FIELD OF THE INVENTION

The present invention relates to a new solvent extraction process for the extraction of rare-earth elements and the separation of a mixture of rare-earth elements into the individual rare- earth elements. Said process comprises the extraction of at least one rare-earth element from an organic phase A into another immiscible organic phase B. The invention further relates to a non-aqueous solvent extraction process wherein at least one rare-earth element is extracted from an organic phase A that comprises chloride ions into another immiscible organic phase B, wherein said organic phase B can be a non-polar organic solvent that comprises an extractant such as trialkylphosphine oxide ( for instance Cyanex 923).

BACKGROUND

Solvent extraction is widely used for the separation of mixtures of rare-earth elements (REEs) on an industrial scale. However, the separation processes are not very efficient because the separation factors between adjacent REEs are small, on average only 1.5 to 2. Often more than 1000 mixer- settlers are required to separate a mixture containing all the REEs into individual REEs with a purity of more than 99.9%. Additionally, most flow sheets for the separation of REEs are not environmentally friendly. When the separations are carried out with acidic extractants such as substituted phosphoric, phosphonic or phosphinic acids, large amounts of chemicals are consumed for pH control and for stripping of REEs from the loaded organic phase, and also large volumes of waste water are generated. Another issue with the use of some acidic extractants such as bis(2- ethylhexylphosphoric acid) (D2EHPA) is that the concentration of the REEs in the loaded organic phase cannot be too high, because otherwise gel formation will occur. Therefore, the aqueous feed solution cannot be too concentrated and large volumes of feed solution have to be used for separation of bulk quantities of REEs. When neutral extractants such as tri-«-butyl phosphate (TBP) or basic extractants such as Aliquat 336 are used, much higher metal loadings of the organic phase are possible and stripping can be achieved easily by water, but these extractants can extract REEs only efficiently from nitrate feed solutions and not from chloride feed solutions. HNO3 is much more expensive than HC1, and treatment of nitrate-containing aqueous waste streams is more difficult than treatment of chloride-containing aqueous waste streams.

The difficulties related to the separation of REEs makes that solvent extraction of REEs are an active field of research. Different routes are being explored. A first approach is the addition of a second extractant to the organic phase to enhance the distribution ratios and the separation factors. This approach is called synergistic solvent extraction. A second approach is the addition of complexing agents to the aqueous phase. A third approach is to design new ligands for the extraction of metals. Undiluted ionic liquids are starting to attract more and more attention for the separation and purification of REEs. All these research efforts are focusing on modifications of the existing aqueous and organic phase in conventional solvent extraction systems.

Recently, a new approach to the separation of mixtures of REEs was introduced, by replacing the aqueous phase by a non-aqueous phase (N. K. Batchu, T. Vander Hoogerstraete, D. Banerjee and K. Binnemans, Non-aqueous solvent extraction of rare-earth nitrates from ethylene glycol to n- dodecane by Cyanex 923, Sep. Purif. TechnoL, 2017, 174, 544-553.). In particular, it was shown that REEs can be extracted from a polar phase containing ethylene glycol, lithium nitrate and REE nitrates to a less polar phase consisting of Cyanex 923 diluted in «-dodecane. Cyanex 923 is a commercial mixture of trialkylphosphine oxides, and it has often been used for solvent extraction of REEs. The extraction of REE nitrates from ethylene glycol (+L1NO3) was different from the extraction of these salts from aqueous solutions, in the sense that the separation factors between light rare-earth elements (LREEs) and heavy rare-earth elements (HREEs) were much larger in the ethylene glycol system. This non-aqueous solvent extraction system could be applied for the group separation of HREEs from LREEs, whereas such separations are not possible for extraction from aqueous nitrate solutions. As mentioned above, it is known that REEs cannot be efficiently extracted from aqueous chloride feed solutions by solvating extractants such as Cyanex 923; very low distribution ratios and separation factors are observed. Using the approach by Batchu et al. (Sep. Purif. TechnoL, 2017, 174, 544-553) it is possible to separate the group of HREEs from the LREEs. However, there is still a need for more efficient, more environmentally friendly and more economically feasible separation systems for mixtures of REEs which are better suited to separate REEs, including the separation of individual REEs within one group (HREEs and/or LREEs). Batchu et al. did not make any suggestion to extend their extraction system for extraction of REEs from nitrate feed solutions to extraction of REEs from chloride feed solutions. Moreover, the coordination chemistry of REEs in chloride solutions is very different from that in nitrate solutions. SUMMARY

The separation of a mixture of rare earths by non-aqueous solvent extraction with two immiscible organic phases has been studied. The more polar organic phase was ethylene glycol with dissolved lithium chloride and the less polar organic phase was the extractant diluted in «-dodecane. Cyanex 923 was found to be the most performant extractant amongst the investigated acidic, basic and solvating extractants: Cyanex 272, Cyphos IL 101 , Aliquat 336, bis(2-ethylhexyl)amine, trioctylphosphine oxide (TOPO) and Cyanex 923. The replacement of the aqueous chloride feed solutions by non-aqueous ethylene glycol feed solutions had a profound effect on the distribution ratios and separation factors. The separation factors for extraction of pairs of rare earths from aqueous chloride solutions by Cyanex 923 are too low to be of practical use. On the contrary, a mixture of rare earths can be separated conveniently into four different groups by extraction with Cyanex 923 from ethylene glycol (+LiCl) solutions. The influence of several parameters such as the chloride concentration, the type of chloride salt, the addition of other polar solvents to the ethylene glycol phase, the addition of second extractant to the less polar organic phase, and the addition of complexing agents to the ethylene glycol phase has been studied. Furthermore, a conceptual flow sheet for the fractionation of rare earths from an ethylene glycol (+LiCl) feed solution into different groups by extraction with Cyanex 923 has been proposed. The new extraction system is very suitable, amongst others, for extraction of scandium and for separation of scandium from the other REEs.

Numbered statements of the invention are :

1. A process for separating a mixture of two or more rare-earth elements, comprising the solvent extraction of rare-earth elements from an organic phase A, comprising ethylene glycol and chloride ions, into an immiscible organic phase B.

2. The process according to statement 1 , wherein the organic phase B comprises an extractant diluted into a non-polar organic solvent.

3. The process according to statement 2, wherein the extractant is a neutral extractant. 4. The process according to statement 2 or 3, wherein the extractant is trialkylphosphine oxide.

5. The process according to any of statements 2 to 4, wherein the extractant is Cyanex 923.

6. The process according to any of statements 1 to 5, wherein the chloride ions in phase A are generated, at least partially, from the addition of lithium chloride.

7. The process according to any of statements 1 to 6, wherein the organic phase A is repeatedly extracted with organic phase B.

8. The process according to statement 7, wherein the concentration of chloride ions in organic phase A is varied in the individual extractions, for example by dissolution of salts containing chloride ions or by dilution.

9. The process according to statement 8, wherein the concentration of chloride ions is increased by addition of a salt selected from the group consisting of lithium chloride, calcium chloride, magnesium chloride, nickel chloride, and ammonium chloride or a mixture thereof.

10. The process according to statement 8, wherein the salt is lithium chloride.

1 1. The process according to any of statements 1 to 10, wherein the process is a non-aqueous process.

BRIEF DESCRIPTION OF THE FIGURES

Figure 1. Structures of Cyanex 923, Cyanex 272, bis(2-ethylhexyl)amine, Aliquat 336, Cyphos and IL 101. For Cyanex 923, the main component trioctylphosphine oxide (TOPO) is shown. Other components include R ₂ R'P=0, RR' ₂ P=0 and R' ₃ P=0 (R = octyl, R' = hexyl). For Aliquat 336, the main component methyltrioctylammonium chloride is shown.

Figure 2. Influence of the Cyanex 923 concentration on the solvent extraction of REEs. Experimental conditions: [REE(III)] : ~ 2.5 g/L (each), [LiCl] : 2 M, [Cyanex 923] : 0.1-1 M in «-dodecane with 10% (v/v) 1-decanol.

Figure 3. Influence of the TOPO concentration on the solvent extraction of REEs. Experimental conditions: [REE(III)] : ~ 2.5 g/L (each), [LiCl] : 2 M, [TOPO] : 0.1-0.5 M in «-dodecane with 10% (v/v) 1 -decanol. Figure 4. Influence of the Cyanex 272 concentration on the solvent extraction of REEs. Experimental conditions: : [REE(III)] : ~ 2.5 g/L (each), [LiCl] : 2 M, [Cyanex 272] : 0.1-1 M in «-dodecane with 10% (v/v) 1-decanol.

Figure 5. Influence of the Cyanex 923 concentration on the solvent extraction of a mixture of REEs including Sc(III) from ethylene glycol solutions (a) and from aqueous solutions (b). Experimental conditions: [REE(III)]: ~5 g/L(each), [LiCl]: 2 M, [Cyanex 923] : 0.1-1 M in n- dodecane with 10% (v/v) 1 -decanol.

Figure 6. Influence of the Cyanex 923 concentration on the solvent extraction of a mixture of REEs excluding Sc(III) from ethylene glycol solutions (a) and from aqueous solutions (b). Experimental conditions: [REE(III)]: ~5 g/L (each), [LiCl]: 2 M, [Cyanex 923] : 0.1-1 M in n- dodecane with 10% (v/v) 1 -decanol.

Figure 7. Influence of the chloride concentration (LiCl) on the solvent extraction of REEs. Experimental conditions: [REE(III)] : ~5 g/L (each), [LiCl] : 0.1-4 M, [Cyanex 923] : 1 M in n- dodecane with 10% (v/v) 1 -decanol. Figure 8. Influence of the source of chloride ions on the solvent extraction of REEs. Experimental conditions: [REE(III)] : ~5 g/L (each), [CI ] = 1 M, [Cyanex 923] : 1 M in «-dodecane with 10% 1-decanol (v/v).

Figure 9. Solvent extraction of REEs from ethylene glycol (+LiCl) solutions using a mixture of Cyanex 923 and Cyanex 272 (a), TBP (b), Aliquat 336 (c) in «-dodecane. Experimental conditions: [REE(III)] : ~5 g/L(each), [LiCl] :2 M, [Cyanex 923 + other extractant] : 1 M.

Figure 10. Influence of addition of polar solvent (50%(v/v)) to ethylene glycol on the solvent extraction of REEs. EG* = EG feed without added polar solvent. Experimental conditions: [REE(III)] : -2.5 g/L (each), [LiCl] : 2 M, [Cyanex 923] : 1 M in «-dodecane with 10% (v/v) 1- decanol. Figure 11. Conceptual flow sheet for the fractionation of REES from ethylene glycol(+LiCl) feed into four different groups by Cyanex 923 in «-dodecane. LO = loaded organic phase. DESCRIPTION

Materials and reagents

LaCi3-7H ₂ 0 (99.9%), CeCl ₃ -7H ₂ 0 (99.9%) and GdCl ₃ -6H ₂ 0 (99.9%) were purchased from Alfa Aesar (Ward Hill, USA), PrCl ₃ -7H ₂ 0 (99.9%), NdCl ₃ -6H ₂ 0 (99.9%), YC1 ₃ -6H ₂ 0 (99.9%), YbCl ₃ -6H ₂ 0 (99.9%) from Strem Chemicals (Newburyport, USA) and DyCl ₃ -6H ₂ 0 (99.9%),) from abcr GmbH (Karlsruhe, Germany). Sc ₂ 0 ₃ was kindly supplied by Solvay (La Rochelle, France). ScCl ₃ - H ₂ 0 was prepared from Sc ₂ 0 ₃ following a reported method. Ethylene glycol (99.9%), «-dodecane (>99%), 1-decanol (98%), N,N-dimethylfomamide (99%), methanol (>99.9%), lactic acid (90% in H ₂ 0), citric acid (99%) and methanesulfonic acid (98%) were obtained from Acros Organics (Geel, Belgium). Sodium chloride (>99%) was obtained from Fisher Scientific (Geel, Belgium), ammonium chloride (99%), anhydrous calcium chloride (95%) from Chem-Lab NV (Zedelgem, Belgium), anhydrous magnesium chloride (99%) from Alfa Aesar (Ward Hill, USA) and lithium chloride (99.9%) from Sigma-Aldrich (Diegem, Belgium). HC1 (37%) and dimethyl sulfoxide (100%) were supplied by VWR Chemicals (Haasrode, Belgium). Trihexyl(tetradecyl)phosphonium chloride (Cyphos ^® IL 101, >97%), Cyanex ^® 923 and bis(2,4,4- dimethylpentyl) phosphinic acid (Cyanex ^® 272) were provided by Cytec Industries (Canada). Tricaprylmethylammonium chloride (Aliquat ^® 336, 88.2-90.6%) was purchased from Sigma- Aldrich (Diegem, Belgium). Trioctylphosphineoxide (TOPO, 99%) and bis(2-ethylhexyl) amine (99%) were obtained from Acros Organics (Geel, Belgium). The silicone solution in isopropanol for the treatment of the TXRF quartz glass carriers was obtained from SERVA Electrophoresis GmbH (Heidelberg, Germany). The gallium standard (1000 mg/L in 2-5% H 0 ₃ ) was purchased from Chem-Lab NV (Zedelgem, Belgium). All chemicals were used as received without any further purification.

Instrumentation

A fiat magnetic stirrer (MIX 15 eco model, 2mag magnetic ^e motion) was used to mix the two phases in the solvent extraction experiments. The metal concentration in the ethylene glycol phase was determined by a total reflection X-ray fluorescence (TXRF) spectrometer (Bruker S2 Pico fox). ^! H NMR spectra were recorded on a Bruker Avance 300 spectrometer (operating at 300 MHz) and t ³¹ P NMR spectra on a Bruker Ascend 400 spectrometer (operating at 162 MHz). The samples for NMR spectra were prepared in acetone -de. The chemical shifts are noted in parts per million (ppm), referenced to tetramethylsilane for and to 85% H3PO4 (external reference) for ³¹ P NMR. The water content was determined by coulometric Karl Fischer titration (Mettler-Toledo DL39 titrator). The viscosity of ethylene glycol was measured using an Anton Paar rolling-ball viscometer (Lovis 2000 ME). The concentration of ethylene glycol in «-dodecane samples was determined using a gas chromato graph, combined with a flame ionization detector (GC-FID) and a Turbomatrix 16 headspace autosampler (HS) (Perkin Elmer, USA). A Perkin Elmer CP-SIL 8 CB (5%-phenyl-95%-dimethylpolysiloxane) capillary column, 50 m long, 0.32 mm I.D. and 1.20 μηι film thickness was used as GC column. The PE TotalChrom Version 6.3.2.0646 software package was used to determine the peak areas in the gas chromatogram. Solvent extraction procedure

Non-aqueous solvent extraction experiments were carried out by mixing equal volumes of the more polar organic phase (REEs in ethylene glycol +LiCl) and the less polar organic phase (Cyanex 923 in «-dodecane) in glass vials at 1000 rpm and room temperature (RT, 293 ± 2 K) for 15 min. An aqueous solution and Cyanex 923 in «-dodecane were the two phases for the conventional solvent extraction experiments. The behavior of the diluents n-dodecane and kerosene is very similar. Kerosene is a mixture of hydrocarbons (C10 to Ci6 compounds) and it is widely used in industry as a diluent for solvent extraction processes. The pure compound n- dodecane is often used for fundamental solvent extraction studies, because it is more convenient for characterization studies with for instance NMR or GC-FID. 1-Decanol was added to Cyanex 923/ «-dodecane as phase modifier both non-aqueous and conventional solvent extraction experiments. After equilibrium, the concentrations of the REEs in ethylene glycol were determined by TXRF after proper dilutions and addition of a gallium internal standard. The concentrations of the REEs in the «-dodecane phase were calculated via the mass balance. The quartz glass sample carriers for TXRF measurements were pretreated with 20 iL of a silicone solution in isopropanol, dried in oven for 2 min at 333 K followed by addition of 2 iL sample to carrier and dried at 333 K for 30 min. The samples were measured for 300 s in the TXRF spectrometer. All samples were measured in duplicate.

The distribution ratio D is defined as the ratio of total metal concentration in the less polar organic phase («-dodecane, CDD) to the total concentration in more polar organic phase (ethylene glycol, CEG) at equilibrium: (1)

For the separation of two metals MA and MB, a separation factor β can be defined: where D _A > D B,

The percentage extraction %E is calculated as follows: where VEG is the volume of the ethylene glycol phase and VDD is the volume of the n- dodecane/Cyanex 923 phase.

The composition of the feed was about 5 g/L of each REE ion (Sc(III), La(III), Ce(III), Pr(III), Nd(III), Gd(III), Dy(III), Y(III), Yb(III)) in ethylene glycol(+LiCl) for most of the experiments, unless otherwise specified and about 2.5 g/L for experiments of screening of extractants, addition of polar solvents (50% (v/v)) and complexing agents. The accurate concentrations in the solution has been measured by TXRF.

Results and discussion

Mutual solubility of organic phases

Since two organic phases are being used for the solvent extraction of REEs, it is important to determine their mutual solubility. Mutual solubility experiments were conducted by mixing equal volumes of ethylene glycol and Cyanex 923/«-dodecane as a function of the LiCl concentration, feed concentration and extractant concentration. The solubility of Cyanex 923 in ethylene glycol phase was determined by ³¹ P NMR and the solubility/co-extraction of ethylene glycol in Cyanex 923/«-dodecane was measured by gas chromatography. The solubility of Cyanex 923/«-dodecane in the ethylene glycol phase is very low, as shown by the absence of resonance lines corresponding to P=0 in the ³¹ P NMR spectra. The solubility/co-extraction of ethylene glycol by 1 M Cyanex 923 (+10% v/v l -decanol)/«-dodecane was measured to be 46.3 g/L from the gas chromatography measurements. The addition of 2 M LiCl to the ethylene glycol phase suppressed the solubility/co- extraction of ethylene glycol to 29.7 g/L. The solubility of ethylene glycol in 1 M Cyanex 923 (+10% v/v 1 -decanol)/ «-dodecane was further suppressed to 23.5 g/L when extractions were carried out from a feed solution of 50 g/L of REEs together with 2 M LiCl. Additionally, an experiment was carried out to compare the extraction of water molecules by Cyanex 923 from aqueous solutions with the extraction of ethylene glycol by Cyanex 923 from ethylene glycol solutions. The co-extraction of water molecules by 1 M Cyanex 923 (+10% v/v 1 -decanol)/ n- dodecane as determined by Karl Fischer titration was 24.6 g/L after equilibrating with distilled water and was suppressed to 22.7 g/L from aqueous solutions containing 2 M LiCl. The co- extraction of water was even further reduced to 21.7 g/L from the aqueous feed solution containing ~50 g/L of REEs. The co-extraction of water and ethylene glycol are comparable.

The viscosity of the ethylene glycol was measured as a function of LiCl concentration at 298 K. The viscosity of the ethylene glycol phase gradually increased from 18 mPa-s for 0 M LiCl to 21 1 mPa-s for 4 M LiCl. Similarly, the viscosity of the ethylene glycol feed (containing ~45 g/L of REE) increased from 30 mPa-s for 0 M LiCl to 308 mPa-s for 4 M LiCl. An increase in temperature resulted in suppression of viscosity from 124 mPa-s at 293 K to 21 mPa-s at 333 K for ethylene glycol feed containing ~45 g/L of REE and 2 M LiCl.

Screening of extractants

In order to find a suitable extractant for the separation of REEs from non-aqueous ethylene glycol(+ 2M LICl) solutions containing 9 REE ions (Sc(III), La(III), Ce(III), Pr(III), Nd(III), Gd(III), Dy(III), Y(III) and Yb(III)) each of a concentration of ~2.5 g/L, a screening experiment was carried out using acidic, neutral and basic extractants. The extractants tested were Cyanex 272, Cyanex 923, trioctylphosphine oxide (TOPO), bis(2-ethylhexyl)amine, Cyphos IL 101 and Aliquat 336 (Figure 1). All extractants were dissolved in «-dodecane diluent. Preliminary experiments showed that the phase separation was not clear so that 10% (v/v) 1 -decanol was added to «-dodecane as phase modifier. In order to determine the extraction behavior of 10% (v/v) 1- decanol/«-dodecane diluent, one extraction experiment was carried out with 10% (v/v) 1- decanol/«-dodecane diluent without Cyanex 923 added to the diluent phase. It was found that the extraction of all REEs was close to zero. This shows that 10% (v/v) 1 -decanol acts only as phase modifier and plays no role as extractant. The concentration of each extractant was varied from 0.1 M to 1 M, except for TOPO, for which the maximum concentration that could be obtained was 0.5 M.

Sc(III) has a strong affinity for Cyanex 923 and TOPO, resulting in high extraction efficiencies (Figures 2 and 3). The extraction percentages were higher for Cyanex 923 than for TOPO. The order of extraction of REEs using these phosphine oxide reagents is: Sc(III) > Yb(III) > Υ(ΠΙ) > Dy(III) > Gd(III) > Nd(III) > Pr(III) > Ce(III) > La(III). The extraction behavior of Y(III) was similar to that of the HREEs, in between Yb(III) and Dy(III). It was observed that the extractants Cyphos IL 101 , Aliquat 336 and bis(2-ethylhexyl)amine were unable to extract REEs from ethylene glycol(+LiCl) solutions under the experimental conditions tested. Sc(III) was selectively extracted and negligible extraction (%E = 0-5%) of all other REEs was observed for Cyanex 272 (Figure 4).

Comparison of extraction from non-aqueous and aqueous solutions

Solvent extraction of a mixture of REEs including Sc(III) (each ~5 g L) has been carried out from ethylene glycol (+ LiCl) and aqueous chloride solutions using Cyanex 923 dissolved in «-dodecane (Figure 5). Sc(III) was selectively extracted from aqueous chloride solutions at all concentrations, because Cyanex 923 has a strong affinity for Sc(III). The percentage extraction of all other REEs was close to zero, even for 1 M Cyanex 923. Therefore, a mixture of REEs cannot be separated under these conditions. In the case of ethylene glycol solutions, Sc(III) was selectively extracted at Cyanex 923 concentrations up to 0.5 M. It must be noted that Sc(III) is selectively extracted from both aqueous and ethylene glycol solutions. However, the extraction of Sc(III) is more efficient from ethylene glycol solutions than from aqueous chloride solutions. Under similar experimental conditions (0.3 M Cyanex 923), the %E of Sc(III) is 67.7% from ethylene glycol solutions, where as it is 55.6% from aqueous chloride solutions. Additionally, only Sc(III) was extracted from aqueous solutions. On the other hand, HREEs such as Yb(III),Y(III) and Dy(III) were also extracted together with Sc(III) from ethylene glycol at Cyanex 923 concentrations higher than 0.5 M. For instance, the %E of Yb(III) was 91.2% for ethylene glycol solutions and 0.0% from aqueous solutions for extraction with 1 M Cyanex 923. This clearly shows that the extraction of HREEs is much more efficient from ethylene glycol solutions than from aqueous solutions. Negligible extraction percentages were found for the LREEs (La, Ce, Pr, Nd) from ethylene glycol solutions as well. Solvent extraction of rare earths from aqueous chloride solutions by neutral or solvating extractants like Cyanex 923 is inefficient since Cyanex 923 usually transfers metal ions coordinated to their anions (salt extraction). It is well known that rare-earth ions are strongly hydrated in aqueous solutions (with eight or nine coordinated water molecules) and no chloride ion is present in the inner coordination sphere, even at very high chloride concentrations Due to the lack of coordinating chloride anions, the extraction of rare earths by Cyanex 923 from aqueous chloride solutions is inefficient. For extraction by Cyanex 923, the solvating molecules around the rare- earth ions must be replaced fully or partly by the Cyanex 923 molecules to form adducts. In the case of ethylene glycol (+LiCl) solutions, the rare-earth ions are solvated by ethylene glycol molecules, even though some water molecules can still be present if no anhydrous rare-earth salts are being used. The solvation energy required to remove the solvation sphere around the metal ion in ethylene glycol solutions is lower than the energy required for the removal of the hydration sphere around the metal ion in aqueous solutions. Therefore, the extraction of rare earth ions by Cyanex 923 from ethylene glycol (+LiCl) is more efficient than the extraction from aqueous chloride solutions. It is also important to note that the difference in extraction behaviour of rare earths from ethylene glycol and aqueous solutions depends on the type of rare earth ion (LREE or HREE) and coordinating anion.

Because Cyanex 923 has a strong affinity for Sc(III), it is relevant to investigate how the extraction behavior changes after the removal of Sc(III) from aqueous and non-aqueous solutions. Experiments were carried out with a Sc(III)-free feed solution in order to investigate the extraction behavior of REEs after removal of Sc(III) (Figure 6). It can be seen that the extraction of the REEs is almost the same for aqueous chloride solutions with and without Sc(III) present in the feed solution. There is no significant extraction of the REEs, with the exception of Yb(III). The extraction of the HREEs was much more efficient than the extraction of LREEs from ethylene glycol solutions. This makes it possible to separate HREEs efficiently from LREEs. A high separation factor of 191 was obtained for the HREE group (Yb+Y+Dy, £>Yb+Y+D _y = 7.63) over the LREE group (La+Ce+Pr+Nd, £>La+ce+Pr+Nd = 0.04) at 1 M Cyanex 923. Figure 6a also shows that there is a possibility to separate Yb(III) from rest of the REEs by extraction with 0.3 M or 0.5 M Cyanex 923. This can be evaluated in terms of separation factors. The separation factors for Yb over Dy+Y are 9.5 at 0.3 M Cyanex 923 and 11.2 at 0.5 M Cyanex 923. These separation factors are sufficiently high to design separation processes for HREEs based on extraction of REE chlorides from ethylene glycol by Cyanex 923. On the other hand, the REEs cannot be separated by extraction with Cyanex 923 from aqueous chloride solutions. Therefore, it can be concluded that the replacement of water by ethylene glycol not only enhances the distribution ratios but also the separation factors. Effect of chloride concentration

The influence of the chloride concentration on the extraction of REEs (~5 g/L(each)) was studied by increasing the LiCl concentration from 0.1 M to 4 M, with 1 M Cyanex 923 in «-dodecane as the extracting solvent (Figure 7). It can be seen that the extraction of the REEs increases with an increase in LiCl concentration, which is a typical behavior for solvating extractants. It was observed that Sc(III) is extracted quantitatively, even at very low chloride concentrations. The separation factors for Yb over Dy+Y are 8.5 for 1 M LiCl and 9.2 for 2 M LiCl. For the least extractable elements (LREEs), the extraction efficiency was increased to 20% at high chloride concentrations. High chloride concentrations are not a favourable condition for the separation of REEs. It is clear that by carefully controlling the concentrations of LiCl and Cyanex 923, a mixture of REEs including Sc(III) can be separated conveniently into three or four fractions. One option is to separate Sc(III) from the other REE ions at low chloride concentrations, followed by separation of Yb(III) and then separation of Y(III) and Dy(III). All LREEs remain in the ethylene glycol phase. In the second option, after the removal of Sc(III), the HREEs (Yb, Y, Dy) can be separated as a group from other REEs. Therefore the selection of a suitable process depends on the actual composition of the feed solution.

Effect of source of chloride

The influence of source of the chloride on the separation of REEs (~5 g/L(each)) byusing different chloride sources such as LiCl, NaCl, NH ₄ C1 (1 M) and MgCb, CaCb (0.5 M) was investigated for 1 M Cyanex 923. The solubility of LiCl in ethylene glycol is more than that of any other salt studied. It is possible to dissolve more than 4 M of LiCl in ethylene glycol whereas the maximum solubilities are 1 M for NaCl and NH ₄ C1, and 0.5 M for MgCb and CaCb. Based on its good solubility in ethylene glycol, LiCl was selected as a source of chloride. Unlike in aqueous solutions, there is no significant salting-out effect on the extraction of REEs (except for Yb(III)) from ethylene glycol feed solutions. A change of cation of the chloride salt has a negligible effect on the extraction and for on the separation of REEs (Figure 8). Addition of a second extractant (synergist) to the less polar organic phase

Synergstic systems are often used in solvent extraction to study any improvement in the extraction or separation by the addition of a second extractant. Synergistic extraction is generally defined as an enhancement in extraction/separation by the cooperative effect of two extractant molecules in comparison to single extractants. The separation factors may increase by certain combination of extractants. Therefore, experiments were carried out by adding a second extractant to Cyanex 923: Cyanex 272, TBP or Aliquat 336 (Figure 9). The total concentration of extractant (Cyanex 923 + other extractant) was kept constant at 1 M.

The addition of Cyanex 272 to Cyanex 923 had a positive effect on the extraction of Yb(III), which means the Cyanex 272 acts as a synergist. On the other hand, addition of TBP or Aliquat 336 to Cyanex 923 had a negative effect on the extraction of Yb(III), which indicates that these extractants act as an antagonist.

Addition of polar solvents to ethylene glycol

The influence of addition of the polar solvents water, dimethyl sulfoxide (DMSO), N,N- dimethylformamide (DMF) and methanol on the extraction of REEs (~2.5 g/L (each)) was studied by the addition of 50% (v/v) polar solvent to an ethylene glycol feed solution containing 2 M LiCl (Figure 10). In mixtures of solvents, ions can be solvated preferentially by one of the solvents in the mixture, The preferential solvation of rare-earth ions can lead to better selectivities for extraction. It was found that the extraction of Sc(III) remained unchanged at 100 % extraction in all polar solvents. This shows that Cyanex 923 is a promising candidate to extract Sc(III) from different non-aqueous solvents. Note that the percentage extraction of Yb(III) is drastically decreased from 99.5% to 46.5% by the addition of 50% (v/v) water to ethylene glycol. In other words, the percentage extraction of Yb(III) increased from 0.0% in water (Figure 5b) to 46.5% in 50% (v/v) ethylene glycol. There was no significant change in the extraction behavior of Yb(III) in other solvents (DMSO, DMF and methanol). In the case of DMSO, interesting results were obtained for the extraction of Y(III) and Dy(III). The %E was strongly supressed from 90 % to 25% for Y(III) and from 85% to 14% for Dy(III) when going from ethylene glycol to ethylene glycol/DMSO mixtures. The co-extraction of Y(III) and Dy(III) was suppressed by the addition of DMSO to the ethylene glycol feed solution. The addition of DMSO might be a viable option for the separation of HREEs such as Yb from Dy+Y. The separation factor of Yb over Dy+Y increased from 26 (in ethylene glycol) to 36 (in DMSO). There is no significant change in extraction behavior upon addition of 50%(v/v) DMF or methanol to the ethylene glycol (+LiCl) phase.

Influence of complexing agent

It was reported that addition of complexing agents like ethylenediaminetetraacetic acid (EDTA), diethylenetriaminepentaacetic (DTP A), lactic acid and citric acid to aqueous chloride solutions of rare earths has a positive influence on the separation factors in the solvent extraction with acidic extractants. Therefore, we investigated the complexation behaviour of rare-earth ions with these complexing agents in ethylene glycol. Unfortunately, EDTA and DTPA did not in ethylene glycol. The influence of other complexing agents on the extraction and separation of REEs was studied by addition of lactic acid and citric acid (0.5 M) to ethylene glycol solutions containing REE chlorides (Table 1). Sc(III) was still quantitatively extracted, but the percentage extraction of all other REEs, except Yb(III), was close to zero. The addition of the complexing agents lactic acid and citric acid to ethylene glycol feed had thus a negative influence on the extraction and separation of REEs. The extraction of some heavy rare earths is suppressed by the addition of EDTA to aqueous chloride solutions.

Table 1. The extraction percentage of REEs from ethylene glycol with complexing agents ¹

%E Sc Y La Ce Pr Nd Gd Dy Yb

Citric acid 99.7 1.8 0.0 0.0 0.0 0.0 0.0 0.0 10.9

Lactic acid 99.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 5.6

No complexing 100 90.2 7.2 7.6 11.5 11.8 34.2 85.3 99.5 agent

Experimental conditions: [REE(III)] : ~2.5 g/L (each), [Cyanex 923] : 1 M in «-dodecane with 10% (v/v) 1 -decanol.

Stripping and recycling studies

Stripping studies were carried out for Yb(III)-loaded 1 M Cyanex 923 (5.3 g L of Yb(III)) aqueous HCl and methanesulfonic acid solutions. Stripping percentages were 99.9 and 99.3% for 1 M methanesulfonic acid and 1 M HCl, respectively. Considering the high stripping percentage for methanesulfonic acid (CH3SO3H) and the fact that methanesulfonic acid is considered as a sustainable acid and solvent, methanesulfonic acid was selected for recycling studies. First, Yb(III) was extracted from ethylene glycol phase by 1 M Cyanex 923. The Yb(III)-loaded Cyanex 923/«- dodecane phase was stripped with 1 M methanesulfonic acid. In the second cycle, the regenerated Cyanex 923 was used for the extraction of Yb(III) from ethylene glycol, followed by stripping with 1 M methanesulfonic acid. The system was studied for five cycles and it was found that the percentage extraction (98.5 ± 1%) and percentage stripping percentage (99.8 ± 1%) were unaffected.

Conceptual flow sheet

A conceptual flow sheet is proposed for the group separation of REEs (Figure 1 1). First, Sc(III) is selectively separated by 0.5 M Cyanex 923 from ethylene glycol feed without addition of LiCl, followed by extraction of Yb(III) using 0.5 M Cyanex 923 from ethylene glycol feed with 2 M LiCl and Y(III)+Dy(III) using 1 M Cyanex 923 from ethylene glycol feed with 2 M LiCl. Therefore it is possible to fractionate the REEs in to four different groups using one single extractant: Sc(III), Yb(III), Y(III)+Dy(III) and LREEs. The ethylene glycol raffinate (after complete removal of rare earths) can be recycled back to the feed. Conclusions

This invention shows that the extraction of REE chlorides from ethylene glycol solutions is significant different from extraction from aqueous feed solutions. The replacement of water by the organic solvent ethylene glycol allows for the efficient separation of REEs from a chloride feed solution. Cyanex 923 was found to be the best extractant for the separation of REEs from ethylene glycol (+LiCl) solutions. Cyanex 923 has strong affinity for Sc(III), Yb(III) and Y(III). The addition of LiCl has a positive effect on the percentage extraction. The type of chloride salt has a negligible effect. Cyanex 272 acts as synergist and TBP, Aliquat 336 as antogonists in their mixtures with Cyanex 923 for the extraction of REEs. The addition of the polar solvent DMSO to ethylene glycol (+LiCl) solutions increased the separation factor of Yb over Y+Dy. The addition of complexing agents to ethylene glycol phase decreased the extraction of REEs. Quantitative stripping of Yb(III) could be carried out by HC1 or CH3SO3H. A conceptual flow sheet was proposed for the separation of REEs into four different groups: Sc(III), Yb(III), Y(III)+Dy(III) and LREEs.