2.1 Materials

LiCl, KCl (99%) and NiCl₂ (99.9%) (Sigma-Aldrich) were used without further purification; the nickel-chromium alloy (Menghe resistor factory, Changzhou, China) contained 72.62 wt% nickel and 21.84 wt% chromium. Other impurities (5.54 wt%) included Si, Sb, Mn, Al, Fe, Ti, and Mo. The LiCl-KCl eutectic salt was prepared according to previously reported procedures [23,24] and the NiCl₂-LiCl-KCl melt was obtained by mixing LiCl-KCl eutectic and NiCl₂ followed by heating at 723 K for 2 h.

2.2 Electrochemistry

Electrochemical experiments were carried out using a set up described in detail in a previous report [25,26]. Briefly, the measurement set up consisted of an argon glove box, an electric resistance furnace with a stainless-steel vessel and an off-gas treatment unit. The concentrations of O₂ and H₂O were maintained at < 1 ppm inside the argon glove box. LiCl-KCl-NiCl₂ mixtures were loaded into an alumina crucible placed in the stainless-steel vessel, which was heated to 723 K using a furnace that precisely maintained this temperature with variations less than ~ 1 K. All the electrochemical experiments were carried out at 723 K.

A three-electrode system was used for the electrochemical investigations. The working electrode for cyclic voltammetry (CV) and square wave voltammetry (SWV) measurements was a tungsten wire (99.9%, Φ1, Alfa-Asar), cleaned by ultrasonic treatment in diluted HNO₃ (1.6 mol/dm³) and dried overnight under vacuum. Electrorefining was carried out under constant potential while simultaneously tracing the chronoamperogram. To improve deposition efficiency, a stainless-steel screw (M6, active electrode area 7.22 cm²) was used as the working electrode during electrorefining. A standard Ag/AgCl (5 mol% AgCl in LiCl-KCl) electrode served as the reference electrode and a graphite rod (Φ3) was used as the auxiliary electrode in all the experiments. All the electrochemical experiments were performed on an Autolab PGSTAT N302 (Metrohm) electrochemical workstation controlled by NOVA (1.10 version) software. Transient techniques, such as CV and SWV, were used to examine the electrochemical behavior of Ni²⁺/Ni⁰ in LiCl-KCl molten salt.

2.3 Sample analysis

The obtained metal deposits were dissolved in nitric acid and the concentration of nickel was determined in the solutions by ICP-AES (Optima8000, PerkinElmer). XRD (X’PERT POWDER, PANalytical, copper K_α source) and XRF were performed to characterize the deposited products. The morphology of the deposits was studied by SEM (Phenom ProX).

3.1 Electrochemical behavior of Ni²⁺ in LiCl-KCl molten salt

There are two main steps in the electrorefining of a crude material in a molten salt: The dissolution of the anodic crude material under a certain positive current, and the deposition of dissolved metal ion on the cathode. Impurities with oxidation potentials (E_diss ^impurity) more positive than that of nickel (E_diss^Ni) remain on the anode. If E_diss ^impurity is more negative than E_diss^Ni, dissolved impurities will remain in the electrolyte and high purity nickel can be deposited on the cathode [27,28]. Although the electrochemical behavior of chromium (major impurity in the alloy) in molten chloride has been reported [29], there is little information on the electrochemistry of nickel in this system. Therefore, we first investigated the electrochemical behavior of Ni²⁺ in LiCl-KCl to determine both the reduction potential of nickel and the number of electrons exchanged in the reduction process.

Fig. 1 shows the cyclic voltammogram of molten LiCl-KCl containing 0.58 wt% NiCl₂ at 723 K. A clear negative current peak at -0.20 V is seen with the corresponding positive current peak appearing at 0.0 V. Since no peak is observed in the cyclic voltammogram of pure LiCl-KCl (used as a blank) in the voltage range of -1.2–0.4 V, we infer that both the peaks should arise from the redox reactions of nickel in LiCl-KCl. A series of cyclic voltammograms were obtained by varying the scan rate from 0.05 to 0.25 V/s. As shown in Fig. 2a, the cathodic peak slightly shifts in the negative direction with increasing scan rate. The mid-peak potential (E_mp, defined as half of the sum of cathodic and anodic peak potentials) remains nearly independent of the scan rate as shown Fig. 2b, which further confirms that the electrochemical process is reversible [30]. For a simple reversible reaction, the half-wave width, W_1/2, obtained in a square wave voltammogram depends on the number of electrons exchanged and the temperature, as given by equation 1 [31].

Fig. 1Full-screen View

Cyclic voltammogram of LiCl-KCl-NiCl₂ (0.58 wt%) at the scanning rate of 0.1 V/s at 723 K. Working electrode: tungsten wire (S=0.32 cm²); reference electrode: Ag/AgCl; auxiliary electrode: graphite rod (Φ3)

Fig. 2Full-screen View

(a) Cyclic voltammograms of LiCl-KCl-NiCl₂ (0.58 wt%) at different scanning rates, (b) plot of the relation between mid-peak potential and scanning rate, (c) square wave voltammogram with a frequency of 20 Hz; working electrode: tungsten wire, (S=0.32 cm²), reference electrode: Ag/AgCl; auxiliary electrode: graphite rod (Φ3)

where R is the ideal gas constant (8.314), T is the temperature in K, n is the number of electrons exchanged during the electrochemical reaction on the electrode, and F is the Faraday constant (96485). From Eq. 1 and obtaining W_1/2 from Fig. 2c, the number of electrons exchanged is calculated to be 1.8, suggesting that Ni²⁺ is directly reduced to metallic Ni in LiCl-KCl by a two-electron process and the redox peaks at -0.2 and 0.0 V arise from the redox couple Ni²⁺/Ni.

3.2 Electrorefining of nickel from nickel-chromium alloy

Cotarta et al. have studied the electrochemistry of chromium in LiCl-KCl molten salt and measured the redox potential of chromium [29]. Cr(III) can be reduced to Cr metal stepwise through two electrochemical reactions mediated by Cr(II) at 0.1 and -0.9 V (vs. Ag/AgCl). Based on the electrochemical behavior of nickel and chromium, it is noted that both metals can be electrochemically dissolved in LiCl-KCl melt as Ni²⁺ and Cr³⁺ ions at 0 and 0.2 V. According to the CV results in Fig. 1, the reduction potential of Ni²⁺ is located at -0.2 V (vs. Ag/AgCl); the redox reactions involved in the complete process are listed as follows:

Anodic reaction:

Cathodic reaction:

It is seen from the above equations that the difference between the deposition potentials of Ni²⁺ and Cr²⁺ is 0.7 V. According to reports by Li et al. [32-34], the theoretical potential difference between the apparent standard potentials of nickel and chromium depends on the recovery rate η, the number of exchanged electrons, and the initial concentration of the two metal ions. Supposing η_Ni=99.999% and η_Cr=0.01%, a difference > 0.36 V in the deposition potentials of nickel and chromium is sufficient to achieve selective deposition at 723 K. Therefore, it is possible to obtain high purity Ni from a mixture of the two metals in molten salt if a suitable potential is applied. Since the Ni-Cr alloy dissolves in the melt during electrorefining, the deposition of metallic nickel occurs on the cathode while chromium remains in the electrolyte as Cr²⁺ which cannot be reduced at -0.2 V. Hence, we have used the constant potential technique during the electrorefining process.

To achieve further insight into the process of nickel deposition on the cathode during electrorefining, the chronoamperogram was recorded as shown in Fig. 3. The current associated with nickel deposition starts at -200 mA, and decreases sharply to -135 mA in 10 s, which we attribute to the slow formation of the first nickel nuclei on the steel cathode [35,36]. In the next 3590 s, the reduction current rises from -135 to -630 mA as a result of the increase in active surface area of the working electrode as more nickel is electrochemically deposited on the substrate. The current remains at about -630 mA for 1 h until the end of the electrorefining process. The Ni-Cr alloy immersed in LiCl-KCl melt becomes thinner after electrorefining (Fig. 4a) with a mass loss of 0.61 g. On the cathode, the products of electrorefining are mixed with chloride salts that adhere to the electrode. After washing the chlorides with deionized water, dendritic deposits are obtained, as illustrated in Fig. 4b.

Fig. 3Full-screen View

Chronoamperogram during electrorefining under constant potential (-0.2 V) at 723 K in LiCl-KCl-NiCl₂ (0.58 wt%) melt. Working electrode: Stain-less steel screw (M6, active electrode area 7.22 cm²)

Fig. 4Full-screen View

Photos of a) anodic Ni-Cr alloy after electrorefining; b) deposits on the cathode; c) SEM image of the deposits

SEM image of the cathodic deposits shows the presence of rod-like structures with holes, and chips (Fig. 4c). The microstructure of the deposits formed in molten salt is completely different from the island-like surface morphology with cracked patterns formed in aqueous solutions [37]. The rods are several millimeters long and ~80 μm wide, and the total weight of the collected products is 0.46 g. Thus, we find that almost all the dissolved nickel can be recovered, in accordance with equation 2; using equation 3, the current efficiency can be calculated to be 71.27%. According to the cathodic reaction sequence, Cr³⁺ should be reduced to Cr²⁺ prior to the reduction of Ni²⁺ to Ni⁰ at -0.2 V, which means that a fraction of the current passing through the cathode is consumed by the Cr³⁺→Cr²⁺ reaction. This is similar to the process by which, plutonium is separated from lanthanum by electrolysis in molten LiCl-KCl salt [38]. If it were not for the Cr³⁺→Cr²⁺ reaction, the current efficiency of Ni electrorefining from Ni-Cr alloy would be 95.58%. As shown in Table 1,

where m_Ni^deposit and m_Ni^alloy are the masses of nickel in the deposit and in the alloy, and m_Ni^theroy is the calculated mass of nickel corresponding to the current passing through anode.

XRF, XRD and ICP-AES measurements were carried out to determine the composition and purity of the deposits. In the XRF spectrum of the Ni-Cr alloy, peaks due to Ni and Cr are observed at 5.988 and 8.331 keV (Fig. 5). The first peak disappears in the spectrum of deposited products, suggesting that Ni is deposited on the cathode, whereas chromium remains dissolved in the molten salt. In addition, the formation of metallic nickel on the cathode is confirmed from XRD results where only reflections originating from metallic Ni are observed (Fig. 6). The purity of nickel in the deposits was determined to be higher than 99.83% by ICP-AES; the amounts of the other detected impurities are listed in Table 1. It is clear that the Cr content in the deposits decreases by three orders of magnitude compared to that in the starting Ni-Cr alloy and the impurity content in the deposits is significantly reduced upon electrorefining; thus, our results are better than those reported for Ni electrorefining CaCl₂ solution [39].

Fig. 5Full-screen View

XRF spectra of the Ni-Cr alloy and deposits on the cathode

Fig. 6Full-screen View

XRD spectra of the Ni-Cr alloy and deposits on the cathode

3.3 Influence of the nickel ion concentration in LiCl-KCl on electrorefining

The concentration of metal ions in the electrolyte has a significant influence on several aspects of the electrorefining process, such as deposition rate and product size [40,41]. According to our experimental results, at the same potential, the deposition rate of nickel increases as the concentration of nickel ions is raised (Table 2). At high nickel concentration, Ni²⁺ ions are rapidly reduced on the cathode, during which gaseous Cl₂ is formed simultaneously on the anode; similar phenomena have been observed in other molten salt systems [42]. Chlorine gas bubbles formed on the surface of the nickel anode can accumulate at gas-liquid interface [43] and the middle part of the nickel anode at the interface is corroded (Fig. 7) due to the formation of soluble NiCl₂ by the reaction of nickel with Cl₂. Therefore, at high nickel concentrations, it is possible for the anodic nickel rod to fall into the electrolyte during electrorefining due to corrosion of the electrode, which obviously hinders the whole electrorefining process. At the low concentration (that was employed in our experiment), the nickel anode dissolves gradually from the bottom to the top and no obvious corrosion is observed (Fig. 4a). Our results thus reveal that a relatively low concentration of NiCl₂ in the molten electrolyte is favorable for the optimized deposition of Ni in the electrorefining process.

Fig. 7Full-screen View

Photo of the anodic nickel rod after electrorefining at high NiCl₂ concentration (2.23 wt%)