Articles
Direct Tungsten Extraction from Scheelite in a Molten Salt Media
2024 年 92 巻 4 号 p. 043023
詳細
2024 年 92 巻 4 号 p. 043023
Tungsten is a basic metal commodity that is classified as critical raw material (CRM) by the European Commission (EC) with the highest economical importance compared to other CRM. Thanks to its unique properties, secure W supply is critical to all industrial applications involving cutting or component wear, such as mining, machining, construction, tools and dies. Other important uses are in high-strength steels and high-temperature alloys, chemicals, mill products and lighting filaments. The production of W metal or carbide to be used in end products, requires a reduction process of the oxide which has been previously extracted from the primary (ores) resources by complex and energy intensive hydrometallurgical processes.
In the frame of the EC-funded TARANTULA project (GA 821159), innovative methods to obtain W metal directly from scheelite raw material have been investigated. The process comprises two steps, i.e., selective chlorination of the W ore in a molten chloride media using gaseous reactants, and subsequent electrolysis of the dissolved W electroactive species from the same reaction media. The chlorination of natural scheelite in a molten chloride has been demonstrated in the equimolar NaCl-KCl mixture at a working temperature of 727 °C, using both Cl2 (g) and HCl (g). The dissolved W species in the molten chloride were found to be tri-tungstate: W3O102− and/or the chloro-complex, as e.g., W3O10Cl24−. Subsequent electrolysis trials demonstrated the recovery of WC2 deposits on a carbonaceous cathode, while the anode reaction was evidenced to include the discharge of the oxide ions from the dissolved tri-tungstate species.
Tungsten is a basic metal commodity underpinning the global manufacturing sector and its market growth outperforms that of the Global GDP. European and US end-users tend to sign long term off-take supply contracts.1 W is classified as Critical Raw Materials (CRM) by the European Commission, as it has been positioned always at the highest economic importance compared to all other CRM on the EC lists since 2011.2 It has unique properties such as highest melting point among all metals, high density, high tensile strength, as well as excellent heat and corrosion resistance. Thus securing W supply is critical to all industrial applications involving cutting or component wear, such as mining, machining, construction, tools and dies. The increase in demand (quantity and diversity) and prices of metal and mineral commodities have led to a renewed interest in recovering W from underexploited side streams in a sustainable way. Besides, the development of complementary ground-breaking processes for the recovery of W metal at superior recuperation rates compared to state-of-the-art (SoA) processes, are sought.
The production of W metal or carbides to be used in end products, requires a reduction process of the oxide or the salt once those are extracted from the primary (ores) or other secondary (wastes) resources by complex and energy intensive hydrometallurgical processes, which can be different depending on the ore, scheelite or wolframite concentrates, but in general including:3 i) leaching or digestion, which can be alkali leaching with NaOH, pressure leaching with Na2CO3, or acid leaching with HCl; ii) filtration and silicon precipitation; iii) solvent extraction to eliminate sodium; iv) crystallization to produce ammonium polytungstate (APT). The SoA reduction processes are typically involving the use of hydrogen,4 carbon5 or other reactive metals to produce the metal under extreme conditions of high temperature and reduced pressure.
Compared to SoA technologies, electrodeposition requires much simpler equipment and allows better control on product morphology. However, the electrodeposition of refractory elements as W from aqueous solutions is rather challenging, since the electroreduction of the metal will only occur when the hydrogen formation is blocked, which is influenced by both thermodynamic and kinetic effects. Therefore, conventional electrodeposition in aqueous solutions can only yield deposits with alloying elements, not pure refractory metals, unless highly alkaline solutions as plating baths are used.6 Then the use of molten salts as electrolytes offers a valuable alternative to produce W metal or carbide deposits by electrolysis. This can be enhanced by the fact that molten salts offer interesting possibilities as solvents for the treatment of ores by processes that are mid-way between pyro-and hydrometallurgy.
Literature on electrochemical behaviour of dissolved scheelite species in molten salts is scarce. Erdogan and Karakaya investigated the electrochemical reduction of CaWO4 and WO3 to produce tungsten metal powder from pure CaCl2 and CaCl2-NaCl electrolytes at temperatures ranging from 600 to 900 °C.7 The authors demonstrated the formation of metallic tungsten powder in the pure CaCl2 melt at 900 °C. However, significant amounts of tungsten losses occurred as a result of the formation of volatile WO2Cl2 compound. The ability of dissolving tungsten-oxocompounds like scheelite CaWO4 in molten chlorides has been reported in the literature from a theoretical point of view.8,9 The authors studied both oxobasic (O2− donor) and oxoacidic (O2− acceptor) media. In the case of the later, it has been found that scheelite behaves as an oxobase, and that the solubility of hexavalent-tungsten species increases when increasing the pO2−. This can be achieved by addition of a sufficient strong oxoacid gaseous reactant, like Cl2 or HCl gas. The dissolution reaction (chlorination) of scheelite in a molten chloride media can occur effectively at much lower temperatures than that of solid-gas phase chlorination processes, thus proceeding more selectively in the presence of impurities.
This work is dealing with the investigation of an innovative method to produce W metal or carbides directly from scheelite raw material by selective chlorination of the W ore in a molten chloride media using gaseous reactants, and subsequent electrolysis of the dissolved W electroactive species.
90 g of the equimolar NaCl-KCl molten salt mixture was mixed with 10 g of natural scheelite in a quartz crucible. The salts and scheelite material were dried at 120 °C for at least 48 h prior to use. The loaded crucible was placed in an air-tight quartz cell with a water-cooled quartz lid with holes that allowed the insertion of a thermocouple, gas in, sampling tubes and electrodes. The glass reactor was placed in a vertical furnace with a mullite liner and kept under inert atmosphere (Ar 99.999 %). The working temperature was chosen to be 727 °C, as oxoacidic properties of the NaCl-KCl mixture have been widely studied at this specific temperature, which corresponds to 1000 K.10 The temperature inside the melt was measured using a thermocouple type S (Pt-Pt 10 %Rh) shielded by a closed-end alumina or quartz tube and was monitored and recorded by means of a multichannel Keithley 2000 Multimeter. Cl2 and HCl gas, of purity 99.999 % and 99.8 %, respectively, were used during the chlorination experiments.
To absorb the unreacted Cl2 and/or HCl gas coming from the cell, 6M sodium hydroxide solutions inside gas washing bottles were placed in the out-gas line. This proved to be a very efficient scrubber, and as long as the sodium hydroxide was not saturated, no Cl2/HCl gas would pass these bottles. Also, a liquid N2 (LN) cooling trap was positioned at the end of the gas line to an (eventual) controlled condensation of high vapour pressure chlorides. A U-tube was connected in parallel with the out-gas line to ensure an alternative escape route for the off gas in case of clogging of the NaOH scrubbers or LN trap and thus avoiding pressure build up in the cell. Moreover, several other parts related to HSE issues were installed, i.e. gas alarms, pressure gauge and transmitter and pressure relieve valve. The gauge allowed for continuous monitoring of the cell inlet pressure and thus early observations of unexpected pressure build ups, the relieve valve would ensure that in such an event, this would be released in a controlled manner and over-pressures avoided. The whole set-up was installed in a walk-in hood and was explained elsewhere.11
Samples of the molten bath were taken prior to and after the chlorination experiments, using a quartz tube with a quartz frit to avoid solid particles. The samples were cooled down in a desiccator and there stored until analysed by ICP-MS (Inductively Coupled Plasma Mass Spectrometry). The oxygen content of the samples was analysed by inert gas fusion technique using a LECO ON836 apparatus. Argon gas was introduced to strip the melt of chlorination gas (Cl2 or HCl) and remove it from the gas line before opening the cell for further manipulation, like for taking melt samples.
After the chlorination trials were finished, and in order to determine the electroactive species formed after chlorination, electrochemical characterization of the melt was carried out, mainly by CV (cyclic voltammetry). This was done with the help of potentiostat/galvanostat PARSTART 4000+, using a 3-electrode set-up with 3 mm diameter GC (glassy/vitreous carbon) rods as working and counter electrodes, and 1 mm W wire as reference electrode (RE). Potentiostatic electrolysis on the GC substrate was applied at different cathodic potentials, and the deposits obtained were analysed by SEM-EDS (scanning electron microscope combined with energy-dispersive X-ray spectroscopy).
For the solubilization of an insoluble oxide in a molten chloride solvent, i.e., chlorination, a sufficiently high pO2− in the molten salt needs to be stablished by the chlorinating gaseous agent. This can be easily achieved by the oxidation of the O2− ions from the oxide into oxygen by the addition of a sufficiently strong oxidizing reactant, like Cl2 gas. In this case, the (general) equilibria showed in Eq. 1 is established:
| \begin{equation} \text{M$_{2}$O$_{x}$ (s)} + \text{$x$ Cl$_{2}$ (g)} \leftrightarrows \text{2 MCl$_{x}$ (diss)} + \text{$x/2$ O$_{2}$ (g)} \end{equation} | (1) |
Chlorination of natural scheelite (CaWO4) in the equimolar NaCl-KCl mixture (scheelite/salt ratio: 1/9) at 727 °C, using gaseous Cl2 (100 ml min−1) as chlorinating agent, was carried out at various chlorination times, i.e., 1, 2, 3, 4 and 5 h.
The Ca and W contents in the bath samples extracted after each chlorination run were determined by ICP-MS, while the oxygen contents were obtained by inert fusion method. Taking into account the Ca, W and O contents in the initially added scheelite, the amounts of the elements in the bath was determined. The results obtained in terms of M recovery (yield in bath) are presented in Fig. 1.
Chlorination yields of elements Ca, W and O vs. time (h) using Cl2 (g) as chlorinating agent of scheelite in the equimolar NaCl-KCl at 727 °C.
The results showed that there is a certain solubility of the scheelite in the molten chloride at the operating temperature, and that ∼15 % of scheelite was dissolved after almost 22 h in contact with the NaCl-KCl melt, however, the process is very slow.
Following the amounts of Ca and W present in the bath samples, one can say that >95 % of the scheelite was dissolved after 2–3 h, according to Eq. 2:
| \begin{align} &\text{CaWO$_{4}$ (s)} + \text{3 Cl$_{2}$ (g)} \notag\\ &\quad\rightarrow \text{WCl$_{6}$ (diss)} + \text{3/2 O$_{2}$ (g)} + \text{CaO (diss)} \end{align} | (2) |
However, the amount of W in the bath samples are decreasing at chlorination times higher than 2–3 h. This is probably due to evaporation losses, as confirmed by the significant amount of solid recovered in the cold zones of the reactor. Moreover, the amount of O content in the samples also decreased with time, which may lead one to think that the dissolved tungsten species exist as oxo-compounds.
Figure 2 illustrates the O/W ratios obtained from the bath samples (before and after chlorine was added) and recovered condensate from the reactor lid. These values point out that most likely, the scheelite is dissolved as tri-tungstate species, i.e., W3O102− and/or the chloro-complex as e.g., W3O10Cl24−. This shows that the dissolved scheelite species formed as in Eq. 2 are preferably coordinated by oxide, rather than chloride ions, forming soluble poly-oxo tungstate structures, as suggested by Combes and Tremillon.12 Moreover, the results point out that the solid condensate is WO2Cl2, as confirmed by XRD analysis (cf. Fig. 3). The formation of gaseous WO2Cl2 has been demonstrated previously in molten chlorides.6
O/W ratios analysed in the bath samples before and after chlorine was added, and in the recovered condensate. W content analysed by ICP-MS and O by inert fusion method.
XRD pattern of the condensate. (red stars) WO2Cl2; (blue circles) NaKCl; (magenta upside down triangle) NaCl.
The overall efficiency of W recovery from the scheelite input material was found to be close to 100 %, of what ∼50 % was converted to dissolved tri-tungstate species in the molten chloride, and ∼50 % recovered from the gas phase as condensed WO2Cl2.
3.2 Chlorination trials using HCl gasHCl gas can also be used as chlorinating agent to solubilise a metal oxide. In this case, HCl is a strong enough oxoacid reactant, which combines with the O2− anion of the insoluble compound, giving H2O as the conjugated oxobase. The (general) equilibria stablished is showed in Eq. 3.
| \begin{equation} \text{M$_{2}$O$_{x}$ (s)} + \text{$2x$ HCl (g)} \leftrightarrows \text{2 MCl$_{x}$ (diss)} + \text{$x$ H$_{2}$O (g)} \end{equation} | (3) |
Chlorination of natural scheelite (CaWO4) in the equimolar NaCl-KCl mixture (scheelite/salt ratio: 1/9) at 727 °C, using gaseous HCl (30 ml min−1) as chlorinating agent, was carried out at various chlorination times, i.e., 1, 2, 3, 4 and 5 h, following the same procedure as in the case of gaseous Cl2, and the dissolution reaction in this case is showed in Eq. 4:
| \begin{align} &\text{CaWO$_{4}$ (s)} + \text{6 HCl (g)} \notag\\ &\quad\rightarrow \text{WCl$_{6}$ (diss)} + \text{3 H$_{2}$O (g) } + \text{CaO (diss)} \end{align} | (4) |
The results obtained were comparable with both gaseous chlorinating agents, as showed in Figs. 4 and 5, and the formation of dissolved poly-oxo tungstate structures is also evidenced in this case. However, the dissolution reaction is somewhat slower than with Cl2 and 3 h are needed to chlorinate ∼95 % of the scheelite material. Besides, the O content of the condensate is significantly higher than when using chlorine. This is possibly due to the hydrolysis of the WO2Cl2 compound with the H2O (g) released during the chlorination reaction, as shown in Eq. 5:13
| \begin{equation} \text{WO$_{2}$Cl$_{2}$ (s)} + \text{2 H$_{2}$O (g)} \rightarrow \text{WO$_{3}{\cdot}$H$_{2}$O (s)} + \text{2 HCl (g)} \end{equation} | (5) |
In this case, the overall efficiency of W recovery was found to be ∼60 %, and of that, ∼10 % was recovered from the gas phase, the rest remaining in the chloride melt as dissolved tritungstate species. 60 % overall efficiency is truly underestimated, probably due to the difficulties in the recovery of the volatile W species from the off-gas system, and so uncertenties in the mass balance calculations.
Chlorination yields of elements Ca, W and O vs. time (h) using HCl (g) as chlorinating agent of scheelite in the equimolar NaCl-KCl at 727 °C.
O/W ratios analysed in the bath samples before and after HCl was added, and in the recovered condensate. W content analysed by ICP-MS and O by inert fusion method.
The electroactive W species in the NaCl-KCl melt after chlorination with either gaseous Cl2 or HCl were investigated by CV. Figure 6 shows the voltametric curves obtained on a GC substrate, compared to those obtained from a pure NaCl-KCl melt on GC and W electrodes.
Cyclic voltammetric curves obtained on a GC substrate after the chlorination trials of scheelite (black continuous line) using HCl (g). The CVs obtained in the pure NaCl-KCl mixture (equimolar composition) are also shown on GC (grey continuous line) and W (grey discontinuous line) substrates. Sweep rates 200 mV s−1.
The electrochemical stability of the molten NaCl-KCl mixture is limited on the cathodic side by the reduction of sodium ions to sodium (A), which remains on the tungsten substrate giving rise to the stripping peak (A′). On the GC substrate the cathodic potential window is reduced due to intercalation of Na on the GC structure, leading to an underpotential deposition of sodium ions, i.e., deposition at activities lower than 1 (B). In this case, the re-oxidation signal is not the typical stripping peak (B′), as in the case of the W substrate, showing that the anodic oxidation of the Na intercalated in the GC structure is a slow process. Moreover, using a GC working electrode allowed to study the anodic potential window of the melt, which is limited by the oxidation of chloride ions into chlorine gas (C′). The Cl2 gas bubbles remain somewhat adsorbed on the GC substrate, thus giving rise to a reduction wave on the reverse scan (C).
Comparing the CVs from the pure melt, the voltametric curves obtained from the molten bath after the chlorination trials showed two sets of distinguishable electrochemical signals at potential ranges −0.75 to 0.25 V (zone 1) and −1.75 to −1 V (zone 2). Besides, the anodic signal B′ is sharper, indicating that not only Na ions are reduced, but also Ca ions obtained in the chlorination of the scheelite, as showed in Eqs. 2 and 4. This supports the formation of the tungstate species, stemming from the chlorination products obtained as showed in Eqs. 2 and 4, i.e., dissolved W species and dissolved CaO, and subsequent formation of dissolved poly-oxo tungstate compounds. In addition, there is a certain depolarization on the anodic oxidation signal C′ compared to that of the pure melt. This indicates the anodic discharge of the O2− ion-species prior to the chlorine evolution reaction, according to the Eq. 6:
| \begin{equation} \text{W$_{3}$O$_{10}{}^{2-}$ (diss)} + \text{1/2 C} \rightarrow \text{1/2 CO$_{2}$ (g)} + \text{3 WO$_{3}$} + \text{2e$^{-}$} \end{equation} | (6) |
Figure 7 shows CVs on GC substrate at different cathodic potential vertex. The shape of the corresponding anodic signals (D′ and E′) show the formation of insoluble species, most likely related to the formation of W metal and/or carbide in both cases from different dissolved W-containing species.
Examples of cyclic voltammograms obtained on a GC substrate at different shift potentials from a NaCl-KCl after the chlorination trials of scheelite with HCl (g). Sweep rate 200 mV s−1.
Senderoff and Mellors reported the existence of various polynuclear and mononuclear ions of W and Mo,14 and that complex decomposition equilibria exist, with slow or faster kinetics depending on the concentrations and temperatures. Although the W and O contents of the bath samples from the chlorination trials gave a O/W ratio of ∼3.5 (cf. Figs. 2 and 5), one cannot rule out the existence of several polynuclear and even mononuclear W ions, with an average corresponding to the tri-tungstate ions.
The reduction of aluminium metal from different chloroaluminate species has been reported to be from the different Al-Cl complexes, being the dimer Al2Cl7− ion easily reducible than the AlCl4− ion,15 with the consecutive reduction reactions showed in Eqs. 7 and 8:
| \begin{equation} \text{Al$_{2}$Cl$_{7}{}^{-}$} + \text{3e$^{-}$} \rightarrow \text{Al} + \text{7 AlCl$_{4}{}^{-}$} \end{equation} | (7) |
| \begin{equation} \text{AlCl$_{4}{}^{-}$} + \text{3e$^{-}$} \rightarrow \text{Al} + \text{4 Cl$^{-}$} \end{equation} | (8) |
Therefore, similar mechanism can be possible in the case for the W-dissolved complex species as reported by Combes and Tremillon.12
3.4 Recovery of W/WC by electrolysisPotentiostatic electrolysis on a GC substrate was conducted at a constant applied potential of −0.4 V vs. W wire RE for 40 minutes. The deposit obtained had poor adherence and in dendritic form, as showed in Fig. 8A. SEM-EDS analysis of the cross-section evidenced the formation of WC2 (cf. Fig. 8C). Moreover, the GC rod used as CE was consumed (cf. Fig. 8B), thus confirming the formation of CO2 in the anodic discharge reaction from the tri-tungstate compounds, cf. Eq. 6.
(A) Picture of the GC rod cathode after electrolysis of the W-O-Cl species obtained after chlorination of scheelite with HCl (g) in the NaCl-KCl at 727 °C. Potentiostatic electrolysis at −0.4 V vs. W RE applied potential. (B) Picture of the consumed GC rod counter electrode after the electrolysis; (C) SEM image of the cross-section of the GC cathode with the deposit obtained and EDS point analysis (average representative value in wt%).
Analysis of the deposits obtained at cathodic potentials ∼ −1.2 V vs. W RE for almost 100 minutes, did not provide conclusive results of the existence of W and/or WCx, and further investigations of the E/E′ electrochemical exchange are needed, as well as optimization of the best conditions to obtain smooth layers.
An innovative approach for the direct extraction of W metal from scheelite has been investigated. The process consists of two steps, namely, selective chlorination in a molten chloride media using gaseous reactants, and subsequent electrolysis of the dissolved W electroactive species from the same reaction media as electrolyte. Both gaseous Cl2 and HCl were chosen as chlorination agents, as it is easier to remove their excess, thus not affecting the subsequent electrolysis process. The chosen molten chloride media was the equimolar NaCl-KCl mixture at a working temperature of 727 °C. Similar results were obtained with both Cl2 (g) and HCl (g), and the dissolved W species from the natural scheelite feedstock were found to be tri-tungstate: W3O102− and/or the chloro-complex as e.g., W3O10Cl24−. Some W was also recovered from the gas phase, condensed in the cold zones of the reactor, as WO2Cl2 or hydrated WO3 compounds, the later obtained in the case of HCl (g) due to the corresponding associated oxo-base being H2O (g), introducing humidity in the gas phase. The overall efficiency of W recovery was almost 100 % in the case of Cl2 (g), ∼50 % recovered from the vapour phase, and ∼60 % in the case of HCl (g), ∼10 % recovered from the vapour phase. The latest value is truly underestimated, probably due to the difficulties in the recovery of the volatile hydrated W-species from the off-gas system.
Electrochemical characterization of the dissolved W species obtained after the chlorination trials showed two different reduction reactions, both leading to W metal from different polynuclear and/or mononuclear W ions. Subsequent electrolysis trials demonstrated the recovery of WC2 deposits on a carbonaceous cathode, and that the anode reaction involves the discharge of the oxide ions from the dissolved tri-tungstate species.
This work has been funded by the European Union’s EU Framework Programme for Research and Innovation Horizon 2020 under Grant Agreement No. 821159.
Ana Maria Martinez: Conceptualization (Lead), Data curation (Lead), Funding acquisition (Lead), Investigation (Lead), Methodology (Lead), Project administration (Lead), Resources (Lead), Supervision (Lead), Validation (Lead), Writing – original draft (Lead), Writing – review & editing (Lead)
Anne Støre: Data curation (Equal), Formal analysis (Equal), Investigation (Supporting), Methodology (Equal), Resources (Equal), Validation (Equal)
Karen Osen: Conceptualization (Equal), Data curation (Equal), Investigation (Equal), Supervision (Equal)
The authors declare no conflict of interest in the manuscript.
H2020 European Research Council: 821159
A part of this paper has been presented in 2023 Joint Symposium on Molten Salts (Presentation 1A09).
A. M. Martinez, A. Støre, and K. Osen: Equal Contribution
