Introduction
Scandium has of late, become a metal of interest due to its use in lightweight alloys which have applications in the aerospace industry as well as hydrogen sponges used in fuel cells. These new applications along with the existing use of scandium in lasers, lighting, catalysts, and ceramics have led to an increase in demand for the metal [1]. Scandium , however, rarely congregates into economically viable ores or deposits and therefore is exclusively extracted as a by-product of other processes [2, 3]. At present, the world production of scandium is in the order of 10 to 15 Tpa in the form of scandium (III) oxide [1]. This is way below its demand and in 2003 only three mines produced scandium ; Uranium and Iron mine in Zhovti Vody in Ukraine, rare earth mines in Bayan Obo, China and the Apatite mines in the Kola Peninsula, Russia [2].
Potential other sources of scandium include bauxite residue [4, 5], the tailings of tantalum and niobium processing plants and electronic waste from which mainly precious metals like gold are currently being recovered. Patents by Black et al. [6] and Odekirk [7] describe the process of extracting scandium from tantalum residues, and both processes involve the use of toxic acids like hydrofluoric acid and a series of solvent extraction steps. These intensive process conditions are expensive for low-grade feeds and environmentally harmful.
An increasing environmental awareness and the growing demand for metals necessitates the development of more efficient metal recovery processes. Ionic liquids have recently been identified as alternative solvents for metal extraction , and a large amount of research into their use in hydrometallurgical processes has been conducted due to their desirable and environmentally friendly properties, and flexibility [8, 9]. Of particular interest for this research is the use of ionic liquids to directly dissolve metal oxides. Ionic liquids, which are molten salts at room temperature or ambient conditions, are completely composed of organic cations and inorganic or organic anions. They have several chemical and physical advantages over conventional chemicals and molecular organic solvents due to their ionic nature [10]. Among others, some of the advantages of ionic liquids are; negligible flammability and vapour pressure, high solvation ability, high selectivity and can be used as designer solvents. When the cation and anion of an ionic liquid are changed, the properties thereof can be adjusted to suit any requirement [11, 12].
A popular choice of ionic liquid used for both solvent extraction and leaching applications is betainium bis(trifluoromethylsulfonyl)imide or [Hbet][Tf2N] . Scandium has been extracted using this ionic liquid from solution and its efficacy has been demonstrated by many researchers such as Hoogerstraete et al. [13], and Onghena and Binnemans [14]. The use of [Hbet][Tf2N] as a lixiviant to extract the metals directly from solids has been described in literature for rare earth elements, and recently in a proof of concept study for scandium extraction from bauxite residue by Davris and co-workers [5]. Dupont and Binnemans [15] described processes in which [Hbet][Tf2N] was used as a lixiviant to leach the rare earths Nd, Dy, Y and Eu directly from their oxide from lamp phosphor waste. Their paper illustrates that [Hbet][Tf2N] can be used successfully as a lixiviant on low-grade rare earth ores or residues. The oxides of iron , cobalt , silicon , titanium , and aluminium are shown to be only slightly soluble in [Hbet][Tf2N] making it a potential lixiviant for scandium extraction from Columbite-Tantalite tailings which often contain these impurities [16, 17]. Davris and co-workers [5] also showed that about 35% of scandium could be recovered from bauxite residue with [Hbet][Tf2N] after 24 h of leaching at 333 K, 25% v/v water and a solid to liquid ratio of 1:10. However, this study only showed a preliminary investigation and little is known about the reaction between [Hbet][Tf2N] and scandium oxide.
![$$ Sc_{2} O_{3} + 6\left[ {Hbet} \right]\left[ {Tf_{2} N} \right] \to 2\left[ {Sc\left( {bet} \right)_{3} } \right]\left[ {Tf_{2} N} \right]_{3} + 3H_{2} O $$](../images/468727_1_En_230_Chapter/468727_1_En_230_Chapter_TeX_Equ1.png)
The only by-product of this reaction is water which is indicative of a potentially environmentally friendly process. It has also been shown by Onghena et al. [4] that scandium can be recovered and the ionic liquid recycled through subsequent scrubbing, stripping and precipitation .
Experimental
Reagents and Materials
Betaine hydrochloride (≥99%) and lithium bis(trifluoromethylsulfonyl)imide used in the synthesis of the ionic liquid , [Hbet][Tf2N] , were sourced from Sigma-Aldrich. AgNO3 solution and metal oxides, Sc2O3 (99.9%), Ta2O5 (99.99%), Nb2O5 (99.99%) and Al2O3 (99.99%) were also obtained from Sigma-Aldrich. All chemicals were used as supplied with no further purification .
The composition of the synthetic Columbite processing tailings matrix was obtained as follows. The processing of tantalum and niobium containing ores is usually undertaken with hydrofluoric acid and as a result the metals in the tailings from these processes are in a fluoride form [3]. These metal fluorides can however easily be converted to oxides by reaction with H2SO4 and subsequent roasting. The produced HF can then be recycled back to the Columbite processing plant. Consequently, the tailings matrix tested in this work is of an oxide form. The composition of the Columbite processing tailings to be tested was obtained from an unpublished source, and were normalised to include only Sc2O3 (8.52 wt%), Ta2O5 (4.28 wt%), Nb2O5 (48.86 wt%) and Al2O3 (38.34 wt%). The Sc2O3, Ta2O5 and Nb2O5 were chosen since they were metals of interest and the Al2O3 was used as it made up the second largest weight percent of the sample. Despite being the largest component iron was excluded since it can easily be extracted in a pre-processing step such as magnetic separation as described by Deveau and Young [20].
Synthesis and Characterization of [Hbet][Tf2N]
The ionic liquid synthesis method of Hoogerstraete et al. [13], was adapted for this study. 26.753 g (0.174 mol) betaine hydrochloride and 50 g (0.174 mol) lithium bis(trifluoro-methylsulfonyl) were added to 50 mL of water. The mixture was stirred at room temperature and left to refrigerate overnight. After phase separation , the ionic liquid was washed with small amounts of deionised water (about 5 ml aliquots) until no chloride ions were detected by the silver nitrate test. Residual trace amounts of water were evaporated at 383 K.
The structure of betaine bis(trifluoromethylsulfonyl)imide ([Hbet][Tf2N]) was confirmed by molecular spectroscopy . 1H NMR, 13C NMR and 19F NMR spectroscopy were measured on a Bruker AMX 400 NMR MHz spectrometer and reported relative to tetramethylsilane (δ 0.00). The Infrared spectrum was recorded using a Bruker, Tensor 27 platinum ATR-FTIR spectrophotometer in the range 4000−400 cm−1.
Experimental Procedure
All leaching experiments were conducted in 4 mL glass vials with TPFE screw-top lids. For every experiment, a measured quantity of metal oxide powder was added to the vial along with a corresponding stoichiometric volume of ionic liquid and enough water to facilitate dissolution as suggested by Nockemann et al. [17]. The stoichiometric volume of ionic liquid required for a metal oxide, with the formula MxOy, was determined as x times the oxidation state of the metal or x times y. All samples were agitated at 240 rpm at a set constant temperature using an orbital and reciprocating shaking incubator with digital temperature control .
The effect of temperature was investigated at 343, 353 and 363 K over a 72 h period. The water to ionic liquid (IL) ratio was kept constant at 5 mL water to 1 g IL [17] at a stoichiometric equivalent metal oxide ratio. The effect of water to IL ratio on extraction was investigated at 363 K at 2, 5, 7 mL water—1 g IL at a stoichiometric equivalent metal oxide ratio. The effect of IL to metal oxide ratio were also investigated by varying the stoichiometric required ratios from 0.5, 1 and 2 by keeping the mass of metal oxide constant and changing the solid liquid ratio. In addition to the single metal extraction experiments, an experiment was conducted on a metal matrix containing a mixture of metal oxides representative of Columbite processing tailings (composition described above) at 363 K, 5 mL: 1 g for 48 h.
After extraction , the samples were filtered with syringe filters (0.2 µm) to remove any undissolved oxides and diluted with deionized water. ICP-OES analyses were performed on the samples to analyse for metal concentration. The results were used to calculate the percentage extraction achieved which was defined as the weight of scandium ions present in solution (leached amount) over the weight of scandium initially present in the oxide sample.
Results and Discussion
The synthesized ionic liquid , [Hbet][Tf2N] , was characterised to confirm its identity and used in a number of extraction experiments with synthetic scandium oxide. Scandium was also leached from a mixture of metals typically found in the tailings of the niobium and tantalum extraction process from Columbite-Tantalite ore.
Characterisation of IL
FT-IR spectrum for synthesised [Hbet][TF2N]
On the 1HNMR spectrum of the synthesized [Hbet][Tf2N] the protons of the methylene group were observed as a singlet at 4.18 ppm. These protons appear downfield since they are bonded to a quaternary nitrogen atom carrying a positive charge. The same effect is evident on the three methyl groups directly attached to the quaternary nitrogen; their chemical shift appears at 3.30. The signal of the hydroxyl group is observed at 9.24.
1H NMR (400 MHz, CDCl3): δ 9.24 (1H, s, OH), 4.18 (2H, s, 2-Me), 3.30 (7H, s, 3 N-Me) ppm.
On the 13C NMR of betaine bis(trifluoromethylsulfonyl)imide spectrum the signal of carbon atoms of methyl groups lies at 53.92 ppm and the peak of methylene group resonate at 63.11 ppm. The quaternary carbon atoms of the bis(trifluoromethylsulfonyl)imide moiety are observed at 118.1 ppm. The signal of carbonyl group is observed at 165.7 ppm. The 19F NMR shows as peak at −80 ppm corresponding to the six fluorine atoms of the bis(trifluoromethylsulfonyl)imide moiety.
13C NMR (100 MHz, CDCl3): δ 165.7 (COOH), 121.2, 118.1, 63.11, 53.92, 30.92 ppm. 19F NMR (376 MHz, CDCl3): δ 79.67(-) (s, 6F). IR (νm cm−1): 3301, 1753, 1485, 1348, 1179, 1131, 1053.
Extraction of Scandium from Synthetic Scandium Oxide
Effect of temperature on the extraction of sandium (water to IL ratio of 5 mL: 1 g; 1:1 stoichiometric ratio of IL:Sc2O3 and 240 rpm
It can be seen that the extraction does not follow a linear relationship with time, but that for all temperatures the extraction over the 72 h period seems to have two kinetic regions, slower kinetics initially, followed by faster constant kinetics (linear trend) as time progress. This might be an indication that different mechanisms (diffusion and chemical kinetics) might be controlling the overall kinetics as the reaction progress. As temperature increase so does the extraction , with the highest extraction of 95% of scandium at 363 K for a stoichiometric ratio and water to IL ratio of 5 mL: 1 g after 72 h. Apart from its influence on the chemical reaction kinetics, temperature also has an effect on the viscosity of the water IL mixture. Low extractions at 333 K (32% compare to 95% at 363 K after 72 h) might be due to the fact that at 333 K the temperature is not sufficiently above the upper critical solution temperature (UCST) of 328 K to ensure that the water and ionic liquid form a completely homogeneous solution and not a biphasic mixture. Hoogerstraete et al. [13] reported that there is high diffusion in homogeneous mixtures as compared to biphasic mixtures which increases the overall kinetics and is observed from Fig. 2 where extraction increase significantly above 333 K. Also, according to the Stoke-Einstein equation, diffusion is inversely proportional to viscosity , therefore decreasing viscosity results in increasing diffusion [22]. These results are also in line with those reported by Dupont and Binnemans [15] for the leaching of systems of rare earth metals. These papers indicate that temperature additionally to ensuring a homogeneous mixture also reduces the viscosity of the ionic liquid thereby improving the movement of ions in the system as the leaching is believed to be mass transfer controlled. It should be stated that previous research on the extraction of Scandium with [Hbet][Tf2N] focussed solely on its extraction from a liquid solution and that these results show that it is possible to extract scandium directly from its oxide solid using the IL.
Effect of stiochiometric ration of IL: Sc2O3 on the extraction of sandium (363 K, 48 h and 240 rpm)
Effect of water to IL ratio on the extraction of scandium from scandium oxide at 363 K
As shown in Fig. 4, the extraction of scandium increase over the first 24 h of leaching for all water to IL ratios with 5 mL: 1 g showing the highest extraction of 30% compared with 27% (7 mL: 1 g) and 9% (2 mL: 1 g). Thereafter the reaction rate of the 2 mL: 1 g concentration increases rapidly and stays linear with time until the reaction is terminated after 72 h. Similarly, the reaction rates increase after 36 h for the 7 mL: 1 g and 5 mL: 1 g ratios. It is interesting to note that after 36 h of leaching , the extraction rate of scandium in a 7 mL: 1 g solution becomes higher than the rate of the 5 mL: 1 g solution and its overall extraction exceeds that of 5 mL: 1 g. The extraction with a 7 mL: 1 g solution also slows down and levels out after about 60 h of leaching , reaching a maximum extraction value of 98% of Sc.
As the total amount of liquid was kept constant and the solid to IL ratio was kept at the stoichiometric ratio of 1 mol Sc2O3: 6 mol IL, it can be said that theoretically there were enough ionic liquid present in each solution to react with all the Sc2O3. Taking this into consideration, it appears as though the water to IL ratio initially dictates the rate controlling step, but that with time as the amount of solid and IL decrease (reaction taking place) it appears as though the solid to ionic liquid ratio also starts to contribute to the observed rate. This is because the slope of the extraction vs time curve for the 5 mL: 1 g is smaller than that of the 2 mL: 1 g and 7 mL: 1 g curve for times exceeding 36 h. After 72 h of leaching , the total amount of extracted Sc is very similar for the different water to IL ratios 98% (7 mL: 1 g), 95% (5 mL: 1 g) and 92% (2 mL: 1 g), even though the extraction reaction is not complete for the 5 mL: 1 g and 2 mL: 1 g solutions as no levelling out can be observed for these systems. From an operational point of view, it therefore appears as if one can speed up the reaction of scandium with [Hbet][Tf2N] by increasing the water to IL ratio if the reaction time is kept between 36 and 60 h.
Extraction of Scandium from Metal Matrixes
Extraction percentages obtained after leaching for 48 h at 363 K from a matrix of metal oxides
Metal oxide | Composition of matrix (wt%) | Extraction (%) |
|---|---|---|
Sc2O3 | 8.52 | 47.75 |
Ta2O5 | 4.28 | 0.00 |
Nb2O5 | 48.86 | 0.20 |
Al2O3 | 38.34 | 15.66 |
The extraction of aluminium (15.7%) could be attributed to the presence of water in the mixture as reported by Dupont and Binnemans [15]. According to them, an increase in water compromises selectivity and hence the selection of the water to IL ratio of 5 mL:1 g even though a ratio of 7 mL:1 g gave better extraction results in the single metal experiments. As tantalum and niobium have similar behaviour due to their close ionic radii, it is expected that the extraction from their oxides would also be similar. A series of single metal extraction experiments on tantalum oxide (Ta2O5) (results not reported here) which were previously conducted, showed that tantalum is insoluble in the ionic liquid [Hbet][Tf2N] at these reaction conditions. This is very encouraging as Heftetjernite, ideally ScTaO4, and Ixiolite with the general formula which approaches Sc(Ta,Nb)O4 where scandium concentration reaches up to 18% can be processed to potentially selectively extract scandium with [Hbet][Tf2N] [25, 26].
Conclusion
The system proposed in this work for the leaching of scandium from Columbite processing tailings is an improvement on the environmentally damaging conventional methods for scandium recovery and may be of use in meeting the expected increasing demand for the metal. The proposed leaching system uses the ionic liquid [Hbet][Tf2N] , which has favourable and nontoxic properties, as a lixiviant rather than the strong mineral acids required in conventional processes. Preliminary investigations revealed that it was possible to leach scandium directly from its oxide with [Hbet][Tf2N] at reasonable process conditions (363 K) in an acceptable period of time (72 h) and achieve high extraction percentages (>92%). It was also shown that scandium could selectively be extracted with [Hbet][Tf2N] from a mixture of metal oxides, representative of metals present Columbite processing tailings over tantalum and niobium . This indicates that [Hbet][Tf2N] is a promising lixiviant for use in the recovery of scandium from Columbite processing tailings or other ores such as Heftetjernite or Ixiolite which contains these metals.
Acknowledgements
The authors would like to thank Michelle Kange, Michael Gustavo and Jenilee Ferreira for their assistance with some of the experimental work.