Electrochemical sulfur removal from chalcopyrite in molten NaCl-KCl

Accepted Manuscript Title: Electrochemical sulfur removal from chalcopyrite in molten NaCl-KCl Author: Mingsheng Tan Rui He Yating Yuan Zhiyong Wang Xianbo Jin PII: DOI: Reference:

S0013-4686(16)31593-6 http://dx.doi.org/doi:10.1016/j.electacta.2016.07.088 EA 27703

To appear in:

Electrochimica Acta

Received date: Revised date: Accepted date:

21-3-2016 13-7-2016 15-7-2016

Please cite this article as: Mingsheng Tan, Rui He, Yating Yuan, Zhiyong Wang, Xianbo Jin, Electrochemical sulfur removal from chalcopyrite in molten NaCl-KCl, Electrochimica Acta http://dx.doi.org/10.1016/j.electacta.2016.07.088 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Electrochemical sulfur removal from chalcopyrite in molten NaCl-KCl Mingsheng Tan, Rui He, Yating Yuan, Zhiyong Wang, Xianbo Jin* College of Chemistry and Molecular Sciences, Wuhan University, Wuhan, 430072, P. R. China *Corresponding author: Tel & Fax +86 27 68756319 E-mail address: [email protected] (X. B. Jin)

1

ABSTRACT Electrolysis of solid chalcopyrite (CuFeS2) against a graphite inert anode has been studied in equimolar NaCl-KCl melt at 700 °C. During electrolysis, S2- ions are released from the solid CuFeS2 cathode, transfer to the graphite anode and discharge to S2 gas. The reduction mechanism of CuFeS2 was investigated by cyclic voltammetry, potentiostatic and constant voltage electrolysis together with spectroscopic and scanning electron microscopic analyses. The reduction contains mainly three stages: the insertion of Na+ or K+ into CuFeS2, forming LxCuFeS2 (L=Na or K, x≤1); the partial reduction of LxCuFeS2 to Lx-wCuFe1-yS2-z and Fe; the complete reduction to a mixture of Cu and Fe, which can be magnetically separated. After the separation, pure Cu can be obtained by leaching out the residual Fe with acid. Electrolysis at a cell voltage of 2.4 V has led to a rapid reduction of CuFeS 2. The current efficiency and energy consumption were 85 % and 1.68 kWh/kg-CuFeS2, respectively. Keyword: chalcopyrite metallurgy, metal sulfides, molten salts, solid state electrolysis, copper.

2

Introduction Over 200 copper minerals were found in the earth’s crust, while only about 10 % have industrial value. Chalcopyrite (CuFeS2) is the most important mineral for Cu extraction. Presently, chalcopyrite is treated by the smelting process: CuFeS2 is firstly calcinated to FeO, CuO, Cu and SO2; after slagging the FeO with SiO2, the melting CuO is reduced to Cu by anthracite or natural gas. Obviously, this traditional process is environmentally unfriendly, which suffers from double emission of SO2 and CO2 [1]. To circumvent the handling of hazard SO2, various hydrometallurgical techniques have been tried, aiming to decompose the metal sulfide into metal and sulfur [2]. One approach is direct leaching, which involves the use of large amount of acid, and at the same time, oxidant is needed to simultaneously convert the sulfur ions to S to avoid H2S emission [2]. Another approach is direct electrochemical leaching. Similar to the industrial electrolysis of Ni3S2, the chalcopyrite concentrate is used as solid anode, during electrolysis, the metal ions are liberated into the solution, transfer to and deposit at the cathode. Unfortunately, the anode will be passivated promptly by the deposition of non-conducting sulfur [2, 3]. Molten salt electrolysis of Cu2S has been reported for decades. Obviously, pure Cu2S melt is not a suitable electrolyte due to its high electronic conductivity [4]. It was reported that addition of CuCl, an ionic melt, to a mole fraction of 70% or higher can significantly decrease the electronic conductivity of Cu2S [4, 5]. The Cu2S + CuCl 3

electrolyte with the mole fraction of Cu2S not exceeding 30% was thus suggested as electrolyte for the electrowinning of Cu [5]. The electrolysis temperature was suggested to be higher than the boiling point of sulfur (444.6 °C), and therefore, the anodic product could be always sulfur vapor to be simply swept out by argon. But relatively high current efficiencies were only obtained at very low apparent current densities (up to 15 mA/cm2) [5]. In the case of Cu2S + FeS + CuCl electrolyte, it is interesting that Cu can be selectively deposited. However, the low productivity and the formation of dendrite copper deposition at the cathode inhibits the practical application of these electrolytic processes [5]. More recently, molten salt electrolysis of Cu2S has been investigated in Cu2S + BaS melt at 1105 °C [6]. The advantages of this high temperature process includes high electrolysis current (2.5 A/cm2), and deposition of liquid copper to avoid formation of dendrite at the cathode, but the low current efficiency (9 ~ 28%) has to be improved. Alternatively, electrochemical split of various metal sulfides to respective metals and elemental sulfur have been realized by solid cathode electrolysis in molten chlorides [7-10]. During electrolysis, the metal sulfides and the resulting metals stay in solid so no occurrence of dendrite formation at the cathode. It is the S2- ions that leave the cathode, diffuse through the electrolyte, and discharge to S2 gas at the graphite anode. Unlike the molten salt electrolysis discussed above that require fusion or dissolution of metal sulfide, the solid metal sulfide cathode electrolysis requires an insignificant dissolution of the metal sulfide in the electrolyte. As reported, the electrolysis can proceed quickly as long as the electrolyte can promptly transfer S2- ions. 4

Solid cathode electrolysis of metal sulfides were initially investigated in CaCl 2 based molten chlorides that are often used for solid oxide electrolysis [11-13]. However, although the CaCl2 based electrolytes are good for O2- transfer, they have limited dissolution capacity for S2- ions, and CaS byproduct remains in the cathode even after prolonged electrolysis at temperatures above 800 °C. In addition, the residual CaO in the electrolyte due to the easy hydrolysis of CaCl2 results in anodic consumption of graphite consequently carbon contamination to the molten electrolyte and the metal product [9, 10]. On the contrary, NaCl and KCl almost cannot dissolve O2- ions but have great S2- solubility (for example, 39 mol/L in NaCl at 712 °C [14]). They are also much less hygroscopic than CaCl2. In molten NaCl-KCl, it has been reported that electrolysis of WS2 (MoS2) to W (Mo) and sulfur can proceed rapidly with high current efficiency, and no corrosion to the graphite anode takes place after long-term electrolysis [7, 8, 15]. This work aims to removal of sulfur from chalcopyrite by solid cathode electrolysis in the equimolar NaCl-KCl melt at 700 °C. The reduction mechanism of CuFeS2 has been elucidated according to experimental results from cyclic voltammetry (CV), potentiostatic and constant voltage electrolysis. We will also demonstrate the novel magnetism-assistant separation of Cu from the final cathodic product, a mixture consisting of mainly metallic Cu and Fe. Experiment The as-received analytical grade NaCl (352g) and KCl (448g) (99.5%, Sinopharm Chemical Reagent Co., Ltd, China) were mixed together and loaded into a graphite 5

crucible (inner diameter: 9.0 cm; height: 23.5 cm, AR grade, Jing Shi Ceramics Co., Ltd. Shanghai, China), which was then placed in a stainless steel reactor under argon (99.999%, Wuhan Iron and Steel (Group) Corp., China) protection. The temperature of the reactor was controlled by a programmable furnace. The equimolar NaCl-KCl mixture was heated to 300 °C and kept at 300 °C for 10 h to remove moisture, then at a heating rate of 3 °C/min the temperature was raised to 700 °C for the electrochemical experiments. Pre-electrolysis was carried out by applying 2.3 V between a nickel foil cathode and a graphite anode (15 mm in diameter, 99.99%, JingShi Ceramics Co., Ltd. Shanghai, China) in order to remove possible impurities in the melt. Cyclic voltammograms (CVs) of solid chalcopyrite (ore concentrate, containing about 87% CuFeS2 in mass with SiO2 and Al2O3 as the main impurities, Jinchuan Group Co., Ltd., Gansu, China) in the equimolar NaCl-KCl melt were recorded using a computer-assisted IviumStat (IVIUM Technologies B.V., Eindhoven, Netherlands) electrochemical workstation. The working electrode was a Mo metallic cavity electrode [16] (MCE, including two caves with 0.5 mm in both diameter and thickness) filled with the CuFeS2 powders (the total load of chalcopyrite was about 0.45 mg, corresponding to about 0.39 mg CuFeS2). The potential was controlled against a quartz sealed Ag/AgCl reference electrode [17]. For the constant potential or cell voltage electrolysis, the chalcopyrite powders were pressed (8 MPa) into cylindrical pellets (2.0 cm in diameter, ~1.90 g chalcopyrite or ~1.65 g CuFeS2), then sandwiched between two molybdenum meshes to form assemble cathodes. The 6

electrolysis was controlled by a multichannel four-electrode potentiostat (Neware, Shenzhen, China). After each electrolysis, the cathode was pulled out, cooled in argon, washed with distilled water and dried under vacuum at 80 °C. The product was characterized by X-ray diffraction spectroscopy (XRD; Shimadzu 6000, Kyoto, Japan, Cu-Kα, λ=1.54178 Å), and scanning electron microscope (FE-SEM; FEI Sirion Field Emission Gun SEM system, Eindhoven, Netherlands) together with energy dispersive X-ray (EDX; EDAX ,GENESIS 7000, AMETEK, Inc., Mahwah, NJ, USA) analysis. Results and discussion Fig. 1a and b show the CVs of the blank Mo cavity electrode (MCE) [16, 18] and the MCE loaded with CuFeS2 powders at 700 °C. At the blank MCE, the large cathodic current at about -2.4 V followed by a re-dissolution peak at about -2.3 V on the anodic branch (Fig. 1a), labeled as redox couple c/a, can be attributed to the reduction/re-oxidation of the cations from the melt. The small anodic peak at 0.05 V (labeled as a0) possibly corresponds to the oxidation of molybdenum. Two consecutive CVs of the CuFeS2 are shown in Fig. 1b. There are five reduction peaks c1- c5 and an oxidation peak a1 on the first CV. Considering that the SiO2 and Al2O3 impurities in the chalcopyrite concentrate are almost inert during the study, the multiple reduction peaks between 0 and -2.0 V suggest the complexity of electrochemical reduction of CuFeS2 in the NaCl-KCl melt. However, these reduction peaks could be hardly seen in the following cycles, indicating that all the CuFeS 2 powders in the cavity have been reduced at the first negative scan. The total integral charge between -0.3 V and -2.0 V (before the formation of Na or K) was calculated to 7

be about 0.8 C, in good agreement with the theoretical charge needed (~0.82 C) for the full reduction of ~0.39 mg CuFeS2. The peak a1 in Fig. 1b should then be the oxidation of metallic Cu and Fe, which was confirmed by comparing to the CVs of MCEs loading with Fe powders (Fig. 1c) and Cu powders (Fig 1d). In order to further confirm that all the sulfur of CuFeS2 in the MCE can be completely removed by the single cathodic scan, the reduction product after the scan was analyze. Fig. 2a shows the raw chalcopyrite powders, which consists of much nonuniform particles with some larger than 50 micrometers. After the first cathodic scan, the powders loaded in the Mo cave (Fig. 2b) became more porous (Fig. 2c), in line with the expected volume change after removal of S2- ions. The morphology also changed significantly (Fig. 2d). EDX analysis shows that the original strong peak of S in the precursor disappears in the product. These findings demonstrate that the reduction of CuFeS2 can proceed very quickly, which can be explained from five aspects: (1) the load of CuFeS2 in each Mo cave is relatively low (~ 0.2 mg); (2) the relatively large overpotential when scan to -2.2 V; (3) the electronic conductivity of the metal sulfide allows the reduction to take place at all the metal sulfide/melt interfaces; (4) the high porosity metal generated by removal of S2- allows fast ionic diffusion and (5) the NaCl + KCl melt has a very high S2- solubility (~39 mol/L in NaCl at 712 °C) [7, 8, 14,15]. In our previous study, we found that MoS2 in the NaCl-KCl melt could undergo electrochemical insertion of Na+ and/or K+ at the initial stage of reduction [7]. Na+ and K+ can also be inserted into CuFeS2 most likely combining with the valence 8

change from Fe (III) to Fe (II), forming compounds like NaCuFeS2 and KCuFeS2 [19-21]. CuFeS2 may undergo insertion of Na+ and/or K+ during reduction as well. CVs in different potential ranges were recorded to clarify the situation. Fig. 3a shows three consecutive CVs of CuFeS2 in the potential range of c1 and c2 (from 0.05 to -1.0 V) at a scan rate of 50 mV/s. It is clear that there are mainly two redox couples, i.e. c1/a1 and c2/a2. These peaks are fairly stable in the three cycles, and particularly, c1/a1 can be subdivided, featuring the continuous insertion/de-insertion behaviors. However, when the negative potential limitation was extended to -1.3 V, all the reduction and re-oxidation currents decreased significantly with the cycle number, suggesting the gradual consumption of CuFeS2. Considering that the c3/a3 redox couple is as reversible as the c2/a2 couple, c3 could be another insertion reaction. On the other hand, it is clear that the biggest reduction peak c4, which might correspond to the formation of metallic product, begins at about -1.2 V. In order to make further understanding of the electrochemical reduction process, a series of potentiostatic electrolysis of CuFeS2 pellets were carried out at different potentials between -0.9 V and -1.6 V, and the cell voltages were recorded simultaneously. The CuFeS2 in each pellet was about 1.65 g in mass, and the theoretical charge Q needed for complete reduction was about 3471 C. As shown in Fig. 4, when the imposed potential was -0.9 V, the current-time plot exhibited a typical initial large spike followed by a quick decay to the background level. The corresponding cell voltage change should include the variations of anodic polarization and Ohm drop, both would decrease with the current. We thus observed a high initial 9

cell voltage followed by a quick decrease to a plateau voltage near 0.7 V. Combined with the CV study, the small reduction current at -0.9 V is probably due to the formation of LxCuFeS2 (x≤1, L = Na or K) as a result of electrochemical insertion of Na+ and/or K+. The integral cathodic charge of about 261 C (notice that background current is neglectable) for the 1.65 g CuFeS2 pellet corresponds to the x in LxCuFeS2 of about 0.3. Indeed, several small peaks appeared at the XRD pattern of the -0.9 V electrolysis product (Fig. 5), likely due to the formation of inserted compound. The shape of current-time curve at -1.2 V was similar to that at -0.9 V (Fig. 4), with the reduction charge increased to about 560 C, corresponding to x = 0.55 in LxCuFeS2 if all the charge are for the insertion reaction. The ~ 15o XRD peak of the -0.9 V product shifted to ~13.5° (the characteristic peak of LCuFeS2), also indicating an increased x after the -1.2 V electrolysis. However, there was still a noticeable cathodic current after the 1 h electrolysis, which could be indication of another ongoing reaction, probably the reduction of LxCuFeS2 to Fe considering the appearance of XRD peak at about 45°. To confirm these analysis, elemental mapping of the -1.2 V and 1h electrolysis product was performed. Fig. 6a shows a typical SEM image of the product, which include two typical morphologies: aggregation of big and irregular granules and aggregation of fine dendrites. Distributions of elements Na, K, Cu, Fe and S are displayed in Fig. 6b~f. It is clear that the fine dendrites in the middle consist mainly of Fe, suggesting the formation of metallic Fe. On the other hand, Na, K, Cu, Fe and S distributed homogenously in the big particles, indicating the formation of Na+ and K+ inserted compound, particularly KxCuFeS2, considering the much stronger 10

signal of K than Na. The electrolysis at -1.5 V for 2 h resulted in a reduction charge of about 1560 C, which is much larger than the amount (865 C) needed for the formation LCuFeS2. The XRD pattern of the -1.5 V product shown in Fig. 5 indicates a mixture of LxCuFeS2, Fe and Cu. The XRD peaks of LxCuFeS2 after the 2 h electrolysis were still very strong, suggesting the slow reduction of LxCuFeS2 to Cu and Fe at -1.5 V, and the reduction was still ongoing considering the large cathodic current after the 2 h electrolysis (larger than the background current at -1.6 V). This could be largely due to the great kinetic barrier to overcome considering the recorded cell voltage was as high as 1.7 V, while the theoretical decomposition voltage of CuFeS2 to Cu, Fe and S2 gas is only about 450 mV (CuFeS2 = Cu + Fe + S2 (g), ΔGrΘ= 171.3 kJ/mol, 700 oC). Fortunately, when the cathodic polarization was increased to -1.6 V, the electrochemical reduction of CuFeS2 to Cu and Fe became very quick. The XRD pattern shown in Fig. 5 indicates that the cathode pellet has been completely reduced in the 2 h electrolysis. The current-time curve (Fig. 4) shows an initial large current, corresponding to the insertion reaction, followed by two current plateaus, possibly corresponding to the formation of Fe and Cu respectively. According to the XRD pattern of the -1.2 V electrolysis product, Fe was first generated during the reduction. Theoretically, the reduction of iron sulfide is also slightly easier than copper sulfide at 700 oC, considering the thermodynamic decomposition voltages of FeS and Cu2S are 516 mV (2FeS = 2Fe + S2 (g), ΔGrΘ= 199.4 kJ/mol) and 532 mV (2Cu2S = 4Cu + S2 (g), ΔGrΘ= 205.3 kJ/mol) respectively. Moreover, K, Cu, Fe ternary sulfides with 11

different stoichiometric ratios but the same crystal structure (such as K2Cu3FeS4 and KCuFeS2) have been reported. Thus, the step-by-step reduction of LCuFeS2 to metals is possible. The above findings allow to presume the main steps of the electro-reduction of CuFeS2 in the NaCl-KCl melts, which are represented by reactions (1) to (3) below: x L+ + CuFeS2 + x e = LxCuFeS2

(1)

LxCuFeS2 + (2z-w) e = L(x-w)CuFe(1-y)S(2-z) + yFe + wL+ + zS2-

(2)

L(x-w)CuFe(1-y)S(2-z) + (4-2z+w-x)e → Cu + (1-y)Fe + (x-w)L+ + (2-z)S2-

(3)

(L= Na+ or K+, 0≤w
voltage larger than 2.3 V (see the recorded cell voltage during electrolysis at -1.6 V, Fig. 4). Subsequently, constant cell voltage electrolysis was carried out to mimick the common industrial practices. The CuFeS2 in the cathode pellet was about 1.65 g, and the applied cell voltage between the solid cathode and the graphite anode was 2.4 V to ensure that both the thermodynamic and kinetic difficulties encountered in the reduction of the intermediates could be overcome. The current response is shown in Fig. 7, suggesting several reduction stages. These currents can be interpreted by linking to the XRD analyses (Fig. 8) of the partially and fully reduced products. The first stage lasted for about 3 ~ 6 min with the largest current, suggesting a very fast reaction (Fig. 7). The XRD pattern of the product after electrolysis at 2.4 V for 5 min (Fig. 8) shows exactly KCuFeS2 phase (JCPSD No. 51-0744), indicating that the x (in LxCuFeS2) could increase to 1 in a short time at a high voltage, and the electrochemical insertion of K+ ion via Reaction (1) should be of priority. This is also in line with our observation in Fig. 6, which indicate that the inserted ion is mainly K+. The 15~30 min product consisted of mainly Fe and a small quantity of Cu, indicating that the second stage (6~30 min) mainly corresponds to the reduction of LxCuFeS2 to Fe, which is still fast. The third stage between 30 and 80 min was relatively slow, and the 60 min product shows significantly increased Cu phase. These three reduction stages are also in good agreement with the above observation for electrolysis at a constant potential of -1.6 V. Both the constant potential and cell voltage electrolysis indicate that the electrochemical reduction of solid CuFeS2 in the NaCl-KCl melt undergoes an insertion reaction to LxCuFeS2, followed by a partial metallization with 13

Fe generated and finally a full metallization to Cu and Fe. The SEM image of the metallic product from the 120 min electrolysis at 2.4 V is shown in Fig. 9. There are basically two kinds of morphologies. One is aggregations of small particles with sizes less than 5 μm. The other is belt-like, with the width of about tens of micrometers. The EDX analysis indicate that the aggregations of small particles consist mainly of Fe while the big belts were mainly Cu. On the other hand, the anodic gas S2 has been led out by argon. After condensation, the anodic product was in a pale yellow color, which was proved to be S8 by XRD (not shown here). No corrosion to the graphite anode was found after electrolysis for one week. The observation that graphite is an ideal non-consumable anode is not only in line with the results of previous electrolysis of solid WS2 and MoS2 in the same electrolyte [7, 8], but also agrees with those observations during electrolysis of dissolved metal sulfides in molten chlorides [23, 24]. According to Fig. 7, there existed some background current after the completion of the chalcopyrite reduction, which could be of an electronic nature as often observed during molten salt electrolysis. The current efficiency was calculated to be about 85 % for the 120 min electrolysis of about 1.65 g CuFeS2 at a cell voltage of 2.4 V, and the energy consumption was about 1.68 kWh/kg-CuFeS2. We thus have demonstrated an efficient process to remove the sulfur ions from solid chalcopyrite and convert them to elemental sulfur. A mixture of Cu and Fe metals is the reduction product of solid chalcopyrite. It is interesting that Cu and Fe in the electrolysis product are phase separated and have great difference in particle sizes. 14

Various separation routes may be designed according to the different physical and chemical properties between the two metals. One possible way could be direct gravity separation, considering their size difference. Electrolytic refining in aqueous solution could be another way, taking advantage of the great different between the equilibrium dissolution/deposition potentials of Cu and Fe, which is as large as 0.78 V considering the Cu2+/Cu and Fe2+/Fe couples for Cu and Fe respectively. It is also possible to separate them by leaching in aqueous solution, noting that some acids can dissolve Fe but not Cu. Here a magnetism-assistant separation method has been implemented. Considering Fe is a magnetic metal but Cu is not, it is very possible to separate the two metals by magnetic force. As shown in Fig. 10a, after magnetic stirring of the electrolysis product in water, the iron powders were collected on the magnetic stirrer. The rest powders were let to settle and collected by filtering. XRD analysis shows that the collected purple bronze powders are mainly Cu, with some small diffraction peaks of Fe (Fig. 10b). The small amount of iron was then removed by leaching in HCl solution, and pure Cu was obtained as confirmed by the XRD analysis (Fig. 10b). These results indicate that after removal of S from the chalcopyrite, the post separation of Cu and Fe could be easy. Conclusions In summary, facile removal of sulfur ions from chalcopyrite (CuFeS2) through solid cathode electrolysis, with a mixture of Cu and Fe generated at the cathode and elemental sulfur generated at the graphite anode, can be achieved in NaCl-KCl at 700 °C. Cyclic voltammetry, potentiostatic and constant cell voltage electrolysis 15

together with XRD, EDX analysis and SEM observation showed that CuFeS2 was first reduced to LxCuFeS2 through the continuous electrochemical insertion of Na+ and/or K+. In the following reduction, Fe was firstly generated and LxCuFeS2 was converted to L(x-w)CuFe(1-y)S(2-z), and finally to a mixture of Cu and Fe. Rapid electrolysis of CuFeS2 has been realized at a cell voltage of 2.4 V with a current efficiency of about 85 %. In the final metallic product, the big belt-like Cu (10 ~ 30 μm in width) and small particles of Fe (less than 5 μm in particle size) were phase separated. Magnetic separation of Cu and Fe from each other was carried out and after acid leaching of the residual Fe in Cu, pure Cu was obtained. These findings can form the base for the development of a new, fast, and environmental friendly electrolytic process for the metallurgy of chalcopyrite and other sulfur-copper mines. Acknowledgements This work is supported by NSFCs (21173161, 20973130), the Fundamental Research Funds for the Central Universities.

16

Reference [1] W.G. Davenport, M. King, M. Schlesinger, A.K. Biswas, Extractive Metallurgy of Copper, Forth ed., Pergamon, Oxford, 2002. [2] S. Venkatachalam, Treatment of chalcopyrite concentrates by hydrometallurgical techniques, Minerals Engineering, 4 (1991) 1115-1126. [3] G. P. Power, The electrochemistry of the nickel sulfides—2. Ni3S2, Electrochim. Acta, 27 (1982) 359-364. [4] J.E. Enderby, A.C. Barnes, A Theory for the Electrical Conductivity of Molten Mixtures of Sulfides and Halides, J. Electrochem. Soc., 134 (1987) 2483-2485. [5] A.K. Garbee, S.N. Flengas, Electrical and Structural Properties of Metal Sulfides in Chloride Melts: The Systems and, J. Electrochem. Soc., 119 (1972) 631-641. [6] S. Sokhanvaran, S.-K. Lee, G. Lambotte, A. Allanore, Electrochemistry of Molten Sulfides: Copper Extraction from BaS-Cu2S, J. Electrochem. Soc., 163 (2015) D115-D120. [7] H. Gao, M. Tan, L. Rong, Z. Wang, J. Peng, X. Jin, G.Z. Chen, Preparation of Mo Nanopowders through Electroreduction of Solid MoS2 in Molten KCl-NaCl, Phys. Chem. Chem. Phys., 16 (2014) 19514-19521. [8] T. Wang, H.P. Gao, X.B. Jin, H.L. Chen, J.J. Peng, G.Z. Chen, Electrolysis of solid metal sulfide to metal and sulfur in molten NaCl-KCl, Electrochem. Commun., 13 (2011) 1492-1495. [9] X.L. Ge, X.D. Wang, S. Seetharaman, Copper extraction from copper ore by 17

electro-reduction in molten CaCl2-NaCl, Electrochim. Acta, 54 (2009) 4397-4402. [10] G.M. Li, D.H. Wang, X.B. Jin, G.Z. Chen, Electrolysis of solid MoS 2 in molten CaCl2 for Mo extraction without CO2 emission, Electrochem. Commun., 9 (2007) 1951-1957. [11] G.Z. Chen, D.J. Fray, T.W. Farthing, Direct electrochemical reduction of titanium dioxide to titanium in molten calcium chloride, Nature, 407 (2000) 361-364. [12] T. Nohira, K. Yasuda, Y. Ito, Pinpoint and bulk electrochemical reduction of insulating silicon dioxide to silicon, Nat. Mater., 2 (2003) 397-401. [13] X.B. Jin, P. Gao, D.H. Wang, X.H. Hu, G.Z. Chen, Electrochemical preparation of silicon and its alloys from solid oxides in molten calcium chloride, Angew. Chem. Int. Ed., 43 (2004) 733-736. [14] N. Shivgulam, D. Barham, W.H. Rapso, Sodium chloride (and) potassium: their effects on kraft smelt, Pulp Pap. Can., 80 (1978) 89-92. [15] H. Gao, L. Rong, M. Tan, Z. Wang, X. Jin, Z.G. Chen, Study of liquid ion diffusion in electroreduction of solid WS2 cathode to nanometer W in molten NaCl-KCl using an instantaneous ion release and diffusion model, Scientia Sinica Chimica, 44 (2014) 1354-1361. [16] G.H. Qiu, M. Ma, D.H. Wang, X.B. Jin, X.H. Hu, G.Z. Chen, Metallic cavity electrodes for investigation of powders, J. Electrochem. Soc., 152 (2005) E328-E336. [17] P. Gao, X.B. Jin, D.H. Wang, X.H. Hu, G.Z. Chen, A quartz sealed Ag/AgCl reference electrode for CaCl2 based molten salts, J. Electroanal. Chem., 579 (2005) 321-328. 18

[18] J.J. Peng, Y. Deng, D.H. Wang, X.B. Jin, G.Z. Chen, Cyclic voltammetry of electroactive and insulative compounds in solid state: A revisit of AgCl in aqueous solutions assisted by metallic cavity electrode and chemically modified electrode, J. Electroanal. Chem., 627 (2009) 28-40. [19] M. Oledzka, K.V. Ramanujachary, M. Greenblatt, Physical properties of quaternary mixed transition metal sulfides: ACuFeS2 (A = K, Rb, Cs), Mater. Res. Bull., 31 (1996) 1491–1499. [20] S. Conejeros, P. Alemany, M. Llunell, V. Sánchez, J. Llanos, L. Padilla-Campos, Structural Stability of Quaternary ACuFeS2 (A = Li, K) Phases: A Computational Approach, Inorg. Chem., 51 (2012) 362-369. [21] M. Oledzka, K.V. Ramanujachary, M. Greenblatt, New Low-Dimensional Quaternary Sulfides NaCuMS2 (M = Mn, Fe, Co, and Zn) with the CaAl2Si2-Type Structure:  Synthesis and Properties, Chem. Mat., 10 (1998) 322-328. [22] J. Llanos, A. Buljan, C. Mujica, R. Ramírez, Electron transfer and electronic structure of KCuFeS2, J. Alloy. Compd., 234 (1996) 40-42. [23] N.Q. Minh, R.O. Loutfy, N.P. Yao, The electrolysis of Al2S3 in AlCl3-MgCl2-NaCl-KCl melts, J. Appl. Electrochem., 12 (1982) 653-658. [24] Y. Xiao, D.W. van der Plas, J. Bohte, S.C. Lans, A. van Sandwijk, M.A. Reuter, Electrowinning Al from Al2S3 in Molten Salt, J. Electrochem. Soc., 154 (2007) D334-D338.

19

Figure Captions Fig. 1. Cyclic voltammograms (CVs) of (a) the blank Mo cavity electrode (MCE, with the SEM image inset), (b) MCE loaded with chalcopyrite (CuFeS2) powders (with the SEM image inset), (c) MCE loaded with Fe powders and (d) MCE loaded with Cu powders electrode in equimolar NaCl-KCl at 700 °C. Fig. 2. The typical SEM images of (a) chalcopyrite (CuFeS2) powders; (b) MCE loaded with CuFeS2; (c and d) the reduction product of (b) after the first cathodic scan from -0.2 V to -2.3 V (vs. Ag/AgCl) as shown in Fig. 1b. The insets show the corresponding EDX analysis. Fig. 3. Consecutive CVs of the MCE loaded with CuFeS2 powder in the potential range (a) 0.05 ~ -1.0 V and (b) 0.05 ~ -1.3 V at 50mV/s in the equimolar NaCl-KCl mixture at 700 °C. Fig. 4. Typical current-time plots recorded during the potentiostatic electrolysis of pellets at the indicated potentials in molten NaCl-KCl (700 °C). The inset shows the corresponding cell voltage-time curves. Fig. 5. XRD patterns of potentiostatic electrolysis products as indicated potentials at 700 °C in molten NaCl-KCl. Fig. 6. (a) SEM image and the corresponding elemental mapping (b) sodium (Na), (c) potassium (K), (d) copper (Cu), (e) iron (Fe) and (f) sulfur (S) of product obtained at 20

-1.2 V for 1 h. Fig. 7. The current-time plots recorded during the constant cell voltage electrolysis at 2.4 V. Fig. 8. XRD patterns of raw materials and products obtained at 2.4 V for different times (chalcopyrite pellets ca. 1.9 g). Fig. 9. Typical SEM images of the products from constant cell voltage electrolysis of chalcopyrite pellets at 2.4 V for 2 h. The inset is the corresponding select area EDX analysis. Fig. 10. The photos of powders obtained at 2.4 V for 2 h (a) after magnetic stirring, (b) XRD patterns of corresponding powders through different treatment process as indicated.

21

Figure 1 0.08

0.08

(a)

50mVs

-1

a0

a

0.04

Current/A

Current/A

0.00

blank

-0.08 -0.16 c

-2.5

0.0

c3 c2

c5

a0 c1

c4

-2.5

(c)

-2.0 -1.5 -1.0 -0.5 Potential/V (vs.Ag/AgCl)

(d) Fe powder, 10 mV/s

Current/A

Current/A

-0.04

0.02

0.02 0.00

-0.04 -2.5

a1

-0.12 -2.0 -1.5 -1.0 -0.5 Potential/V (vs.Ag/AgCl)

0.06

-0.02

1st cycle 2nd cycle -1 50mVs

0.00

-0.08

-0.24

0.04

(b)

blnak, 50 mV/s -2.0 -1.5 -1.0 -0.5 Potential/V (vs.Ag/AgCl)

0.0

Cu powder, 10 mV/s

0.00

-0.02 blank, 50 mV/s -0.04 -2.4 -2.0 -1.6 -1.2 -0.8 -0.4 Potential/V (vs.Ag/AgCl)

0.0

22

0.0

Figure 2 S

Cu Si Fe O Al C 0

2

A

B

D

C

Fe

Cu Fe Cu

4 6 Energy/keV

8

10

Cu Cu Fe Fe O Al Si C 0

2

Fe 4 6 Energy/keV

Cu 8

10

23

Figure 3

0.03 a2

Current/A

0.02

a1

0.01

1st 2nd 3rd

0.00 -0.01 -0.02

c1

50 mV/s

c2

-0.03 -1.2

-0.8 -0.4 0.0 Potential/V (vs.Ag/AgCl)

0.03

Current/A

(a)

0.02 1st 2nd 0.01 3rd

a2

a3

a1

(b)

0.00 -0.01 c4

c1

-0.02 -0.03

c3 -1.2

50 mV/s

c2

-0.8 -0.4 0.0 Potential/V (vs.Ag/AgCl)

24

Figure 4

4

3

Cell Voltage/V

Current/A

3

2

1

-1.6V, 2h -1.5V, 2h -1.2V, 1h -0.9V, 1h

2

1

0

0

30

60 Time/min

90

120

0 0

30

60 Time/min

25

90

120

Figure 5 1 LxCuFeS2 2 Fe 3 Cu

1

1

1

1

11 1

1



2 3

-1.6V,2h 3

2

3

2 3 11

-1.5V,2h

21

-1.2V,1h

1

1

1 1

1

-0.9V, 1h raw

10

20

30

40 50 60 2Theta/degree

26

70

80

Figure 6 (a)

(b)

(d)

(e)

C

(c)

N F

27

(f)

K S

Figure 7

6

Current/A

4

2 2.4 V, 2h 0

0

20

40

60 80 Time/min

28

100 120

Figure 8





LxCuFeS2

 Cu



 Fe



2.4V, 60min   2.4V, 30min

 

2.4V, 15min



  

2.4V, 5min raw

10

20

30 40 50 60 2Theta/degree

29

70

80

Figure 9 Fe

Fe

Fe

0

2

4 6 Energy/keV

8

10

Cu

Cu Cu

Fe

0

30

2

4 6 Energy/keV

8

10

Figure 10

(a)

Cu

(b)

Acid washing Cu

Cu

Magnetic separation Fe 2.4V, 2h Fe

30

40

31

50 60 2Theta/degree

70

80