Journal of Solution Chemistry

, Volume 47, Issue 8, pp 1395–1417 | Cite as

Study of Cadmium Extraction with Aliquat 336 from Highly Saline Solutions

  • Hans Vigeland Lerum
  • Niels Højmark Andersen
  • Dag Øistein Eriksen
  • Eddy Walther Hansen
  • Dirk Petersen
  • Grethe Wibetoe
  • Jon Petter Omtvedt


A large number of model solutions with high ionic strength were synthesised to mimic industrial conditions and were used as a first approach to study Cd extraction in the presence of chloride at high salinity, as experienced in real industrial solutions. The extractant used throughout in this work was Aliquat 336, a quaternary ammonium salt well known to the hydrometallurgical industry. The effects of some selected anions in addition to chloride (i.e., perchlorate, nitrate, and sulfate) were studied. The distribution of cadmium was measured using 109Cd as a tracer. Liquid-scintillation spectroscopy quantified the concentration of 109Cd in both phases. Raman and NMR spectroscopy were employed to gain further insight into the extraction chemistry. A careful analysis of all Cd extraction data showed that within specific windows of the reactant concentrations the chemical reactions could be represented by simplified equations, as discussed thoroughly in the text. Equilibrium constants for the extraction of \({\text{CdCl}}_{3}^{ - }\) from chloride and chloride/sulfate media were determined to be log10Kext = 4.9 ± 0.8 and log10Kext = 5.7 ± 0.5, respectively. For the nitrate environment, an exchange reaction involving a LiNO3 ion pair is proposed and agrees with the experimental data, but was not proven. 14N-NMR and Raman spectroscopy confirmed that the relative affinity of Aliquat 336 for the relevant anions followed the order: perchlorate > nitrate > chloride > sulfate. Finally, 14N-NMR enabled the equilibrium constant of the exchange reaction between nitrate and chloride for Aliquat 336 to be determined.


Liquid–liquid extraction NMR Raman spectroscopy Cadmium High salinity Radiotracers 

1 Introduction

Cadmium is a toxic metal that normally occurs at low concentrations in raw materials used for the recovery of metals and fertilizer production. Therefore, low concentrations of this unwanted metal occur in industrial process streams and may disturb the process. It may also end up in the final product in too high a concentration. Environmental and/or health concerns will then require removal of this cadmium. From an economic point of view, the future appears challenging as tighter legislative control and stricter limits for release from industrial plants are expected. Simultaneously, the quality of available ores is decreasing [1], i.e. exploited minerals will become more complex and contain higher amounts of unwanted elements such as cadmium. This will increase the cost of processing in the future. For these reasons, it will be necessary to enforce efficient control on the amounts of Cd in many of today’s industrial processes. Clearly, effective methods to remove even low concentrations of Cd from the solutions will be needed.

In the present work, synthetic industrial solutions, i.e. of high salinity, based on chloride media were used as a starting point for studying the extraction chemistry of trace amounts of Cd. The effects of high salinity and the presence of some selected anions (perchlorate, nitrate, and sulfate) were studied. Radioactive Cd-tracer was used to measure the distribution of cadmium. However, selective removal of metal(s) in low concentration(s) from complex and concentrated solutions is challenging. In order to approach this challenge, the mechanics of the extraction processes needs to be better understood. To achieve this, Raman and NMR spectroscopy were combined to gain further insight into other species’ influence on the extraction chemistry.

The goal of this work was to study extraction of Cd from aqueous solutions with high salinity. Aliquat 336 (A336) was chosen as the extractant as its extraction of Cd from chloride media has been studied before and is well known to industry. Safarzadeh et al. have made a comprehensive review of Cd recovery from various sources [2] while various extraction and separation methods were given by Jha et al. [3]. Quaternary amines have been shown to have selectivity for Cd-chloro complexes compared to those of other metal ions that form anionic chloride complexes such as Hg(II), Zn(II), Cu(II), and Co(II) [4, 5].

A key point in understanding the extraction efficiency for metal ions from brine solutions using the extractant A336 relates to the relative affinity of A336 for the different anions. Changing the anion bound to A336 will change its extraction properties [6]. Subsequently, Cd can be stripped from the organic solution using several different anions (perchlorate, nitrate, sulfate, phosphate) [7]. It has been suggested that A336 extracts both \({\text{CdCl}}_{3}^{ - }\) and \({\text{CdCl}}_{4}^{2 - }\) from solutions with chloride [8]. A336 can be dissolved in several different organic solvents and gives a high extraction of Cd from chloride environments [9]. Miller and Fuerstenau have proposed that A336 will replace water when Cd is extracted and that there will be a low concentration of water in the organic solution [10].

In the first part of this work, a radioactive 109Cd tracer, measured using liquid scintillation spectroscopy, was used to determine extraction distribution ratios. Experimental data on the distribution of Cd between the organic phase and the aqueous phase using different anions, covering a broad range of concentrations, is presented.

In the second part, Raman and NMR spectroscopy are used to study how various anions affect the extraction of Cd, e.g., perchlorate, sulfate and nitrate. These have well known Raman spectra [11, 12] so that tracking the transport of these anions from the aqueous phase to the organic phase is possible. Using NMR, it has been shown that Cd affects the position and width of the 31P peak when pre- and post-extraction spectra are compared [13]. Thus, NMR spectra provide information on Cd binding.

2 Equilibrium Modelling

Cd(II) readily forms aqueous complexes with several different ligands including chloride and sulfate [14]. The kinetics of ligand exchange is fast [15]. Table 1 lists recommended equilibrium constants (when available) and equilibrium reactions, relevant for this work. KiCl is the equilibrium constant for consecutive addition of chloride ligands to cadmium.
Table 1

Formation of Cd chloride and sulphate complexes


Equilibrium constant


Cd2+ + Cl ⇌ CdCl+

log10 K1Cl = 1.98


CdCl+ + Cl ⇌ CdCl2

log10 K2Cl = 0.664


\({\text{CdCl}}_{ 2} \, + \,{\text{Cl}}^{ - } \rightleftharpoons {\text{CdCl}}_{3}^{ - }\)

log10 K3Cl = − 0.34


\({\text{CdCl}}_{3}^{ - } \, + \,{\text{Cl}}^{ - } \rightleftharpoons {\text{CdCl}}_{4}^{2 - }\)

log10 K4Cl = − 0.6


\({\text{Cd}}^{ 2+ } \, + \,n{\text{Cl}}^{ - } \rightleftharpoons {\text{CdCl}}_{n}^{(2 - n)}\)

\({ \log }_{10} \beta_{{n{\text{Cl}}}} = \mathop \sum \limits_{i = 1}^{n} { \log }_{10} K_{{i{\text{Cl}}}}\)

Eq. 1

\({\text{Cd}}^{ 2+ } \, + \,{\text{SO}}_{4}^{2 - } \rightleftharpoons {\text{CdSO}}_{ 4}\)

log10 K1SO4 = 2.36


\({\text{CdSO}}_{ 4} \, + \,{\text{SO}}_{4}^{2 - } \rightleftharpoons {\text{Cd(SO}}_{ 4} )_{2}^{2 - }\)

log10 K2SO4 = 0.96


Equilibrium constants measured at an ionic strength of 4.0 mol·kg−1 are taken from Powell et al. [14]

Note that the fourth consecutive stability constant, K4Cl, for cadmium-chloro complex formation in Table 1 is rather uncertain [14]. Equation 1 connects the concentration of free Cd2+ with that of any of its corresponding chloro complexes:
$$\begin{aligned}{\text{Cd}}^{ 2+ } + m{\text{Y}}^{ - q} \rightleftharpoons {\text{CdY}}_{m}^{(2 - mq)} \\ {\log }_{10} \beta_{mY} = \mathop \sum \limits_{i = 1}^{u} { \log }_{10} K_{iY}\end{aligned}$$

Equation 2, the more general form of Eq. 1, takes into account anions with a higher charge.

Below, chemical equations relevant for the liquid–liquid extractions studied in this work are listed. Aliquat 336 (without its anion) is represented as A336 in the equations, m and n are positive integers (with n − 2 + qm > 0) and q is a positive integer representing the (negative) charge of anion Y. Organic phase constituents are indicated with a bar above the species.

The direct anion exchange with A336 in the organic phase when equilibrated with the aqueous phase can be described as
$$q\overline{{{\text{A336}} \cdot {\text{Cl}}}} + {\text{Y}}^{ - q} \rightleftharpoons \overline{{{\text{A336}}_{q} \cdot {\text{Y}}}} + q{\text{Cl}}^{ - }$$

Importantly, this is valid, with or without Cd being present.

Equation 4 describes the balance between Cd species in the two phases, when A336 is present:
$$\overline{{{\text{A336}}_{n} \cdot {\text{CdCl}}_{n + 2 - qm} {\text{Y}}_{m} }} \rightleftharpoons n\overline{{{\text{A336}} \cdot {\text{Cl}}}} + {\text{Cd}}^{2 + } + \left( {2 - qm} \right){\text{Cl}}^{ - } + m{\text{Y}}^{ - q}$$
Equation 5 illustrates the balance between aqueous cadmium-chloro complex species, free chloride, organic cadmium-chloro complexes bound to A336, and A336 bound to chloride:
$$n\overline{{{\text{A336}} \cdot {\text{Cl}}}} + {\text{CdCl}}_{m}^{(2 - m)} \rightleftharpoons \overline{{{\text{A336}}_{n} \cdot {\text{CdCl}}_{n + m} }} + \left( {m - 2} \right){\text{Cl}}^{ - }$$
Equation 6 relates the organic cadmium-chloro complexes bound to A336 to the more complicated organic complex that might be dominant:
$$\overline{{{\text{A336}}_{n} \cdot {\text{CdCl}}_{n + 2 - qm} {\text{Y}}_{m} }} + qm{\text{Cl}}^{ - } \rightleftharpoons \overline{{{\text{A336}}_{n} \cdot {\text{CdCl}}_{n + 2} }} + m{\text{Y}}^{ - q}$$

By careful addition and/or subtraction of these equations, any organic complex species can be balanced towards any possible complex formed in the water phase. Further, by assuming domination of certain species over the rest, a simple approximate expression involving the phase distribution ratio (explained below) can be obtained. Of course, an exact expression can be derived, but this is not necessary if the dominant species can be pinpointed.

Speciation diagrams for Cd were calculated as a function of sulfate and chloride concentrations based on data from Table 1. For these calculations, the chemical activity coefficient was set to 1, i.e., by assuming ideal solutions. Figure 1 shows the speciation when only one anion, either sulfate or chloride, is present.
Fig. 1

Panel A shows the speciation of Cd with sulfate estimated using Eqs. B3 and B5 (Appendix B). Panel B is the speciation of Cd with chloride using Eqs. B2 and B4 (Appendix B). For both panels, equilibrium constants from Table 1 were used and the activity coefficients were set to unity

Figure 2 shows a similar diagram but for a mixture of sulfate and chloride ions. Here it was assumed that mixed ligand species, such as CdCl2(SO4)2−, will not occur and that Li2SO4 and LiCl are strong electrolytes. If ideal solutions (chemical activity coefficient set to 1) are assumed, then speciation as shown in panel A results. A more realistic calculation that takes into account chemical activity is shown in panel B. Here, the chemical activities were calculated using Pitzer equations according to [16, 17, 18, 19]. Clearly, the deviations from ideal solutions will increase with the ionic concentration. It is important to note that in both simulations (see Fig. 2) there is a drastic shift in the dominant species of Cd at approximately 1 mol·kg−1 sulfate. However, if the sulfate concentration is below 1 mol·kg−1 then cadmium chloride complexes will dominate. Details about the calculation are given in Appendix B.
Fig. 2

Speciation of Cd in a mixture of sulfate and chloride ions. Panel A shows results for ideal solutions with activity coefficients set to unity. Here the speciation was calculated using Eqs. B4, B5 and B6 in Appendix B. Panel B shows results when chemical activities are taken into account and calculated with the Pitzer equations [16, 17, 18, 19]

The curves presented in Figs. 1 and 2 aid in predicting at which concentrations the extraction will be most efficient. As there are several different possible species present, a general equilibrium expression for extraction of cadmium with A336·Cl is presented. This is given by Eq. 4, where n and m are integers satisfying certain constraints (n + mq > 2 and 1 ≤ n) and hence define the stoichiometric constants of the finite number of possible reactions.

The distribution ratio (D) of Cd in the liquid–liquid extraction is defined as the total concentration of all Cd species in the organic phase divided by that of all Cd species in the aqueous phase:
$$D = \frac{{[\overline{{{\text{Cd}}_{\text{org}} }} ]}}{{[{\text{Cd}}_{\text{aq}} ]}}$$
D can be inserted into Eq. 5, with the assumption that the D value is limited to one species dominating in the aqueous phase and one species dominating in the organic phase, leading to:
$$K_{\text{ex}} = \frac{D}{{[\overline{{{\text{A336}} \cdot {\text{Cl}}}} ]^{n} [{\text{Cl}}^{ - } ]^{2 - m} }}$$

Plotting log10D against log10 of the variables (e.g. [Cl]) will give a slope that should represent the stoichiometry of the reaction. If several species are present in the solution, then the extraction data can be used to determine the dominant species interacting with the extractant. Assumption of a dominant cadmium species can then be used to estimate an equilibrium value of the extraction.

3 Experimental Procedure

3.1 Chemicals

Toluene was selected as organic solvent for the liquid–liquid extraction experiments even though it is not used in industrial processes due to its carcinogenic and flammable nature. The reason for using it in this work is that it is well defined and “pure”, contrary to other industrial solvents that frequently are a mixture of different organic compounds. More important was the fact that toluene will not quench the light transfer process of the liquid-scintillation spectroscopy measurements. Thus, high quality spectra from 109Cd could be obtained. In addition, toluene has a well-known Raman spectrum with clear sharp signals.

Toluene used for the extraction experiments was of technical grade (from Univar) while p.a. grade quality (from VWR) was used for the Raman spectroscopy measurements. LiCl (99%), CdCl2 (99.5%), Li2SO4 (99%), NaClO4 (99.0%), and HCl (37% fuming) were all of p.a. grade. Due to the hygroscopic nature of LiNO3 (99%), a saturated solution was made. The solution was gently stirred for a couple of days to ensure equilibrium. The saturated aqueous phase was used in the experiments. A carrier free, aqueous solution of 109Cd was supplied by Ecker and Ziegler. Aliquat 336 (assay 88–90%) was supplied by Alfa Aesar and used as supplied.

The liquid scintillator cocktail was Ultima Gold XR, produced by Packard and used as supplied.

3.2 Liquid–Liquid Extraction

Extractions were performed at ambient temperature (21 ± 2 °C) using equal volumes of the organic and aqueous phases. The phases were prepared by adding the radioactive tracer 109Cd to the inactive aqueous Cd-solution of selected composition and the A336 was diluted with toluene. The two phases were then mixed and shaken for at least 25 min using a vortex shaker. Measurements of D values at various shaking times had beforehand shown that 20 min was sufficient to obtain equilibrium for the distribution of Cd between the two phases. Separation between the aqueous and the organic phases was performed by centrifugation at 4000 rpm for a minimum of 2 min.

Subsequently, the equal volumes of the two phases were sampled and transferred to 20 mL scintillation vials. To avoid organic entrainment in the aqueous phase (the bottom one), the organic and the interphase were removed before sampling of the aqueous phase.

Table 2 shows the concentrations selected for the four systems studied. System 1 was designed to determine the extraction with chloride as the only anion present. System 2 was designed to study the impact of increase in ionic strength on the extraction using LiNO3. Systems 3 and 4 used Li2SO4 and NaClO4, respectively. LiCl was used to vary the chloride concentration for all four systems.
Table 2

The aqueous systems used for Cd extraction with Aliquat 336 in toluene as an extractant

System #

[LiCl] mol·kg−1

Ionic strength mol·kg−1

Matrix salt

Matrix ion mol·kg−1

A336 mmol·kg−1























All solutions have 0.1 mmol·kg−1 Cd and 0.1 mmol·kg−1 HCl

All experiments were performed at a minimum in triplicate.

3.3 Liquid Scintillation Spectroscopy

Scintillation spectroscopy with a Hidex 900 liquid scintillator spectrometer was used to measure the conversion-electron spectrum from the 88 keV transition following decay of 109Cd. The amount of 109Cd in each sample was then determined by integrating the appropriate peak in the spectrum. An efficiency calibration, measuring quenching of the liquid scintillation process as s function of the position of the 88 keV electron conversion line of 109mAg [20], was determined by measuring the shift in peak position when CCl4 was added as quencher [21].

The spectra were analyzed using the peak-fit function in the Origin software package version 9.1 from OriginLab Corporation. D values were determined provided the peak in the spectra for both phases had an area of at least 10 times as high as the background.

3.4 Raman Spectroscopy

The anions under consideration in this study are characterized by having Td (\({\text{ClO}}_{4}^{ - }\), \({\text{SO}}_{4}^{2 - }\)) or D3h (\({\text{NO}}_{3}^{ - }\)) symmetry, with the exception of \({\text{HSO}}_{4}^{ - }\) that is expected only to have Cs symmetry if any at all. The vibrational modes of the two first symmetry classes fall into the groups (Γ = A1 + E + 2T) and (Γ = A1′ + A2′ + 2E′), respectively, while the vibrational modes of \({\text{HSO}}_{4}^{ - }\) belong to the group (Γ = 8A′ + 4A″). However, what these anions have in common is that their most intense Raman active modes (A1, A1′ and A′) are non-degenerate and strongly polarized. Hence, the wavenumber positions of these Raman bands, characterized by being ‘breathing’ modes, all appear as single bands unless they fall exactly at the same place as Raman bands of other species. Therefore, detection of these in a Raman spectrum will serve as an efficient tool for identification of specific anions, both in the water and the organic phases. In principle, infrared spectroscopy can provide similar information, but here band degeneracies and possible splitting may complicate the picture, and for the case of aqueous solutions, the most interesting spectral ranges would be very difficult to deal with because of strong absorption. The ‘breathing mode’ Raman bands contain valuable information about local molecular surroundings, reflected in characteristics of the bandwidths as well as their absolute position and intensities. Molecular association between anions and species can be revealed by studying model solutions [11, 12, 22].

The Raman spectra were recorded with a Horiba Jobin–Yvon T64000 multichannel spectrometer adjusted to work in the triple subtractive mode using three gratings with 1800 rules/mm. The liquid samples were kept in glass vials illuminated in a 90° macro-chamber setup at room temperature (20 °C). The light source for excitation was a Spectra-Physics Millennia Pro 12SJ, Nd:YVO4 laser emitting 200 mW at 532.1 nm, leading to a power of ~ 40 mW at the sample. Both the entrance and the second intermediate slits of T64000 were set to 300 μm. Scattered light was collected by a BIDD CCD cooled to − 125 °C. These settings lead to a spectral width of 6.8 cm−1. No attempt was made to control polarization within these runs. The wavenumber position was calibrated against paracetamol [23].

Table 3 lists the Raman samples with concentration and composition. The organic phases have been in contact with the aqueous phases as described in Sect. 3.2. Aqueous phases were measured without contact with the organic phase.
Table 3

List of Raman samples


Phase measured

Org. conc. mol·kg−1

Aq. conc.



System 1


0.25 A336


System 2


0.25 A336

0.09 LiCl

6.41 LiNO3

System 3


0.25 A336

0.01 LiCl

2.17 Li2SO4

System 4


0.25 A336

0.12 LiCl

6.81 NaClO4

System 2



0.09 LiCl

6.41 LiNO3

System 4



0.12 LiCl

6.81 NaClO4

Samples with both organic (org.) and aqueous (aq.) components, Org. was contacted with written Aq. phases

3.5 NMR Spectroscopy

14N-NMR can be used to identify the nitrogen species present in both the organic and the aqueous phases. For instance, the chemical shift δ at − 320 ppm is assigned to the quaternary nitrogen in A336, which is only present in the organic phase. Likewise, the 14N-chemical shift at δ = 0 is assigned to the nitrate anion in the aqueous phase as well as to the nitrate anion associated with A336 in the organic phase. An underscore will be used to specify which nitrogen is being referred to by writing \(\overline{{\underline{{{\mathbf{A336}}}} \cdot {\mathbf{NO}}_{\mathbf{3} }}}\) or \(\overline{{{\mathbf{A336}} \cdot \underline{{\mathbf{N}}} {\mathbf{O}}_{\mathbf{3}} }}\) (the bar above the compound indicates that it is present in the organic phase).

Since the 14N-signal intensity is directly proportional to the number of 14N nuclei,
$$I_{0} \left[ {\overline{{\underline{{{\text{A336}} \cdot {\text{Cl}}}} }} } \right]_{0} = I\left( {\left[ {\overline{{\underline{{{\text{A}}336}} \cdot {\text{Cl}}}} } \right]} \right) + I\left( {\left[ {\overline{{\underline{{{\text{A336}}}} \cdot {\text{NO}}_{3} }} } \right]} \right)$$
where I denotes the 14N-NMR signal intensity of A336 and \(I_{0} (\overline{{\underline{{{\text{A336}}}} \cdot {\text{Cl}}}} )\) represents the initial NMR signal intensity of A336·Cl within the organic phase, before any extraction has taken place.
Since the 14N-NMR chemical shift of \(\underline{\text{N}} {{\text{O}}}_{3}^{ - }\) is significantly different from the chemical shift of \(\underline{\text{A336}} \cdot {\text{Cl}}\), it is possible to quantitatively differentiate between \(\overline{{\underline{{{\text{A336}}}} \cdot {\text{Cl}}}}\) and \(\overline{{{\text{A336}} \cdot \underline{\text{N}} {{\text{O}}_{3}} }}\). The equilibrium constant K for the ion exchange reaction between A336·Cl and nitrate is:
$$K = \frac{{\left[ {\overline{{{\text{A336}} \cdot {\text{NO}}_{3} }} } \right][{\text{Cl}}^{ - } ]}}{{\left[ {\overline{{{\text{A336}} \cdot {\text{Cl}}}} } \right][{\text{NO}}_{3}^{ - } ]}}$$
Hence, combining Eqs. 9 and 10 and noting that \(\left[ {\overline{{{\text{A336}} \cdot {\text{NO}}_{3} }} } \right]_{0} = 0\) and \(\left[ {\overline{{{\text{A336}} \cdot {\text{Cl}}}} } \right]_{0}\) ≪ \(\left[ {{\text{LiNO}}_{3} } \right]_{0}\) (see Table 4) yields:
Table 4

Sample characteristics of the three solutions studied by 14N-NMR


\([{\text{Cl}}^{ - } ]\)

\(\left[ {{\text{NO}}_{3}^{ - } } \right]\)

\(f_{\text{eq}} = \frac{{\left[ {\overline{{{\text{A336}} \cdot \underline{\text{N}} {{{\text{O}}_{3}}} }} } \right]_{\text{eq}} }}{{\left[ {\overline{{\underline{{{\text{A336}}}} \cdot {\text{Cl}}}} } \right]_{0} }}\)















The initial concentration \(\varvec{ }\left[ {\overline{{{\mathbf{A336}} \cdot {\mathbf{Cl}}}} } \right]_{0}\) was the same in all samples and equal to 0.1 mol·kg−1

$$f_{\text{eq}} = \frac{{\left[ {\overline{{{\text{A336}} \cdot {\text{NO}}_{3} }} } \right]_{\text{eq} }}}{{\left[ {\overline{{{\text{A336}} \cdot {\text{Cl}}}} } \right]_{0} }} = \frac{{I\overline{{(\underline{\text{N}} {{\text{O}}}_{3}^{ - } )}} }}{{I(\overline{{\underline{{{\text{A336}}}} \cdot {\text{Cl}} + \underline{{{\text{A336}}}} \cdot {\text{NO}}_{3} }} )}} = \frac{K}{{K + \frac{{\left[ {\text{LiCl}} \right]_{0} }}{{\left[ {{\text{LiNO}}_{3} } \right]_{0} }}}}$$
here \(f_{\text{eq}}\) shows how much of the total \(\overline{{A336 \cdot {\text{Cl}}}}\) has reacted with nitrate and formed \(\overline{{A336 \cdot {\text{NO}}_{3} }}\). The \(\left[ {\overline{{A336 \cdot {\text{Cl}}}} } \right]_{0}\), \(\left[ {\text{LiCl}} \right]_{0}\) and \(\left[ {{\text{LiNO}}_{3} } \right]_{0}\) represent the initial concentrations of \(\overline{{{\text{A336}} \cdot {\text{Cl}}}}\), LiCl and LiNO3, before any extraction was initiated, and I describes the 14N-NMR signal intensity.

14N-NMR spectra were acquired using a Bruker DRX 500 spectrometer operating at 11.74 T, equipped with a 5 mm BBO (BB/1H/2H) probe with Z-gradient. 5 vol% deuterium oxide (D 99.9%, Cambridge Isotope Laboratories, Inc.) was added to the aqueous phases and 5 vol% benzene-d6 (D 99.6%, Cambridge Isotope Laboratories, Inc.) to the organic phases in order to achieve lock. Spectra were recorded using a single-pulse program with a RF-pulse duration of 13.25 μs (90° pulse), a repetition time of 5.0 s between pulses and an acquisition time of 1.5 s. An exponential apodization of 10 Hz and a zero filling from 32 K to 128 K were applied before the Fourier transformation. The frequency spectrum was base-line corrected before further analysis. All spectra were acquired at 25 °C. The intensity (integrated area) of an NMR peak was determined by fitting certain spectral functions (a linear combination of a Gaussian and a Lorentzian peak function) to the observed NMR spectra by a non-linear least-squares technique.

4 Experimental Results

Experimental data is provided in this chapter. First, the general results of the extraction experiments with the Cd radiotracer will be presented. Then the NMR and Raman results that provide insight into the speciation of the extraction chemistry will be given. A detailed discussion of the results from the different extraction systems will be presented in Sect. 5.

4.1 Liquid–Liquid Extraction

Liquid–liquid extraction experiments were performed at several aqueous conditions as defined in Table 2 (systems 1–4). Figure 3 illustrates how the D values for extraction of Cd from chloride solutions vary for:
Fig. 3

Combined 3D and contour plots fitted to the measured D- values as functions of the various aqueous conditions for extraction of Cd into toluene using Aliquat 336 as extractant. Panel A shows extraction when chloride is used to vary the ionic strength between 0.02 and 5 mol·kg−1. Panel B shows the extraction when LiNO3 is used to keep the ionic strength equal to 6.5 mol·kg−1 and LiCl is varied from 0.1 to 5 mol·kg−1. Panel C shows the extraction of Cd where Li2SO4 is used to keep the ionic strength constant at 6.5 mol·kg−1 and the LiCl concentration is varied from 0.01 to 1 mol·kg−1

  • System 1; with no other anions present (panel A)

  • System 2; in the presence of nitrate (panel B), and

  • System 3; in the presence of sulfate (panel C).

The surfaces in Fig. 3 are computer fitted to the experimentally measured D values, which are tabulated in Appendix A. System 4 was not plotted, as the association between perchlorate and A336 was stronger than for the Cd-chloro complexes and therefore extraction was not viable.

As seen in Fig. 3, adding other anions such as nitrate or sulfate perturbs the extraction behaviour quite significantly. This is especially noticeable for nitrate. The reason is most likely because either the extractant is saturated with species having stronger affinity to the A336 extractant than Cd chloro-complexes, or a different and less extractable Cd complex is created.

In order to determine exactly in which way nitrate and sulfate influence the extraction, Raman and NMR spectroscopy were used to gain more detailed insight into the systems. Below, the details of the Raman and NMR spectroscopy will be presented. Afterwards, using the Raman and NMR results, a detailed discussion of the extraction data is provided.

4.2 Raman Spectroscopy

Raman spectroscopy can identify specific anions both in aqueous and organic phase. For the present study the detectable anions of interest are \({\text{ClO}}_{4}^{ - }\), \({\text{SO}}_{4}^{2 - }\), and \({\text{NO}}_{3}^{ - }\). Below, the experimental results from the Raman spectroscopy are presented.

Figure 4 shows the Raman spectra of nitrate (panel A) and perchlorate (panel B) in the organic and the aqueous phase.
Fig. 4

Panel A: the Raman spectrum of toluene containing only Aliquat 336 (A336) (full line), the organic phase in an equilibrium mixture with \(\overline{{{\mathbf{A336}} \cdot {\mathbf{Cl}}}}\) and \({\mathbf{LiNO}}_{\mathbf{3}}\) nitrate (dotted) revealed by the A1′ band at 1040 cm−1, and the spectrum of aqueous LiNO3 (dot dash). Panel B: the Raman spectrum of toluene containing only Aliquat 336 (A336) (full line), the organic phase in an equilibrium mixture with \(\overline{{{\mathbf{A336}} \cdot {\mathbf{Cl}}}}\) and NaClO4 (dotted) showing the A1 mode of perchlorate at 930 cm−1 and finally the spectrum of aqueous NaClO4 (dot dash)

For water, the spectral range from 850 to 1100 cm−1 is Raman silent. Toluene has a strong A1 band at 1003.6 cm−1 and a medium intensity band following the same symmetry at 1030.6 cm−1. In water, the A1′ stretching band of nitrate falls at 1049 cm−1 as expected, but is found to be shifted to 1040 cm−1 in the organic phase. Comparison with the A336·Cl spectrum leaves no doubt that this is specific for nitrate occurring in some form. For the perchlorate experiment (Fig. 4, panel B), the picture is similar, but the shift is smaller. Here the A1 mode of perchlorate in the organic phase appears at 930 cm−1, a clear sign that anion exchange with the organic phase by A336 again has happened. However, the observed shift is smaller than for nitrate, as the perchlorate A1 band falls at 936 cm−1 in the aqueous phase. The A1 bandwidths of perchlorate and nitrate both decrease upon transfer to the organic phase. The decrease is most prominent for perchlorate. An explanation for this could be that the affinity of perchlorate to A336 is stronger and better defined than for nitrate, and that perchlorate is less susceptible to changes in molecular surroundings in the aqueous phase. Also, a weaker binding of nitrate to A336 will lead to an increase in the number of possible conformations and, hence, a larger spectral broadening. Experiments with sulfate solutions were carried out, but here no A1 peak could be detected in the Raman spectra of the organic phases. This indicates that the affinity of sulfate to A336 is very weak, or at least so weak that the concentration remains below the detection limit of Raman spectroscopy, which under this framework is assumed to be around 1 mmol·kg−1. A qualitative estimate of the sulfate band position, based on the nitrate and perchlorate properties, indicates that the A1 band in the organic phase should be found 5–10 cm−1 below 981 cm−1 that is typical for aqueous solutions.

4.3 NMR Spectroscopy

14N-NMR can be used to study the extraction of nitrate between the phases. This was explored to gain insight into the extraction systems presented in this work. The experimental details from the NMR experiments are presented below.

Figure 5 presents the 14N-NMR spectra for:
Fig. 5

14N-NMR spectra of toluene containing only \(\overline{{{\mathbf{A336}} \cdot {\mathbf{Cl}}}}\) (green broken line), the aqueous phase in an equilibrium mixture of \(\overline{{{\mathbf{A336}} \cdot {\mathbf{Cl}}}}\) and \({\mathbf{LiNO}}_{\mathbf{3}}\) (blue dotted line) and the organic phase in the equilibrium mixture of \(\overline{{{\mathbf{A336}} \cdot {\mathbf{Cl}}}}\) and \({\mathbf{LiNO}}_{\mathbf{3}}\) (red solid line) (Color figure online)

  • organic phase containing only A336·Cl,

  • aqueous phase in an equilibrium mixture of A336·Cl and LiNO3, and

  • organic phase in an equilibrium mixture of A336·Cl(org) and LiNO3(aq).

As can be inferred from Fig. 5, \(\overline{{\underline{{{\text{A336}}}} \cdot {\text{Cl}}}}\) reveals two distinct peaks in the 14N-NMR spectrum (green broken line), a narrow peak at a chemical shift δ ≈ − 320 ppm with a half-width of less than 35 Hz and a much broader peak at δ ≈− 330 ppm with a half width of the order of 400 Hz. The narrow peak is the response from A336 and the broad peak is the response of triethylamine used as a reference.

Generally, the 14N-NMR spectrum of the aqueous phase (blue dotted curve) reveals only a single peak, which is consistent with the presence of \(\underline{\text{N}} {{\text{O}}}_{3}^{ - }\). No 14N-peak from \(\underline{\text{A336}} \cdot {\text{Cl}}\) or \(\underline{\text{N}} {{\text{O}}}_{ 3}\) is observed and this simply confirms the lack of any significant solubility of A336 in the aqueous phase. In contrast, the 14N-NMR spectrum (red solid curve) of the organic phase, which was formed after a thorough mixing of the organic solution, \(\overline{{{\text{A336}} \cdot {\text{Cl}}}}\) and the aqueous solution (LiNO3), revealed the presence of both \(\overline{{{\text{A336}} \cdot {\text{Cl}}}}\) and \(\overline{{{\text{A336}} \cdot {\text{NO}}_{3} }}\). The presence of \(\overline{{{\text{A336}} \cdot {\text{Cl}}}}\) was indirectly confirmed by noting the difference between the 14N-peak intensities of \(\overline{{\underline{{{\text{A336}}}} \cdot {\text{Cl}} + \underline{{{\text{A336}}}} \cdot {\text{NO}}_{3} }}\) and \(\overline{{{\text{A336}} \cdot \underline{\text{N}} {{\text{O}}_{3}} }}\), respectively.

Moreover, the ratio \(\left[ {\overline{{{\text{A336}} \cdot \underline{\text{N}} {{\text{O}}_{3}} }} } \right]_{\text{eq}} /\left[ {\overline{{\underline{{{\text{A336}}}} \cdot {\text{Cl}}}} } \right]_{0}\) is easily determined from the 14N-NMR spectrum of the organic phase as it is simply represented by the signal intensity ratio between the 14N-NMR resonance peak at δ ≈ 0 and the resonance peak within the region − 320 ppm < δ < − 310 ppm (Fig. 5), at equilibrium. Table 4 summarizes the sample characteristics of the three samples investigated by 14N–NMR. A plot of \(\left[ {\overline{{{\text{A336}} \cdot \underline{\text{N}} {{\text{O}}_{3} }}} } \right]_{\text{eq}} /\left[ {\overline{{\underline{{{\text{A336}}}} \cdot {\text{Cl}}}} } \right]_{0}\) against \(\left[ {\text{LiCl}} \right]_{0} /\left[ {{\text{LiNO}}_{3} } \right]_{0}\) is presented in Fig. 6. The solid curve represents the non-linear least-squares fit of Eq. 11 to the observed data and results in an equilibrium value of Kex0 = (18 ± 5). The large uncertainty (28%) stems partly from the large distance between the second and third data point, but also from the uncertainty in each data point (15%).
Fig. 6

\(\left[ {\overline{{{\mathbf{A336}} \cdot \underline{{\mathbf{N}}} {\mathbf{O}}_{\mathbf{3} }}} } \right]_{{{\mathbf{eq}}}} /[\overline{{\underline{{{\mathbf{A336}}}} \cdot {\mathbf{Cl}}}} ]_{\mathbf{0}}\) versus \(\left[ {{\mathbf{LiCl}}} \right]_{\mathbf{0}} /\left[ {{\mathbf{LiNO}}_{\mathbf{3}} } \right]_{\mathbf{0}}\) as derived from 14N-NMR spectral analysis (see Fig. 5). The standard error in NMR intensity measurements is of the order of ± 10%, resulting in a standard error in \(\varvec{f}_{{{\mathbf{eq}}}}\) of approximately ± 15% (see error bars). The solid curve represents a non-linear least-squares fit of Eq. 11 to the observed data (red circle) (Color figure online)

Although the number of data points is small and the relative error is high, it can be concluded that the association between nitrate and A336 is stronger than the corresponding association between \({\text{Cl}}^{ - }\) and A336. This is supported by previous work [24]. For instance, by considering extraction of nitrate and applying Eq. 11 with Kex0 = 18 ± 5 (log10 Kex0 =  \(1.26 \pm 0.12\)) on a solution containing initially only \(\overline{{{\text{A336}} \cdot {\text{Cl}}}}\) and \({\text{NO}}_{3}^{ - }\) with \([\overline{{{\text{A336}} \cdot {\text{Cl}}]}}_{0} = \left[ {{\text{NO}}_{3}^{ - } } \right]_{0}\) ≫ \([{\text{A336}} \cdot {\text{Cl}}]\), it can be shown that more than 90% of \(\overline{{{\text{A336}} \cdot {\text{Cl}}}}\) will be converted to \(\overline{{{\text{A336}} \cdot {\text{NO}}_{3} }}\).

5 Discussion

In this chapter, the three extraction systems that were measured and presented in Fig. 3 will be discussed in detail, using the results from the NMR and Raman spectroscopy to gain insight into the speciation and extraction mechanisms.

5.1 System with Chloride as the Only Anion

Here an analysis of system 1 (defined in Table 2) with variable chloride concentration and variable ionic strength is presented.

Figure 7 shows the extraction of Cd from a pure chloride media. Panel A presents D values as function of the A336 concentration, panel B presents D values as a function of chloride concentration. These data can be discussed in the contexts of Eq. 5: Panel A (Fig. 7) shows that one mole of A336 is needed to extract one mole of Cd. Since it is only chloride that participates in the equilibrium. Eq. 5 becomes:
$$\overline{{{\text{A336}} \cdot {\text{Cl}}}} + {\text{Cd}}^{2 + } + 2{\text{Cl}}^{ - } \rightleftharpoons \overline{{{\text{A336}} \cdot {\text{CdCl}}_{3} }}$$
A336 is an anion exchanger, meaning that to extract Cd the Cd-chloro complex must be negatively charged, i.e., contain at least three chloride ions. Therefore, m in Eq. 5 must be 2 (the third chloride is delivered by the A336·Cl complex). If this model is correct then the slope of log10 D versus log10 [Cl] should be approximately 2. This is not in agreement with the experimental data (Fig. 7).
Fig. 7

Extraction of aqueous phase with 0.1 mol·kg−1 Cd extracted from a LiCl solution using Aliquat 336. Panel A shows the D values as a function of A336 concentration. Panel B shows the D values as a function of the chloride concentration. For both panels, straight lines with slope + 1 are drawn through the datasets. The straight lines represent a one-to-one stoichiometric ratio. D values with arrows indicate that the real D value should be higher due to the concentration of Cd in the aqueous phase being below the limits of detection

Figure 1 (panel B) suggested that between chloride concentrations of 0.1 and 1 mol·kg−1 the species of CdCl+ and CdCl2 dominate. This could imply that that there is no free Cd2+ present and therefore Eq. 12 can be written as:
$$\overline{{{\text{A336}} \cdot {\text{Cl}}}} + {\text{CdCl}}^{ + } + {\text{Cl}}^{ - } \rightleftharpoons \overline{{{\text{A336}} \cdot {\text{CdCl}}_{3} }}$$
i.e., the stoichiometric ratio between A336·Cl, free Cl and CdCl+ should be 1:1:1, in fair agreement with the experimental data shown in Fig. 7. Equation 7 takes into account all possible organic species and aquatic species for the extraction, the assumption is then that it can be written as:
$${{D}} = \frac{{\overline{{\left[ {A336 \cdot {\text{CdCl}}_{3} } \right]}} + \overline{{\left[ {A336_{2} \cdot {\text{CdCl}}_{4} } \right]}} + \overline{{\text{[CdCl}_{2} ]}} }}{{\mathop \sum \nolimits_{0}^{n} [{\text{CdCl}}_{{n}} ]}} \approx \frac{{\overline{{[A336 \cdot {\text{CdCl}}_{3} ] }} }}{{[{\text{CdCl}}^{ + } ]}}$$
If Eq. 13 is combined with Eq.14 the equilibrium constant can be estimated as:
$$K_{{{\text{ext}}1}} = \frac{D}{{\left[ {\overline{{{\text{A336}} \cdot {\text{Cl}}}} } \right]_{0} \left[ {{\text{Cl}}^{ - } } \right]_{0} }}$$

If Eq. 14 is assumed to hold, then the equilibrium constant Kext1 can be calculated using Eq. 15. This results in log10 Kext1 = 4.9 ± 0.8. If chemical activities are taken into account then the resulting value is log10 Kp1 = 5.1 ± 0.4

5.2 System with Chloride and Nitrate

Here an analysis of system 2 (defined in Table 2) is presented. The ionic strength was kept constant at 6.5 mol·kg−1 by using LiNO3 and LiCl.

Figure 8 shows the extraction of Cd with both nitrate and chloride present. From the Raman and NMR experiments there is clear evidence that nitrate replaces chloride in the A336 complex. The extraction curves show that roughly 1 mol of chloride is extracted per mole of Cd, as indicated by fitting the extraction data in the double logarithmic plot with a straight line with slope + 1 (Fig. 8, panel B).
Fig. 8

Extraction of the aqueous phase with 0.1 mmol·kg−1 Cd with an ionic strength of 6.5 mol·kg−1. Panel A shows the D values as function of the A336 concentration, panel B shows D as function of increasing chloride (and decreasing nitrate) concentration, and panel C shows D as function of increasing nitrate (and decreasing chloride) concentration. The straight lines (slope = + 1) in panels A and B represent a one-to-one stoichiometric ratio. The straight lines (slope − 2) in panel C represent a one-to-two stoichiometric ratio between cadmium and nitrate

Nitrate hinders extraction: i.e., a high nitrate concentration will exchange nitrate with chloride (Eq. 3, \({\text{Y}}^{ - q}\) equal to \({\text{NO}}_{3}^{ - }\)) and more Cd will remain in the aqueous phase. Remember that the NMR results (Sect. 4.3) indicated that nitrate will replace chloride ions in A336·Cl to a large extent (> 90%). Based on panel A (Fig. 8) it can be assumed that one A336 is required per Cd extracted. Each Cd will be complexed by one chloride ion, in accordance with panel B. From panel C it can be deduced that two nitrate ions are needed to “inhibit” one A336 ion. Therefore, a suggested chemical equation that fits these parameters can be written as:
$$\overline{{{\text{A336}} \cdot {\text{NO}}_{3} \cdot {\text{LiNO}}_{3} }} + {\text{CdCl}}_{2} + {\text{Cl}}^{ - } \rightleftharpoons \overline{{{\text{A336}} \cdot {\text{CdCl}}_{3} }} + {\text{Li}}^{ + } + 2{\text{NO}}_{3}^{ - }$$
$$K_{{{\text{ext}}2}} = \frac{{D\left[ {{\text{Li}}^{ + } } \right]_{0} \left[ {{\text{NO}}_{3}^{ - } } \right]_{0}^{2} }}{{\left[ {\overline{{A336 \cdot {\text{NO}}_{3} \cdot {\text{LiNO}}_{3} }} } \right]_{0} \left[ {\text{Cl}} \right]_{0} }}$$

Using Eq. 16 and data from Fig. 8, and assuming that all A336·Cl have been replaced, a corresponding equilibrium constant Kext2 can be calculated. This results in log10 Kext = 4.2 ± 0.3. If chemical activity is taken into account the result becomes log10 Kext = 4.7 ± 0.3. Wang and Hemmes [25] suggest that LiNO3 will readily form in tetrahydrofuran with an equilibrium constant larger than log10 Kext = 9. In addition, Ref. [26] suggests that the equilibrium constants for LiNO3 in tetramethylurea, methylcyanate and isopentyl alcohol are log10 Kext = 2.22, log10 Kext = 3.35 and log10 log10 Kext = 0.58, respectively. It should be noted that Eq.15 is only presented as a hypothesis. The equation agrees with the experimental data, but it does not prove that the hypothesis necessarily is correct. Further investigations are needed and will be carried out in future work.

5.3 System with Chloride and Sulfate

Here an analysis of system 3 (defined in Table 2) is presented. The ionic strength was kept fixed at 6.5 mol·kg−1 by using LiCl and Li2SO4.

Figure 9 shows the extraction of Cd with both sulfate and chloride present. Notice that for low chloride concentration the sulfate concentration will be high and vice versa. From the Raman experiments there is no evidence of sulfate occurring in the organic phase, and Figs. 1 and 2 show that sulfate has a high affinity for Cd. This indicates that sulfate interacts with Cd rather than with the extractant. From Fig. 9 it seems that the change from an easily extractable chloride species to a less extractable (by A336) sulfate species starts to dominate when the chloride concentration is below 0.1 mol·kg−1. Based on this the following extraction equation is proposed:
$$\overline{{{\text{A336}} \cdot {\text{Cl}}}} + {\text{Cd}}^{2 + } + 2{\text{Cl}}^{ - } \rightleftharpoons \overline{{{\text{A336}} \cdot {\text{CdCl}}_{3} }}$$
This formula dictates that log10 D versus log10 [Cl] should follow a straight line with slope of + 2, as indeed it does (Fig. 9, panel A). Similarly, a slope of + 1 is expected for log10 D versus log10 [A336], which is also in agreement with the experimental data (Fig. 9, panel B). Expressing the equilibrium constant from Eq. 12 can then be done as:
$$K_{{{\text{ext}}3}} = \frac{D}{{\left[ {\overline{{{\text{A336}} \cdot {\text{Cl}}}} } \right]_{0} \left[ {{\text{Cl}}^{ - } } \right]_{0}^{2} }}$$
Using Eq. 17 to calculate the equilibrium constant for the chloride/sulfate system results in log10 Kext3 = 5.7 ± 0.3 and, by taking chemical activity into account, the result is log10 Kp3 = 6.5 ± 0.5.
Fig. 9

Extraction of aqueous phase with 0.1 mmol·kg−1 Cd with an ionic strength of 6.5 mol·kg−1. D values shown as a function of increasing chloride concentration (and decreasing sulfate) (panel A), and of A336 concentration (panel B). Lines in panel A have been drawn with a slope of + 2 and lines in panel B have been drawn with a slope of + 1

5.4 Comparison of K Values

As a check of consistency in our data and the validity of the assumptions leading to Eqs. 14, 16, and 17, the equilibrium constants calculated from the D values to each individual extraction in Figs. 7, 8, and 9 are plotted in Fig. 10. Ideally, the equilibrium constants should show no trends. Figure 10 shows the different equilibrium constants as a function of D ratio. The equilibrium constants have been calculated both with the assumption that the chemical activity coefficient is unity and by using Pitzer equations to estimate the real activity coefficient (\(\upgamma\)) [17, 18, 19]. Panel A shows the values when only chloride was present and the ionic strength was varied. The results suggest that Eq. 13 is valid for 0.02 and 0.5 mol·kg−1 chloride concentrations and A336 concentrations from 1 up to 60 mmol·kg−1. The validity was checked using the student’s t test. Panel B shows the extraction as the ionic strength is kept constant using nitrate. In that system the equilibrium constants does not diverge significantly. Panel C shows the equilibrium constant for the sulfate system. No significant change is observed for D values below 40, which suggest that Eq. 12 fits until the chloride concentration approaches 1 mol·kg−1, where a different mechanism starts to dominate.
Fig. 10

Calculation of the equilibrium constants for the extractions from various media versus D ratios. Panel A is for the LiCl medium, panel B is for the LiCl, LiNO3 mixture and panel C is for the LiCl, Li2SO4 mixtures

Table 5 summarizes the equilibrium constants for Cd extraction. Extraction from a perchlorate matrix (System 4 in Table 2) was measured, but as noted in Sect. 4.1 no extraction was observed as perchlorate has a stronger affinity to A336 than the Cd species. This is in accordance with the observation of De et al. [24] who reported that nitrate’s affinity for A336 is lower than perchlorate’s.
Table 5

the concentration ranges where the calculated equilibrium values are valid


[Cl] mol·kg−1

[A336] mol·kg−1

K extn \(\mu = [] \cdot 1\)

K pn \(\mu = [] \cdot \gamma\)

1[Cl] Kext1



4.9 (0.8)

5.1 (0.4)

2[Cl, \({\text{NO}}_{3}^{ - }\)] Kext2



4.2 (0.3)

4.7 (0.3)

3[Cl, \({\text{SO}}_{4}^{2 - }\)] Kext3



5.7 (0.4)

6.5 (0.5)

Values are shown both for ideal solutions (\(\varvec{\gamma}\) = 1) and when the chemical potentials were calculated using the Pitzer equations

6 Conclusions

Solutions of the quaternary amine A336 dissolved in toluene were used to extract Cd from a large number of different aqueous solutions containing (1) chloride of different concentrations (and hence different ionic strengths), and mixed Li salts of: (2) chloride and nitrate, (3) chloride and sulfate, and (4) chloride and perchlorate, respectively. Notably, solutions (2), (3), (4) were all constrained at the same high ionic strength of 6.5 mol·kg−1.

One objective of the present work was to identify any simple reaction scheme that may describe the extraction of Cd within certain concentration “windows” of the reactants. In order to accomplish this, the three-dimensional data matrices (characterized by the distribution ratio, the A336 concentration, and the salt concentrations) were converted into two-dimensional matrices by simply keeping one of the concentration variables fixed.

Notably, it was found that NMR spectroscopy provided quantitative information on the distribution of anions between phases as well as giving insight into the association property or affinity of A336 for various anions, i.e.: perchlorate > nitrate > chloride > sulfate, an observation which was found useful in explaining the extraction properties within the different systems.

Moreover, 14N-NMR spectroscopy allowed the equilibrium constant Kex0 of the reaction: \(\overline{{{\text{A}}336 \cdot {\text{NO}}_{3} }} + {\text{Cl}}^{ - } \rightleftharpoons \overline{{{\text{A}}336 \cdot {\text{Cl}}}} + {\text{NO}}_{3}^{ - }\) to be determined as Kex0 = 18 ± 5 (or log10 Kex0 = \(1.26 \pm 0.12\)).

Based on the two-dimensional data reduction analysis approach, it was concluded that for chloride concentrations within the concentration range of 0.01–0.5 mol·kg−1 and with no other anions present in the solution, the species CdCl+ dominated in the extraction reaction. Hence, the following simplified reaction scheme: \(\overline{{{\text{A}}336 \cdot {\text{Cl}}}} + {\text{CdCl}}^{ + } + {\text{Cl}}^{ - } \rightleftharpoons \overline{{{\text{A}}336 \cdot {\text{CdCl}}_{3} }}\) (dominated).

If both sulfate and chloride were present at high ionic strength (6.5 mol·kg−1) and the sulfate concentration increased (at the expense of the chloride concentration, which decreased correspondingly at constant ionic strength), then a significant drop in the D values of Cd was observed. This was found not to be caused by sulfate being associated to A336, since no sulfate was detected in the organic phase after extraction, as confirmed by Raman spectroscopy. Instead, it seems that formation of sulfate species started to dominate when the chloride concentration decreased below 0.1 mol·kg−1. If both sulfate and chloride were present at high ionic strength (6.5 mol·kg−1), then the main reaction taking place was identified as: \(\overline{{{\text{A}}336 \cdot {\text{Cl}}}} + {\text{Cd}}^{2 + } + 2{\text{Cl}}^{ - } \rightleftharpoons \overline{{{\text{A}}336 \cdot {\text{CdCl}}_{3} }}\). Finally, the extraction of Cd in the presence of both nitrate and chloride suggested that the following major reaction takes place: \(\overline{{{\text{A}}336 \cdot {\text{NO}}_{3} \cdot {\text{LiNO}}_{3} }} + {\text{CdCl}}_{2} + {\text{Cl}}^{ - } \rightleftharpoons \overline{{{\text{A}}336 \cdot {\text{CdCl}}_{3} }} + \overline{{{\text{A}}336 \cdot {\text{Cl}}}} + {\text{Li}}^{ + } + 2{\text{NO}}_{3}^{ - }\)

The current work demonstrated that both NMR and Raman spectroscopy will be important tools in future characterizations of Cd-extraction from real industrial solutions of high salinity. In particular, the potential of using multinuclear NMR (7Li, 14N, 13C, 1H, 35Cl, 31P and 113Cd) will be considered.



The authors are grateful for the financial support from the Norwegian Research Council and industry companies Yara International, Glencore Nikkelverk, and Boliden Odda. The support was channelled through the Norwegian Research Council project BIA-KPN, Project No. 2366741. We also appreciate the valuable input and constructive discussions with representatives from our industry partners. Thanks to the University of Oslo NMR laboratory for running the NMR experiments.


  1. 1.
    Mudd, G.M.: The sustainability of mining in australia: key production trends and their environmental implications for the future. In:Monash University and Mineral Policy Institute, pp. 1–277 (2009)Google Scholar
  2. 2.
    Safarzadeh, M.S., Bafghi, M.S., Moradkhani, D., Ilkhchi, M.O.: A review on hydrometallurgical extraction and recovery of cadmium form various resources. Miner. Eng. 20, 211–220 (2007)CrossRefGoogle Scholar
  3. 3.
    Jha, M.K., Kumar, V., Jeong, J., Lee, J.-C.: Review on solvent extraction of cadmium from various solutions. Hydrometallurgy 111–112, 1–9 (2012). CrossRefGoogle Scholar
  4. 4.
    Sato, T., Shimomura, T., Murakami, S., Maeda, T., Nakamura, T.: Liquid–liquid extraction of divalent manganese, cobalt, copper, zinc and cadmium from aqueous chloride solutions by tricaprylmethylammonium chloride. Hydrometallurgy 12, 245–254 (1984)CrossRefGoogle Scholar
  5. 5.
    Daud, H., Cattrall, R.W.: The extraction of Hg(II) from potassium iodide solutions and the extraction of Cu(II), Zn(II) and Cd(II) from hydrochloric acid solutions by Aliquat 336 dissolved in chloroform. J. Inorg. Nucl. Chem. 43, 779–785 (1980)CrossRefGoogle Scholar
  6. 6.
    Wassink, B., Dreisinger, D., Howard, J.: Solvent extraction separation of zinc and cadmium from nickel and cobalt using Aliquat 336, a strong base anion exchanger in the chloride and thiocyanate horms. Hydrometallurgy 57, 235–252 (2000)CrossRefGoogle Scholar
  7. 7.
    McDonald, C.W., Moore, F.L.: Liquid–liquid extraction of cadmium with high-molecular-weight amines from iodide solutions. Anal. Chem. 45(6), 983–985 (1973)CrossRefGoogle Scholar
  8. 8.
    Daud, H., Cattrall, R.W.: The extraction of Cd(II) and Zn(II) from acidified lithium chloride solutions by Aliquat 336 in chlorform. J. Radioanal. Nucl. Chem. 43, 599–601 (1980)CrossRefGoogle Scholar
  9. 9.
    Singh, O.V., Tandon, S.N.: Extraction of cadmium as chloride by high molecular weight amines and quaternary ammonium salt. J. Inorg. Nucl. Chem. 37, 609–612 (1975)CrossRefGoogle Scholar
  10. 10.
    Miller, J.D., Fuerstenau, M.C.: Hydration effects in quaternary amines extraction systems. Metall. Trans. B 1, 2531–2535 (1970)Google Scholar
  11. 11.
    Goggin, P.L., Woodward, L.A.: Ramanspectra of methyl mercuric perchlorate and nitrate in solution. Trans. Faraday Soc. 58, 1495–1502 (1962)CrossRefGoogle Scholar
  12. 12.
    Angus, W.R., Leckie, A.H., Ramsay, S.W.: Investigations of Raman spectra Part I—The Raman spectra of sulphuric, nitric and nitrosylsulphuric acids. Proc. Roy. Soc. A 149(867), 327–340 (1935)CrossRefGoogle Scholar
  13. 13.
    Reddy, B.R., Priya, D.N., Kumar, J.R.: Solvent extraction of cadmium(II) from sulphate solutions using TOPD 99, PC 88A, Cyanex 272 and their mixtures. Hydrometallurgy 74, 277–283 (2004)CrossRefGoogle Scholar
  14. 14.
    Powell, K.J., Brown, P.L., Byrne, R.H., Gajda, T., Hefter, G., Leuz, A.-K., Sjöberg, S., Wanner, H.: Chemical speciation of environmentally significant metals with inorganic ligands. Part 4: The Cd2+ + OH, Cl, \({\text{CO}}_{ 3}^{ 2- } ,\;{\text{SO}}_{ 4}^{ 2- } ,\;{\text{and}}\;{\text{PO}}_{ 4}^{ 3- }\) systems (IUPAC Technical Report). Pure Appl. Chem. 83, 1163–1214 (2011).
  15. 15.
    Helm, L., Merbach, A.E.: Water exchange on metal ions: experiments and simulations. Coord. Chem. Rev. 187, 151–181 (1999)CrossRefGoogle Scholar
  16. 16.
    Kim, H.T., Fredercik Jr., W.J.: Evaluation of Pitzer Ion interaction parameters of aqueous electrolytes at 25 °C. 1. Single salt parameters. J. Chem. Eng. Data 33, 177–184 (1988)CrossRefGoogle Scholar
  17. 17.
    Pitzer, K.S.: Thermodynamics of electrolytes. I. Theoretical basis and general equations. J. Phys. Chem. 77, 268–277 (1974)CrossRefGoogle Scholar
  18. 18.
    Pitzer, K.S., Mayorga, G.: Thermodynamics of electrolytes. II. Activity and osmotic coefficient for strong electrolytes with one or both ions univalent. J. Phys. Chem. 77, 2300–2308 (1973)CrossRefGoogle Scholar
  19. 19.
    Pitzer, K.S., Kim, J.J.: Thermodynamics of electrolytes. IV. Activity and osmotic coefficient for mixed electrolytes. J. Am. Chem. Soc. 96, 5701–5707 (1974)CrossRefGoogle Scholar
  20. 20.
    Kristiansen, H.: Radiochemical analysis of cadmium in nitric and phosphoric acid. Masters thesis, University of Oslo (2015)Google Scholar
  21. 21.
    Pacer, R.A.: Liquid scintillation and Cherenkov counting characteristics of 109 Cd. J. Radioanal. Nucl. Chem. Lett. 155(2), 129–140 (1991)CrossRefGoogle Scholar
  22. 22.
    Myhre, C.E.L., Christensen, D.H., Nicolaisen, F.M., Nielsen, C.J.: Spectroscopic study of aqueous H2SO4 at different temperatures and compositions: variations in dissociation and optical properties. J. Phys. Chem. 107, 1979–1991 (2003)CrossRefGoogle Scholar
  23. 23.
    McCreery, R.L.: Raman Spectroscopy for Chemical Analysis. Wiley, New York (2000)CrossRefGoogle Scholar
  24. 24.
    De, A.K., Khopkar, S.M., Chalmers, R.A.: Solvent Extraction of Metals. Van Nostrand Reinhold Series in Analytical Chemistry (1970)Google Scholar
  25. 25.
    Wang, H.C., Hemmes, P.: Ionic association in low-dielectric media. I. Ultrasonic absorption studies of nitrates in tetrahydrofuran. J. Am. Chem. Soc. 95, 5115–5119 (1973)CrossRefGoogle Scholar
  26. 26.
    Högfeldt, E.: Stability Constants of Metal-ion Complexes Part A: Inorganic Ligands, 2nd edn. IUPAC Chemical Data Series. Pergamon Press, Oxford (1983)Google Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2018

Authors and Affiliations

  • Hans Vigeland Lerum
    • 1
  • Niels Højmark Andersen
    • 1
  • Dag Øistein Eriksen
    • 1
  • Eddy Walther Hansen
    • 1
  • Dirk Petersen
    • 1
  • Grethe Wibetoe
    • 1
  • Jon Petter Omtvedt
    • 1
  1. 1.Department of ChemistryUniversity of OsloOsloNorway

Personalised recommendations