Development and Screening of Resins to Recover Rare Earth Elements (REE) and Scandium from Different Sources

Originally presented at Extraction 2018, August 26-29, 2018. Please visit Springer for more information about the conference or how to purchase the paper or proceedings.
Mikhail Mikhaylenko 
Purolite Limited, Moscow, Russia
Natural and technogenic sources of rare earth elements (REE) and other rare and scattered metals are remarkable for their chemical diversity. The pregnant solutions going to processing for the valuable metals have very complex compositions. Both targeted elements and competing impurities are present in the solutions in different chemical forms. Thus processing of such materials demands application of different technological tools addressing numerous specific tasks on the way to the final product. Screening work was undertaken to identify the best resins among developmental and commercial products addressing typical cases in the hydrometallurgy of scandium and REE. The cases outline typical sources of scandium and REE, such as streams of TiO2 production, extracted phosphoric acid, barren uranium solutions and others. Specially designed ion exchange resins can be effective tools.
Keywords: REE, Scandium, Thorium, Sorption, Ion Exchange, Separation, Recovery, Sorption, Desorption, Selectivity
Modern technologies consume many advanced materials based on elements, among others, such as rare earth elements (REE). Yttrium and scandium, thanks to their chemical similarity to REE, are often considered in science and technology as belonging to the REE group. Though the REE market size is relatively small, these elements are critically important for some high-tech applications (1).
Complex compounds of oxide, phosphate, and carbonate or fluoride nature represent most of the REE minerals. Common constituents in such minerals are titanium, iron, thorium, uranium and other metals (2).
Thorium is a common undesired impurity in produced REE concentrates because of its radioactivity. The same is true of uranium (3). The issue of thorium and REE separation becomes especially important in the processing of monazite minerals, which are rich in both thorium and REE (4), (5).
The chemistry of polyvalent cations in aqueous solutions is very complex (6). The trivalent lanthanides are hard acids by the Pearson definition and as such, they complex with the hard bases like F-, CO32-, SO42-, PO43- and OH- (7). All of the latter, except fluoride, can be characterized as “oxygen” ligands. Presumably some common impurities, i.e. thorium, ferric iron, titanium, zirconium behave similarly. It makes the task of their separate extraction complicated.
A variety of technical tasks related to extraction of REE and auxiliary operations are considered in many reviews (8).
The last four resins in the below table are developmental products that are not available yet on a commercial scale.

Ion Exchange Resins Used in the Work
Resin Name Abbreviated Name in the Text Polymer Matrix Known or Supposed Functionality
Puromet MTS9570 S957 St/DVB, mp sulfonic, phosphonic acid
Puromet MTS9500 S950 St/DVB, mp aminophosphonic
AMP AMP St/DVB, mp aminophosphonic (a commercial product)
Puromet MTS9300 S930 St/DVB, mp iminodiacetic acid
Puromet MTA6002PF A600 St/DVB, gel quarternary ammonium
Purolite C160 C160 St/DVB, mp sulfonic acid
Purolite S956 S956 St/DVB, mp phosphonic acid
Purolite C115 C115 Methacrylic/DVB, mp carboxlyic acid
S958 S958 St/DVB, mp phosphonate
GS176 GS176 St/DVB, mp aminophosphonate
GS206 GS206 Acrylic/DVB, mp gem-diphosphonic, amine
GS218 GS218 St/DVB, mp gem-diphosphonic, amine
Distribution parameters Di

Since the goal of the work was quick screening among different resins in different solutions, with limited resources, the sorption in static conditions was used in most cases. Resins contacted the solutions in chosen proportions for fixed times. The experiments were run at ambient temperatures in all cases. Then metals were analyzed by spectrometric methods in the equilibrium solution and solution generated by wet incineration of the resin and leaching of the ash. Based on the analytical results, the distribution parameters Di were calculated:

Artificial solutions simulating real ones were used in the work. Not all lanthanides were included in the experiments. Typically, 1 to 3 elements were taken to represent light, medium and heavy rare earths in different experiments.

Scandium from Streams of TiO2 Production by Sulfate Route
One of the earliest works on ion exchange separation of scandium used strong acid cation exchange resin (9). Resins with different functionality have been examined for scandium separation from REE (10). Later, it was shown that phosphorus-containing resins are more effective for sorption of scandium (11), (12), (13).

Sulfate production of TiO2 pigment is comprised of several stages. When the acid leach solution is diluted, the titanic acid precipitates as a result of hydrolysis. At this step, the hydrolysis sulfuric acid is obtained. This stream contains 200‑250 g/l sulfuric acid, 15-17 mg/l scandium, around 3-5 g/l titanium, 30-40 g/l of total iron, as well as thorium, zirconium, aluminum and other impurities. The hydrolysis acid may be diluted further before disposal. Definitely, such streams are of interest as potential feeds for scandium recovery. Two solutions modeling the real hydrolysis sulfuric acid were prepared (see below).

Compositions of Solutions Modeling the Hydrolysis Sulfuring Acid
Concentration, mg/l Concentration, g/l
Solution Sc Th Cr Si Zr V Ti Fe
Fe3+ Al H2S04
#1 19.7 1.1 68.5 9.5 43.9 376 3.4 39.1 10.8 0.587 208
#2 12.9 0.49 n/a n/a n/a n/a 1.36 26.6 7.1 n/a 192

Sorption was investigated in column with flowrate of 0.5 BV/hour. Due to technical limitations, the sorption runs were stopped overnight and resumed next morning. In total 20 BV of solution was passed through the column.

In the solution #1 the resins GS176, S957, S958 and S956 and in the solution #2 the GS206 and GS218 were tested.

When 20 bed volumes of the feed had passed through the resin beds, the resins were collected and analyzed for metals. Data of the metal loadings on the tested resins are presented in the table below.

Metals Loadings and Upgrades in Resins Against the Feed
Metal Content in Resin, mg/l Upgrade Parameter
Resin Sc Ti Th Fe Zr Cr Sc Ti Th Fe Zr Cr
GS176 130 15,296 22.2 4,074 259 7.4 6.6 4.5 20.2 0.10 5.9 0.11
S957 186 41,143 23.8 9,095 290 14.3 9.4 12.1 21.6 0.23 6.6 0.21
S958 68 5,960 n/a 2,160 272 12 3.5 1.8 n/a 0.06 6.2 0.18
S956 180 22,532 25.8 15,536 356 30 9.1 6.6 23.5 0.40 8.1 0.44
GS206 125 13,875 16.7 4,333 246 20.8 9.7 10.2 34.1 0.61 n/a n/a
GS218 171 10,735 n/a 706 471 8.8 13.3 7.9 n/a 0.10 n/a n/a
Carbonate Desorption Profiles for Sc

The upgrade parameter in the table above was calculated as loading of metal x divided by concentration of metal x in the initial feed solution. The best Sc upgrade was obtained for the GS218 resin. It is remarkable that all tested resins demonstrate upgrade parameters for such impurities as Fe and Cr as low as <1 indicating excellent selectivity over these elements. Thus, right choice of the resin in this application will be a tradeoff between higher scandium loading on one side and purity of the loaded scandium on another side. Definitely, the complexity of the downstream separation of scandium and thorium will influence this choice.

The loaded resins were rinsed with demineralized water and desorption of scandium was done by passing a solution of 200 g/l (NH4)2CO3 and 50 g/l ammonia at a flow rate of 1 BV/hour. In total, 6 BVs were passed. Desorption profiles are depicted in the figure below.
Carbonate Desorption Profiles for Sc

Desorption of scandium is most effective from GS218 resin. In addition, due to higher scandium loading at sorption, the average concentration of scandium in the desorbate is also higher than for other resins. Desorption from S958 is also effective but the grade of the average desorbate is lower. It is interesting to notice that the GS176 desorption profile, which is also effective, is shifted right by about 1 BV. It is as if this resin requires more carbonate for scandium desorption and reflecting a chemical difference of this resin.

Scandium is successfully desorbed from resins GS176, S958 and GS218. Thorium is well desorbed from all resins. Titanium poses the biggest problem at desorption. This metal can be stripped from the resins by fluoride reagents.  However, this is not a desirable method because of the high corrosivity of fluoride ion in acidic media.

Scandium in Uranium In-Situ Leaching Liquors
Scandium and REE are often found in uranium minerals and leached by sulfuric acid together with uranium. A significant amount of natural uranium is recovered by in situ leaching (ISL) technology. Uranium is extracted from the pregnant solution as uranyl sulfate anionic complexes while all elements in cationic form pass in to the sorption filtrate solution. Scandium, thorium and REE remain as cations under these conditions and these metals can be recovered downstream. Table 4 represents a typical analysis of an ISL filtrate solution after uranium recovery.

Composition of the Solution Modeling the Stream After Uranium Recovery (Constituents of the Model Solution, mg/l)
Sc La Y Yb Th Fe2+ Fe3+ Ca Mg Al PO43- H2SO4
1.2 38.3 21.2 6 3.1 450 450 700 200 1,200 100 pH 1.6

Here we can consider La as presenting LREE, Y and Yb for HREE.

Sorption experiments were carried out in batch mode at resin to solution volumetric ratio of 1:400 for 24 hours with continuous agitation. Then metals were analyzed in resin and solution. Two development resins, S958 and GS176, were compared with a commercial aminophosphonic resin AMP. Distribution parameters were calculated not only for individual elements but also for groups of them representing the sum of REE, impurities Me as sum of Th and Fe, and, finally, sum of REE plus Th and Fe considered as impurities with respect to Sc.  Accordingly, distribution coefficients Di were calculated (see table below).

Distribution Parameters Di for Metals Sorbed from the Stream After Uranium Recovery (Distribution parameter Di)
Resin Sc Y Yb La Th Fe REE Me REE + Me
AMP 47 16.8 18.3 16.0 25 26 16.5 26 26
Puromet MTS9580 211 5.0 8.7 0.57 39 42 2.7 42 39
GS176 357 0.13 0.40 0.004 263 17 0.08 18 16

It can be seen that the scandium capacity of GS176 resin is expected to be the highest of all three resins, while at the same time, loading of REE is negligible.

For estimation of selectivity the separation parameters for scandium against other elements and their groups were also calculated (see table below).

Separation Parameter at Sorption from the Stream After Uranium Recovery (Separation Parameter Scandium Versus Specific Metal(s))
Resin Y Yb La Th Fe REE Me REE + Me
AMP 2.78 2.55 2.91 1.88 1.78 2.83 1.78 1.83
Puromet MTS9580 42.5 24.1 370.6 5.41 5.02 77.34 5.02 5.40
GS176 2,706 901 88,630 1.36 20.7 4,370 20.3 21.8

All three resins are selective for Sc against other metals in the solution. However, the most important finding is the very high selectivity of GS176 resin for scandium against REE, especially against LREE. We can write the order of selectivity for scandium against REE this way: GS176≫S958≫AMP. The selectivity row against all metals except scandium will be the following: GS176≫S958>AMP.  The high scandium selectivity of the GS176 resin will reduce the costs of downstream processing of the scandium desorbate. In addition, the GS176 resin can be recommended for selective separation of scandium from REE concentrate solutions in weak acid sulfate matrices.

REE in Wet Process Phosphoric Acid Production
REE are often present in significant amounts in the minerals used for production of phosphate fertilizers. Phosphoric acid is extracted from phosphate minerals by strong mineral acids, most often by sulfuric acid. The material flow splits in two, an acidic solution and a solid waste consisting mainly of so-called phosphogypsum, e.g. calcium sulfate contaminated with other materials. REE are distributed between these two phases. These streams used to attract great attention from researchers looking to recover REE (14), (15), (16), (17), (18). Most of the studies have dealt with processing of the phosphogypsum. In our work, we selected the liquid phase as the object of our investigation.

The table below presents the modeling acid solution. The array of elements represents LREE, MREE and HREE.

Modeling Composition of Phosphoric Acid Leached by Sulfuric Acid (Concentration in Modeling Solution)
g/L mg/L
H3PO4 H2SO4 Fe Na Ca Mg Al Y La Nd Sm Dy Yb
450 10 2.2 85 315 220 60 41.4 378 195 58 11 45

Recovery of REE from phosphoric acid is a very challenging task.  Sorption was carried out in batch at a resin to solution volumetric ratio of 2:200 for 24 hours with continuous stirring. Then metals were analyzed in both phases. Calculated distribution parameters are presented in the table below.

Distribution for Parameters Di for Sorption of REE from Wet Phosphoric Acid, Sulfate Matrix (Distribution Parameters Di)
Resin Y La Nd Sm Dy Yb Al Na Fe Ca Mg REE Me
PPC160 3.55 25.5 11.0 7.28 9.35 1.51 1.03 6.0 0.17 22.9 3.55 16.3 2.6
S957 2.43 4.14 3.85 3.51 5.05 2.31 2.15 2.48 1.23 3.18 1.31 3.8 1.5
S958 0.77 0.25 0.25 0.25 0.79 4.15 1.89 2.8 0.59 0.88 0.36 0.5 0.7
AMP 1.37 1.07 1.25 1.2 2.62 4.17 2.57 0.85 9.21 0.78 0.32 1.4 6.9
Separation Parameters S for (a) Total REE (1)

The figure below shows separation parameters for REE and Yb against all trivial, i.e. having no commercial value in this application, impurities (Al+Na+Fe+Ca+Mg) in the table above).

It can be seen that the strong acid cation exchange resin PPC160 demonstrates the highest loading of LREE and MREE among the resins tested, as well as the best selectivity against iron. At the same time this resin uptakes significant amount of calcium to compare with other resins. S958 and AMP resins prefer HREE, but the AMP resin also loads too much iron. PPC160 has the best selectivity for REE due its strong affinity for LREE and MREE. However if the focus is on recovery of HREE then S958 has the best characteristics in terms of selectivity.

Scandium, with REE and Thorium
Monazite is an important source for both thorium and REE, including scandium. Monazite may be processed by various routes, producing different solid materials and aqueous streams. One such stream is obtained after leaching of the solid residues with nitric acid. Table 9 represents the model composition of such a solution.

Modeling Nitric Acid Solution with Sc, Th and REE (Concentration, g/L)
Sc Th Y La Ce Nd Tb Yb Fe Ca Ti HNO3
0.074 0.346 1.50 10.9 17.5 5.0 0.282 0.110 6.7 6.4 0.235 180

Sorption was carried out in batch at a resin to solution volumetric ratio of 1:100 for 24 hours with continuous stirring. Metals were analyzed in both equilibrated phases. Table 10 shows distribution parameters for the metals of interest and impurities.

Distribution Parameters Di for Sorption of Metals from Nitric Acid Matrix
Resin Sc Th Fe Ca Ti REE + Y Me Sc + Th REE + Y + Me
S958 59.77 15.37 0.91 0.3 9.3 0.1 0.8 21.8 0.3
S956 29.6 17 1.04 3.86 0.3 0.0 2.4 19.1 0.7
GS176 34.13 44.57 0.73 0.11 4.89 0.1 0.5 42.6 0.2
Separation Parameter STh (Left) and SSc (Right)

We can see that all three resins have good loadings of Sc and Th but low loading of REE and Y. GS176 has the best loading of Sc and Th sum. Then it will be possible to separate these two elements downstream using known methods.

The figure below depicts comparison of three resins in regard of separation of Sc and Th from REE with Y, trivial impurities and all metals except Sc and Th. We can state that if one needs to separate Th from all other metals, GS176 resin is the best of the three. For selective recovery of Sc, the S958 and GS176 are equally good. S956 has the best selectivity against all metals except thorium. However, S958 is about 2-fold times higher selectivity for scandium against thorium than S959 resin.

Recovery, separation and purification of REE and scandium from multicomponent solutions are difficult practical tasks. The chemistry of the systems technologists have to deal with are very complicated. Chemical behavior of the metals of interest, as well as of numerous impurities depends on their relative concentrations, anion background, and acidity. It is impossible to make the best choice of ion exchange resin for a particular case without significant experimental work.

The present work is an attempt to outline several technological tools, which can be useful in selected cases. These cases may be used as patterns of general character and applied to similar tasks. It is recommended to consider each actual case as a system comprising the metal of interest, the specific solution matrix and a resin. Changes to any of these three components will most probably require a re-examination of earlier conclusions arrived at from testwork, for example regarding parameters like the resin selected to achieve the desired separation, sequence of specific technological operations.

This work was possible due to support from Purolite Ltd. The author especially acknowledges valuable input from Dr. A. Tatarnikov, Moscow and his group who skillfully performed the experimental work.
ERECON (2014), Strengthening the European rare earths supply chain: Challenges and policy options, Kooroshy, J., G. Tiess, A. Tukker, and A. Walton (eds.).

British Geological Survey,  Natural Environment Research Council, Rare Earth Elements, (2011), A. Waters, P. Lusty, A. Hill (eds.).

Zhaowu Zhu, et al. (2015), Separation of uranium and thorium from rare earths for rare earth production – A review, Minerals Engineering. 77, 185–196.

John J. Barghusen, M. Smutz., (1957) Processing of Monazite Sands, Volume 947 of Ames Laboratory, Iowa State University (ISC) series, United States Atomic Energy Commission, Technical Information Service, and also Ind. Eng. Chem., 1958, 50 (12), 1754–1755

Goode, J. R., (2013), Thorium and rare earth recovery in Canada: the first 30 years. Canadian Metallurgical Quarterly, 52 (3), 234-242

Wood, Scott Al., (1990) The aqueous geochemistry of the rare-earth elements and yttrium, Chemical Geology 82, 159-186

Kenneth N. Han, et al., (2014), Opportunities and challenges for treating rare-earth elements, Geosystem Engineering 17 (3), 178-194

Tarwer, Russell Irwin, (1949) Purification of scandium using an ion exchange resin. Georgia Institute of Technology, USA, thesis

Hubicki, Zbigniew, (1990), Studies on Selective Separation of Sc(III) from Rare Earth Elements on Selective Ion-exchangers, Hydrometallurgy 23 (2-3), 319-331

Sokolova, Yu. V., (2006), Sorption of Sc(III) on Phosphorus-containing Cation Exchangers. Russian Journal of Applied Chemistry 79 (4), 573-578

Kazantsev, E.A., et al., (1975) Zh. Prikl. Khim.  48 (8)

M. Marhol et al., (1974), Selective Ion-Exchangers Containing Phosphorus in their Functional Groups I. Sorption and Separation of Some Bivalent and Trivalent Ions, J. Radioanal. Chem. 21 (1), 177-186

Joanna Kulczycka, et al. (2016), Evaluation of the recovery of Rare Earth Elements (REE) from phosphogypsum waste - case study of the WIZOW Chemical Plant (Poland). Journal of Cleaner Production 113, 345-354

A.V. Valkov, et al., (2014), Phosphogypsum technology with the extraction of valuable components, Procedia Chemistry 11, 176-181

Mugdha Walawalkar, et al. (2016), Process investigation of the acid leaching of rare earth elements from phosphogypsum using HCl, HNO3,and H2SO4, Hydrometallurgy 66, 195-204

Ines Hammas-Nasri, et al., (2016), Rare earths concentration from phosphogypsum waste by two-step leaching method, International Journal of Mineral Processing 149, 78–83

Takeshi Ogata, et al. (2016), Selective recovery of heavy rare earth elements from apatite with an adsorbent bearing immobilized tridentate amido ligands. Separation and Purification Technology 159, 157-160

C.K. Gupta, N. Krishnamurthy, (2005), Extractive Metallurgy of Rare Earths. CRC Press , ISBN 0-203-41302-4.
Habashi, F., (2013), Extractive metallurgy of rare earths, Canadian Metallurgical Quarterly 52 (3), 224-233

Chu Yong Cheng, et al., (2011), Metallurgical processes for scandium recovery from various resources: A review, Hydrometallurgy 108, 100-108

Borisenko, L. F. (1983), Raw-material resources of scandium, International Geology Review 25 (8), 942-946
Xu Shaoquan, Li Suqing., (1996), Review of the extractive metallurgy of scandium in China ( 1978-1991), Hydrometallurgy 42, 337-343