and even third 16

coordination sphere by a large number of water molecules (its hydration sphere). By 17

increasing the chloride concentration, there is a shift towards the formation of and 18

complexes. The charge density of the species is lower than that of M

, 19

decreasing the hydration sphere and hydration energy. The complex has no charge, and 20

the total number of hydrating water molecules in its hydration sphere is at a minimum. At 21

higher chloride concentrations, the species and

are formed, for which the 22

charge density increases again, resulting in larger hydration spheres and higher hydration 23

energies. According to the Hofmeister series, the species with the lowest hydration energy 24

(here MCl

) will preferentially distribute to the organic phase. The intermediate and short

25

living species present in the organic phase or at the interface reacts with one or two 1

molecules of the Lewis base in the organic phase to form or

, which 2

associates with the cation of the extractant. As a result, the maximum in the distribution ratio 3

is found close to the HCl concentration at which the fraction of metal species with the lowest 4

charge density (e.g. MCl

) in the aqueous phase is the highest. The smallest distribution ratios 5

will be found at those HCl concentrations at which the fraction of the metal species with the 6

highest charge densities (e.g. M

or

) is the highest in the aqueous phase. This 7

model is shown in Figure 2.

8

Not only the chloride concentration and linked metal speciation have an influence on the 9

hydration of the metal complex. Also the availability of water molecules that can hydrate the 10

metal complex, i.e. the water activity, has an influence on the extraction. The availability of 11

water molecules for the metal complex can be changed by the choice of cation of the salt 12

which provides the chlorides in the aqueous phase. A cation with a large charge density, such 13

as Li

, is strongly hydrated, making the water molecules less available for the metal complex 14

to be extracted. This decreases the hydration shell and the hydration energy of the [MCl

]

15

species in the aqueous phase at high ionic strength resulting in higher distribution ratios, 16

known as the salting-out effect. This is also visualized in Figure 2. A cation with a smaller 17

charge density (like K

) will not be hydrated to such an extent and more water molecules will 18

be available for binding to the metal complex.

19

1

Figure 2. Model for the extraction of a metal ion from weak (left) and strong (right) salting-out 2

agents. The metal ion is depicted in red and changes from positively charged at low salting-out agent 3

concentration to negatively charged by the complexation of salting-out anions. The salting-out cations 4

are depicted in green, the associated water molecules are blue and the organic cation is illustrated in 5

brown-yellow.

6

This extraction model can give new insights in the extraction of metals by basic 7

extractants, and it gives an explanation for poorly understood extraction phenomena. First, 8

lanthanide ions are extracted by basic extractants as positively charged and hydrated metal 9

complexes Ln(H

O)

at high LiCl concentration, which cannot be explained by the anion 10

exchange mechanism.

However, the new model states that the hydration (in all coordination 11

spheres) of the metal complexes determines the extraction efficiency. The Ln(H

O)

12

complexes are less hydrated at very high LiCl concentrations due to the low water activity 13

and thus extracted more. Secondly, the generally accepted principle that the decrease in metal 14

extraction at higher HCl concentrations is due to competition between metal and HCl 15

extraction via the HCl

species is refuted by the newly proposed model. The decrease in 16

distribution ratio at high chloride concentrations is related to the formation of stronger 17

hydrated metal complexes in the aqueous phase, instead of the competition between HCl and 18

metal extraction. Thirdly, The fraction of the metal species with the lowest charge density in 19

the aqueous phase (e.g. [MCl

]

) is formed at lower chloride concentrations for metals

20

forming strong chloride complexes (e.g. Zn(II)).

Therefore, these metal complexes have 1

their maximum in distribution ratio at lower HCl concentrations. The observation of a 2

maximum in the distribution ratio of Zn(II) in function of the LiCl concentration can also be 3

explained using the same concepts. Zn(II) forms negatively charged chloro complexes at low 4

chloride concentration, where the salting-out effect of LiCl is less pronounced, and there are 5

still enough free water molecules available to hydrate ZnCl

and ZnCl

slightly more than 6

ZnCl

. 7

The absolute distribution ratio of a metals is, of course, also dependent on the stability of 8

the chloro complexes in the organic phase. However, the effect of the stability of the metal 9

complexes in the organic phase on the distribution ratio is almost the same over the whole 10

aqueous chloride concentration range, as the speciation of the metal complex in the organic 11

phase is independent of the chloride concentration in the aqueous phase.

The sole changes 12

in the organic phase that can influence the distribution ratio of a metal are the presence of 13

significant quantities of other compounds in the organic phase These include the presence of 14

HCl in the organic phase when high aqueous HCl concentrations are used and the presence of 15

the same or other extracted metals in the organic phase at high loadings. The latter, although 16

significant for industrial metal separations, is outside the scope of present paper and will be 17

the topic of future work.

18 19

3.3. Literature data on speciation and extraction 20

Linking speciation and extraction allows to investigate the way metals are extracted and 21

can be used to test the applicability of the newly proposed extraction model. A speciation 22

profile of metal chloride complex can be constructed using stability constants reported in the 23

literature. However, stability constants are given at standard condition (i.e. zero ionic

24

strength) and significant deviations the speciation profiles derived from these constants are 1

expected at high ionic strength. To correct for the deviations of chloride, an approximation of 2

the activity coefficient model by Helgeson et al. at 25 °C was used:

3

4

(7) 5

6

with I the ionic strength and Z

the absolute charge of chloride. The complete model can 7

calculate activity coefficients quite accurately, but this approximation is more limited than 8

the complete model of Helgeson et al. The approximation was used as it is user-friendly and 9

an approximate image of the speciation of a metal in chloride solution is sufficient for these 10

preliminary literature studies. The stability constants of a metal forming strong chloride 11

complexes (Zn(II)) and of one forming weak chloride complexes (Ni(II)) were used as 12

examples. The resulting speciation profiles were linked to their corresponding extraction 13

profiles using diluted Aliquat 336 and HCl or LiCl in the aqueous phase, respectively (Figure 14

3 and Figure 4).

15

The maximum in distribution ratio of Zn(II) from HCl media coincides very well with the 16

highest mole fraction of the Zn(II) species with the lowest charge density (ZnCl

). This 17

species is likely the least hydrated and thus the most extracted according to our model 18

presented above. The extraction is lower at chloride concentration were the mole fraction of 19

positively or negatively charged Zn(II) species is higher, but the reduction in distribution 20

ratio is less pronounced when the amount of negatively charged Zn(II) increases. This might 21

be due to the higher HCl concentration, which acts as a minor salting-out agent by hydrating 22

some water molecules and due to the smaller increase in charge density of the anionic 23

complexes, which is related to their larger volume.

24

The extraction of Zn(II) from HCl media is significantly higher compared to the extraction 1

of Ni(II) from HCl media over the whole chloride range, because Zn(II) forms chloride 2

complexes much easier. Despite the absence of negatively charged Ni(II) complexes in the 3

aqueous phase, Ni(II) is still extracted to some extent. This cannot be explained by an anion 4

exchange mechanism, which requires negatively charged species. However, the extraction of 5

Ni(II) is significant at LiCl concentrations were the mole fraction of neutral NiCl

is highest, 6

which is in agreement with our model.

7

8

Figure 3. Extraction and speciation profile of Zn(II) from HCl towards Aliquat 336 in benzene using 9

literature data.

10

1

Figure 4. Extraction and speciation profile of Ni(II) from LiCl towards Aliquat 336 in diethylbenzene 2

using literature data.

3

4

3.4. Metal speciation 5

The speciation and extraction curves presented above are only approximate due to the 6

insufficient consideration of the influence of ionic strength. An experimental investigation of 7

the speciation of metal complexes at the same ionic strength as the actual extraction is much 8

more reliable. Therefore, the speciation of Cu(II) and Co(II) in HCl media was investigated 9

and linked to extraction experiments with the same aqueous solutions, because of the 10

particular trend of the curve of the distribution ratio of Cu(II) and Co(II) vs. HCl 11

concentration. Methyltrioctylammonium chloride (TOMAC) (0.2 M) dissolved in toluene 12

was used for the extractions instead of its industrial equivalent (Aliquat 336) to avoid the 13

influence of impurities and to exactly quantify and control the amount of quaternary 14

compounds in the organic phase. Aliquat 336 has only about 80 wt% quaternary compounds

15

(MeR

NCl with R a mixture of C

and C

hydrocarbons) while also having about 8 wt%

1

alcohols and other impurities not accounted for.

2

The speciation of Cu(II) and Co(II) as a function of the HCl concentration was studied by 3

UV/VIS absorption spectroscopy. UV/VIS absorption spectra were recorded at different 4

aqueous solutions containing Cu(II) (0.5 g L

) or Co(II) (1.0 g L

) and different amounts of 5

HCl (0 to 11.9 mol L

Cl

). The Cu(II) solutions for the speciation measurements were made 6

from copper bis(trifluoromethylsulfonyl)imide (Cu(Tf

N)

) to avoid the presence of chlorides 7

from the metal salt at very low HCl concentrations. No TfN

was present in the solutions for 8

the extraction experiments. The Tf

N anion does not associate with Cu(II) in solution, thus 9

resulting in a fully hydrated Cu(II) complex in water. The color of the Cu(II) solutions 10

changed gradually from blue towards green/yellow (Figure 5), because of the changes 11

between the five different Cu(II) species in the aqueous phase: Cu

, CuCl

, CuCl

, CuCl

12

and CuCl

.

13

14

Figure 5.Color change of 0.5 g L

Cu(II) (added as Cu(Tf

N)

salt) as a function of the HCl 15

concentration (0 to 11 mol L

Cl

) in the aqueous phase.

16

It is known that hydrated octahedral copper ([Cu(H

O)

]

) is a weak absorber in the 17

UV/VIS spectral region investigated here (220 to 500 nm, Figure 6). No absorption is 18

observed in the measured wavelength range except for a small increase in absorbance below

19

absorption maximum at 250 nm.

The individual absorption spectra of the [CuCl

(H

O)

] 1

and [CuCl

(H

O)]

species, having maxima at 270 and 283 nm, respectively, cannot be 2

directly observed in Figure 6 due to overlap of the absorption spectra of the different 3

species.

Note that the geometry changes from octahedral [CuCl

(H

O)

] to tetrahedral 4

[CuCl

(H

O)]

.

The octahedral [Cu(H

O)

]

, [CuCl(H

O)

]

and [CuCl

(H

O)

] complexes 5

do not contribute to the absorption maximum around 380 nm, while the main contribution at 6

380 nm comes from the tetrahedral complex [CuCl

]

.

The optical absorption spectra of 7

Cu(II) in the aqueous phase obtained in this study show large similarities with those reported 8

previously by other research groups.

9

10

Figure 6. UV/VIS absorption spectra of 0.5 g L

Cu(II) (added as Cu(Tf

N)

salt) as a function of the 11

HCl concentration (0 to 11.8 mol L

Cl

) in the aqueous phase.

12

It is not possible to derive the exact speciation of Cu(II) at a given HCl concentration 13

directly from the UV/VIS absorption spectra. Nevertheless, a full speciation profile and de 14

UV/VIS absorption spectra of each of the Cu(II) species can be deduced using a 15

statistical/mathematical technique called Principal component analysis (PCA) and

16

Multivariate Curve Resolution-Alternative Least Square (MCR-ALS).

The mole fractions 1

of the different Cu(II) species as a function of the HCl concentrations, as obtained from the 2

PCA-MCR-ALS analysis of the UV/VIS absorption spectra, can be found in Figure 7. The 3

calculated spectra (See SI) and speciation profile are consistent with models, theoretical 4

calculations and UV/VIS absorption analysis reported by other authors.

Also, the 5

distribution ratio of Cu(II) for extraction from HCl media towards 0.2 mol L

TOMAC in 6

toluene is given in Figure 7, to enable a comparison between the speciation and the extraction 7

of Cu(II).

8

9

Figure 7. The distribution ratio of Cu(II) from HCl media towards 0.2 mol L

TOMAC in toluene 10

(black) and the mole fraction of the different Cu(II) species as a function of the chloride concentration 11

in HCl medium (colored).

12

The lowest distribution ratios for Cu(II) extracted from HCl media are found at very low 13

(e.g. 0.5 mol L

) and very high HCl (e.g. 10 mol L

) concentrations where the mole fractions 14

of [Cu(H

O)

]

or [CuCl

]

are the highest. In other words, the results depicted in Figure 7 15

suggest that an increase in [CuCl

]

concentration lowers the distribution ratio, which is not

16

in accordance with the generally accepted equations (1 and 2) for the extraction of metals by 1

basic extractants via anion exchange.

2

The highest distribution ratio for Cu(II) is obtained at 5.6 mol L

HCl, which is close to 3

the maximum in the mole fraction of [CuCl

(H

O)

] found at 4.8 mol L

HCl. Also, the 4

general trend in extraction efficiency seems to follow the general trend in the speciation of 5

[CuCl

(H

O)

].The difference between the maximum of the distribution ratio and the 6

maximum of the mole fraction of [CuCl

(H

O)

] can be explained by the non-negligible 7

solubility of HCl in the organic phase (vide infra) and changes in the water activity due to 8

changes in the HCl concentration. The HCl concentration in the aqueous phase was corrected 9

for the loss of HCl to the organic phase, but no corrections were made for the increase in 10

chloride concentration in the organic phase. The latter was omitted due to the absence of a 11

quantitative relation between the extraction efficiency and the organic HCl concentration.

12

Next, the abovementioned methodology was used to investigate the relation between the 13

extraction behavior of Co(II) and its aqueous speciation in HCl media. Although Cu(II) was 14

added as Tf

N salt for the spectroscopic study, this lengthy procedure was not repeated for 15

the Co(II) speciation study as no significant influence on the speciation curve of Cu(II) was 16

detected compared to what has been reported in the literature.

The UV/VIS absorption 17

spectra were recorded and two distinctive regions were visible (Figure 8). Octahedral Co(II) 18

complexes weakly absorb in the region between 400 and 550 nm, while tetrahedral Co(II) 19

complexes show a much more intense absorption between 550 nm and 750 nm.

The 20

octahedral Co(II) complexes are mainly present at low HCl concentrations (up to 7.8 mol L

, 21

vide infra) while tetrahedral Co(II) complexes are formed at higher HCl concentrations (from 22

7.8 mol L

, vide infra). The same observations can be deduced from the color change of the 23

Co(II) solutions from pale pink, for [Co(H

O)

]

at low HCl concentrations, to dark blue, for 24

[CoCl

]

complexes at high HCl concentrations.

25

1

Figure 8. UV/VIS absorption spectra of 1.0 g L

Co(II) as a function of the HCl concentration (0.015 2

to 11.8 mol L

Cl

) in the aqueous phase.

3

The large difference in the molar absorption coefficient of the octahedral and tetrahedral 4

Co(II) species makes it more difficult to determine the amount of the different Co(II) species 5

present in solution. Presumably there are 5 different species present (i.e. CoCl

with 0  x

6

4), but many studies on the speciation of Co(II) in aqueous media are inconsistent.

An 7

extensive study on the speciation of Co(II) species in HCl media has been published recently 8

by Uchikoshi, who aimed to remove the ambiguity on the speciation of Co(II).

The author 9

used UV/VIS and X-ray absorption spectroscopy to arrive at the conclusion that only three 10

species exist in aqueous HCl media, being octahedral hydrated [Co(H

O)

]

and 11

[CoCl(H

O)

]

, and tetrahedral [CoCl

]

. This conclusion is consistent with the PCA analysis 12

performed on the UV/VIS absorption spectra in the current study, which revealed that three 13

species were necessary to lower the residuals below the experimental error of 1%. Thus, the 14

MCR-ALS analysis of the Co(II) spectra was performed using three species and the resulting 15

spectra of the individual Co(II) species and the speciation profile of Co(II) in HCl can be

16

found in the SI and Figure 9, respectively. The speciation profile and individual spectra of 1

Co(II) are in good agreement with those reported in the literature.

2

The distribution ratio of Co(II) from aqueous HCl media towards 0.2 mol L

TOMAC in 3

toluene are given in Figure 9 to allow for a comparison between its speciation in aqueous 4

media and extraction towards basic extractants. The maximum in distribution ratio of Co(II) 5

is located at 8.1 mol L

HCl and the highest mole fraction of the Co(II) species with the 6

lowest charge density (CoCl

) is at 7.6 mol L

HCl. As for Cu(II), the similar HCl 7

concentration for the highest mole fraction of [CoCl(H

O)

]

and the maximum in 8

distribution ratio of Co(II) suggest, together with the similar shape of the distribution curve 9

and speciation curve of [CoCl(H

O)

]

, that the species with the lowest charge density (and 10

lowest hydration) is extracted preferentially.

11

Three further comments can be made: (1) Again the maximum in distribution ratio is 12

located at slightly higher HCl concentration compared to the maximum in distribution ratio of 13

the species with the lowest charge density. This can be explained by the presence of HCl in 14

the organic phase or changes in the water activity as mentioned for the Cu(II) extraction; (2) 15

The species with the lowest charge density is not necessarily the neutral species. CoCl

is the 16

species with the lowest charge density, as CoCl

is not present in solution; (3) The absence of 17

a changing speciation of the metal complex in the organic phase for both Cu(II) and Co(II) 18

(see SI for details) shows the validity of linking the extraction as a function of HCl 19

concentration to the changes in speciation of the aqueous phase.

20

1

Figure 9. The distribution ratio of Co(II) from HCl media towards 0.2 mol L

TOMAC in toluene 2

(black) and the mole fraction of the different Co(II) species as a function of the chloride concentration 3

in HCl medium (colored).

4 5

3.5. Influence of salting-out agents 6

As mentioned in the introduction of the new extraction model, a cation of a salting-out 7

agent with a higher charge density would be more strongly hydrated and thus would decrease 8

the amount of free water present in solution. This would more efficiently in decreasing the 9

effective hydration of the metal complex that is extracted. To test qualitatively the effect of 10

slating-out agents and hydration on metal extraction with basic extractants, extraction 11

experiments were performed with Cu(II), Co(II) and Zn(II) from aqueous chloride media 12

towards 0.2 mol L

TOMAC in toluene. Figure 10 shows the distribution ratios of 0.5 g L

13

Cu(II) extracted by 0.2 mol L

TOMAC in toluene as a function of the concentration of 14

mono-, di- and trivalent salting-out agents in the aqueous phase. The extraction of 2.0 g L

15

Zn(II) by 0.2 mol L

TOMAC in toluene was studied in function of the concentration of

16

different monovalent salting-out agents to investigate the effects of salting-out agents in more 1

detail (Figure 11). Zn(II) was chosen, as it forms strong chloro complexes,

resulting in a 2

maximum in distribution ratio at low chloride concentrations. This way, all alkali chlorides, 3

NH

Cl and HCl could be used as salting-out agents up to a concentration larger than the 4

chloride concentration linked with a maximum in distribution ratio of Zn(II).

5

6

Figure 10. Extraction of 0.5 g L

Cu(II) as a function of the initial chloride concentration to 0.2 mol 7

L

TOMAC dissolved in toluene.

8 9

Above 2 mol L

Cl

, the distribution ratios of Zn(II) increases in the following order of 10

monovalent salting-out cation:

11 12

Li+ > Na+ > H+ > K+ ≈ NH4+ > Cs+ (8) 13

14

This series is slightly different from the Hofmeister series.

For instance, the similar 15

extraction efficiency for Zn(II) from KCl and NH4Cl can be explained by their similar ionic

16

radii.

The positioning of HCl in the series (equation 8) is not that clear, as its presence in 1

the organic phase also influences the extraction of metals (vide supra). Nevertheless, the 2

similarities between the Hofmeister series and the general decrease in cationic charge density 3

from left to right in equation 8 suggest that the amount of free water molecules in the aqueous 4

phase has a large effect on the shape of the graph displaying the distribution ratio as a 5

function of the chloride concentration.

6

7

Figure 11. Extraction of 2.0 g L

Zn(II) as a function of the initial chloride concentration to 0.2 mol 8

L

TOMAC dissolved in toluene 9

The distribution of Cu(II) was performed with a few alkali chlorides (LiCl and NaCl), 10

NH

Cl, HCl and some divalent and trivalent chloride salts (CaCl

, MgCl

and AlCl

) to check 11

the observations from the Zn(II) extraction experiments and to the effect of salting-out agent 12

with a cation with a higher charge density than Li

. Below 2 mol L

chlorides, the 13

distribution ratios of Cu(II) are very similar for the different chloride salts. This is expected, 14

because the chloride activity in the aqueous phase and the amount of free water in the

15

aqueous phase would be rather similar in the different systems at these relatively low salt 1

concentrations.

2

Above a concentration of 2 mol L

chloride, the distribution ratios of Cu(II) from CaCl

, 3

MgCl

or AlCl

are higher than those of NH

Cl and HCl at equal initial chloride 4

concentrations, which is expected based on the Hofmeister series. Surprisingly, the 5

distribution ratios of Cu(II) for extraction from solution with salting-out agents of divalent 6

and trivalent cations are lower at equal initial chloride concentrations than the distribution 7

ratios with NaCl and LiCl as salting-out agent, even though the charge density of the cations 8

are higher. This can be explained by the association reaction between the salting-out cations 9

and anions in the aqueous phase resulting in a significant decrease in ionic strength of the 10

aqueous phase.

For instance, the association/dissociation reaction of CaCl

can be written 11

12 as:

13

(9) 14

15

The situation at the right side of equation (9) is found at low ionic strength, but the 16

association reactions between Ca

and Cl

cannot be neglected at higher ionic strengths 17

(from 1 mol L

). Equation (9) clearly shows that the charge of the ions is significantly 18

reduced by this association reaction: one +II charged metal ion (Ca

) and one –I charged 19

chloride ion (Cl

) are replaced by one calcium(II) monochloride cation (CaCl

) with a +I 20

charge. This decrease in charge density lowers the ionic strength of the aqueous phase 21

significantly and might explain the decrease in distribution ratios of Cu(II) to values below 22

those found for extraction from LiCl media at similar initial chloride concentrations. The 23

degree of association cannot easily be predicted because of the changing ionic strengths. For 24

instance, Johnsen et al. calculated that about 46% of the Ca

and 50% of Mg

are not

25

bonded with Cl

in sea water, whereas 85% of Na

ions are still in its dissociated form under 1

the same conditions.

2

Although the similarities between hydration effects, the charge density of the salting-out 3

agent, the Hofmeister series and the extraction are very clear, the experimental evidence 4

given is only an indirect proof. A direct observation of the hydration sphere or energy of 5

salting-out agents and transition metals in function of the salting-out concentrations would be 6

very interesting to study the proposed extraction model more in detail and make the model 7

quantitative. This will be the subject of further investigation.

8

Another way to explain the difference in metal extraction from solutions with different 9

salting-out agents would be via a change in metal speciation when changing the salting-out 10

agent. Different salting-out agents can influence the speciation of a metal because of two 11

reasons: (1) a change in the activity of all species in solution due to different interactions with 12

the cation of a different salting-out agent, and (2) a change in free chloride concentration due 13

to the association of mono-, di- and certainly trivalent salting-out agents. The speciation of 14

Co(II) in LiCl was determined via UV/VIS absorption spectroscopy and PCA-MCR-ALS, 15

similarly to the determination of Co(II) species in HCl. The two speciation profiles were 16

compared (Figure 12) to determine the difference in speciation due to a change in monovalent 17

salting-out agent. The association of both LiCl and HCl is similar and considerably lower 18

compared to that of divalent and trivalent salting-out agents. This allows to observe the effect 19

of activity on the speciation of Co(II), although effects of changing salting-out agent 20

association cannot be completely excluded..

21

1

Figure 12. Speciation profile calculated with PCA-MCR-ALS of Co(II) in HCl (top) and LiCl (bottom) 2

as function of the total chloride concentration.

3

The speciation profile of Co(II) in LiCl is very similar to that of Co(II) in HCl. The curves 4

of [CoCl(H

O)

]

and [CoCl

]

are shifted to higher chloride concentrations by 1 mol L

5

unit, but this cannot explain the observed difference in extraction behavior of Co(II) from 6

HCl and LiCl towards basic extractants at high chloride concentrations. To clarify, the 7

distribution curve of Co(II) from HCl shows a maximum at 8 mol L

HCl, while the 8

distribution curve of Co(II) from LiCl keeps on increasing with increasing chloride 9

concentration.

A mole fraction close to one of [CoCl

]

complexes is still formed at high 10

LiCl concentrations, which agrees with the observed blue color of the measured solutions.

11

The increase in mole fraction of species with a higher charge density would still decrease the 12

extraction efficiency due to an increased hydration of these Co(II) species. However, the high 13

LiCl content decreases the amount of free water molecules drastically, due to association of 14

water molecules with the Li(I) cation, which has a very high charge density. This results in a

15

net decrease in hydration of Co(II) in the aqueous phase at high chloride concentrations, so 1

that extraction is more efficient.

2

The speciation of metal complexes in divalent and trivalent salting-out agents might be 3

shifted more over the chloride concentration range due to the association of the salting-out 4

agents. However, the results presented above suggest that this will not change the 5

applicability of our extraction model. The determination of the speciation of metal complexes 6

in a wide range of mono-, di- and trivalent salting-out agents is of interest for a more 7

quantitative explanation of the observed extraction phenomena and will be investigated in the 8

future.

9 10

3.6. Extraction behavior of HCl 11

Apart from the metal extraction itself, also the behavior of HCl in the extraction systems 12

requires further investigation. A first impression of the HCl extraction behavior towards 13

basic extractants can be provided by performing extraction experiments on HCl itself 14

towards water-saturated Aliquat 336 and 0.24 mol L

Aliquat 336 in toluene (resulting in 15

0.2 mol L

quaternary compounds) (Figure 13). Aliquat 336 was chosen here, because 16

these results can be used directly in other works to correct for the HCl concentration in 17

both phases, as Aliquat 336 is an industrially relevant extractant. However, the metal 18

extractions were performed using TOMAC to increase the accuracy for investigating the 19

extraction mechanism of basic extractants. A smaller scale study was performed using 20

water saturated TOMAC as extractant for comparison with Aliquat 336 and the HCl 21

extraction followed the same trend (see SI).

22

Both the HCl concentration in the water-saturated Aliquat 336 and 0.24 mol L

Aliquat 23

336 phase increased linearly with the equilibrium HCl concentration in the aqueous phase, up 24

to a value above the stoichiometric concentration of basic extractant in the organic phase.

25