1. Introduction

As part of global greenhouse gas reduction policies, the development of carbon-neutral processes for steel and non-ferrous metals has rapidly increased in recent years [1, 2]. As of 2014, greenhouse gas emissions during the metal smelting process were 159.45 million tons of CO₂Eq, accounting for 15% of total industrial greenhouse gas emissions in Korea [2]. Therefore, carbon neutralization in the metal smelting industry is necessary to realize carbon neutrality and reduce the greenhouse gases formed during power generation. In case of steel, smelting technology based on the hydrogen reduction of iron is being considered, especially for non-ferrous metals that cannot be thermodynamically reduced. For example, for lightweight metals, including aluminum and magnesium, and rare/rare earth metals, such as Ti, Zr, and lanthanide metals, only conceptual research and no technological development at the commercial production level has been carried out.

Pyroprocessing recovers metallic uranium from spent nuclear fuel using LiCl-KCl molten salt. When recycling PWR (Pressurized Water Reactor)-spent nuclear fuel, uranium and TRU (Transuranium elements) oxide are reduced to metal at the cathode using LiCl molten salt. When an insoluble anode of platinum is used, oxygen ions move to the anode and are captured as oxygen gas. Pyroprocessing is classified as an electroreduction process, corresponding to a metal smelting process in general metal industries, an electrorefining process for high purity, and an electrowinning process, which corresponds to electrolytic extraction in the copper and zinc industries. Pyroprocessing is known as a highly nuclear proliferation-resistant process, which makes it difficult to separate TRU materials, but it is also a very eco-friendly process as it does not leave a carbon footprint [3-5]. Therefore, when pyroprocessing is applied to the rare metal industry, it can be an eco-friendly process that generates O₂ instead of existing pollutants, such as CO_x and Cl₂. The pyroprocessing of PWR-spent nuclear fuel can be an electrochemical or chemical metal thermal reduction of oxides, such as the FFC (Fray-Farthing-Chen) and OS (Ono-Suzuki) processes developed to recover metals from metal oxide raw materials. If there are any physicochemical considerations, they can easily be applied to high-melting-point rare metals such as Ti, Zr, and Hf [6-9]. However, in the FFC, OS, and EMR (Electrolytic Metal Recovery) processes, titanium oxide is reduced to metal at the cathode, and oxygen and carbon react at the anode to generate CO₂. In addition, all these processes have low current efficiency, and by using finely divided raw materials, a metal phase with a high specific surface area is recovered. Thus, the oxygen content is higher than the industry standard, and there are limitations in that the production cost is higher than that of commercial processes [7, 10-12]. Moreover, metal oxides can be directly reduced using a solid oxide membrane (SOM) ceramic anode for conversion to an eco-friendly process, which was not commercially available owing to the chronic problem of ceramic anode thermal shock [11-13]. As such, there is currently no commercial Ti smelting process to replace the Kroll process [14, 15], which was invented in the 1940s. However, as the Kroll process is a typical energy-intensive and pollutive smelting process that generates CO_x and Cl₂, research is ongoing to circumvent these drawbacks.

Some important problems must be addressed before applying pyroprocessing to the development of carbonneutral smelting technology for rare metals: 1) reducing impurities, especially oxygen, in the final metal, 2) developing an inert anode to avoid CO₂ generation, and 3) increasing production throughput and process efficiency for mass production. As one of the measures for improving the productivity of refractory metals—such as Ti, Zr, and Hf—and reducing oxygen impurities, this study introduces the thermodynamic basis for the liquid copper-assisted electrolysis (LCE) process and the current research direction for carbon neutralization.

2. Experimental

2.1 Materials and Equipment

The salt system constituents were CaF₂ (purity 99.5%), CaCl₂ (purity 98%), and CaO (purity 99%). All the chemicals were purchased from Samchun Chemical Co., Ltd. (Seoul, South Korea). YSZ-8 tubes (inner dia. 9 mm, outer dia. 13 mm, height 600 mm) for the SOM anode material were obtained from Nikkato Corporation (Tokyo, Japan). A Kanthal wire with a diameter of 1 mm and height of 650 mm was used as an anode current collector. A Cu rod (dia. 8 mm, purity 99.9%), Cu wire (dia. 2 mm, purity 99.9%), and tungsten wire (dia. 1 mm, purity 99.9%), obtained from Alfa Aesar (Ward Hill, MA, United States of America) and Sigma Aldrich (Burlington, MA, United States of America), were used as electrodes. Cu chips, Ca granules, and Ag granules (purity 99.9%) were obtained from Junsei Chemical Co., Ltd. (Tokyo, Japan). A tungsten crucible (purity 99.95%) was obtained from Zhengzhou Shibo Nonferrous Metals Products Co., Ltd. (Henan, China), and an Al₂O₃ crucible and tubes were obtained from Mesto (South Korea).

Every experiment was performed in a glove box with a vertical tubular reactor constructed to prevent oxidation of the molten salt components and structural materials. The glove box was operated in an argon atmosphere, where the concentration of oxygen and moisture was controlled to be less than 2 ppm. Before the experiment, all salts were placed in a tungsten crucible and heated to 300℃ for 12 h to remove all moisture. During the experiment, the molten salt was heated to 1,050℃ at a rate of 1℃∙min⁻¹ in an argon atmosphere. The assembled cathode and anode were placed in a vertical tubular reactor within an electrical furnace. Electrochemical measurements and electrolysis were performed using Autolab PGSTAT302N and NOVA computer software.

2.2 Hot Corrosion Test

A hot corrosion test was performed to understand the corrosion behavior based on the grain size of the SOM anode. The YSZ-8 pellets were prepared using CIP (Type 1011-8-16-30, Loomis, Products Company, USA) at 2,500, 3,000, and 5,000 psi pressures. Each specimen was sintered for 3, 6, 8, and 11 h at 1,500°C and 11 hours at 1,600°C. The prepared YSZ-8 sintered bodies with different grain sizes were immersed in CaCl₂-CaF₂-CaO molten salt at 1,150°C for 24, 72, and 120 h to evaluate the temporal reactivity of the molten salt.

2.3 Electrochemical Procedures

2.3.1 Electroreduction and Metallothermic Reduction

The electroreduction behavior of Calcium ions in the molten salt was evaluated by chronopotentiometry (CP) using a Cu cathode and YSZ oxygen-conducting ceramic membrane as a waste salt recycling process. The cathode and anode potentials were monitored using a tungsten wire as the pseudo-reference electrode. The electrolytic system was maintained at 1,050℃ throughout the experiment. During the tests, CaCl₂ and CaO were melted in a Ti crucible with an inner diameter of 100 mm, placed in a stainlesssteel vessel, and heated externally using an electric furnace. Thus, the prepared CaCu alloy was used to reduce the metal oxide feedstock such as TiO₂, ZrO₂ and HfO₂. During this process, CaCl₂ in a Ti crucible with a CaCu alloy laid on the bottom was heated to 1,050℃, and oxide feedstocks were introduced.

2.3.2 Electrorefining

The Ti, zirconium subchloride, and Ba₂HfF₈ molten salts were prepared by the PbCl₂ reaction or BaF₂ precipitation method reported in previous studies [16, 17]. Electrorefining Ti, Zr, and Hf from the Cu(Ti, Zr, Hf) ingots from metallothermic reduction via CaCu was performed using CP. In this experiment, Cu(Ti, Zr, Hf) ingots, a Cu plate, and a tungsten wire electrode were used as the anode, cathode, and reference electrodes, respectively.

2.3.3 Process Monitoring System

Electroreduction was conducted using a power supply unit (ODA Technologies Co., Ltd., Santa Clara, CA, United States of America, EX20-30), and the applied potential was recorded using a data logger (Agilent Technologies, Inc., Santa Clara, CA, United States of America, 34970A, BenchLink Data Logger 2). CP was performed using a potentiostat (Metrohm Autolab, Utrecht, Netherlands, PGSTAT302N, Nova ver. 1.10) and data logger.

2.3.4 Post-electrorefining Treatment

The electrodeposited Zr and Ti recovered from the chloride molten salt were washed with 1mol% HCl solution, while the electrodeposited Hf was ground in a glove box under an argon gas atmosphere. The molten salt was effectively removed via salt distillation for 24 h at 1,300℃ under vacuum (10⁻² torr). The metal powders were recovered, and Ti, Zr, and Hf buttons were prepared by arc melting under vacuum (10⁻⁵ torr).

2.3.5 Material Characterization

The morphology and chemical composition of the final products were characterized using a field emission scanning electron microscope (FE-SEM, JEOL Ltd., Tokyo, Japan, JSM-7000F) equipped with an energy-dispersive X-ray spectroscopy (EDS) system. The oxygen concentrations were analyzed by ONH & CS Determinator (Determinator- ONH, ELTRA, Germany, ONH-2000). X-ray diffraction (XRD, Rigaku, Tokyo, Japan, D/MAX-2200 Ultima/PC) was used to analyze the structural phase evolution of the final products. After dry polishing in a glove box, the specimen was analyzed in an airtight Ar-filled chamber to prevent oxidation. The following conditions were applied in the analyses: FE-SEM/EDS was conducted using an applied voltage and vacuum pressure of 15 kV and 7.04 × 10⁻⁴ Pa, respectively, ONH & CS Determinator was conducted under 3,000°C with an ElectrodeImpulse furnace (maximum 8 kW, vertical alignment) and the XRD scan rate, step, and half-width range (2θ) were 3.5°/min, 0.05°, and 10–90°, respectively. The TiO₂ used in the experiment was SRL-lGR-V5 from Iluka Resources (detailed composition is listed in Table 1). HfO₂ (purity 99.9%, particle size 0.05–0.1 μm) was obtained from Ansto Minerals (Australia). ZrO₂ was supplied by Alkane Resources (Australia); the detailed composition is provided in Table 2.

Table 1

Chemical composition of TiO₂ (wt%)

Table 2

Impurity composition of ZrO₂ (ppm)

2.4 High-Purity Titanium Production by LCE Process

The LCE process was developed as an environmentfriendly smelting process for rare metals (For example, Me = Ti, Zr, or Hf) to replace the Kroll process [18]. This study mainly deals with the reduction of Ti, and the core of this process is the reduction of the oxygen concentration by recovering the Ti alloy in liquid form. As shown in Fig. 1, this process comprises three steps: electroreduction of CaO, metallothermic reduction, and electrorefining. Electroreduction is a process used for producing the CaCu reducing agent for rutile (TiO₂) reduction. In this step, CaO dissolved in the CaCl₂ waste molten salt is electrolytically reduced using an inert anode to produce oxygen gas at the anode and metallic Ca at the cathode as well as to reuse salt. When Cu is used as the cathode material, the cathode alloys with Cu along with Ca electrodeposition. As a result, a CaCu alloy with a relatively low melting point compared to the process temperature is produced. In the metallothermic reduction step, CuTi is produced through the reaction of the produced CaCu alloy with TiO₂ (Ca₂Cu + TiO₂ = CuTi + 2CaO). At this time, CaO, which is dissolved in the molten salt, is produced as a by-product of the reaction The molten salt containing CaO produced in this manner is again used for producing CaCu in electroreduction as a waste salt recycling process. In the electrorefining step, the CuTi alloy produced in the previous step is recovered as electrodeposited Ti. The Cu remaining after Ti is recovered is reused as the Cu cathode in the first step. After removing the residual molten salt with 1mol% HCl solution, the electrodeposited recovered through the arc melting process to finally produce Ti metal with a purity of 99.9% or greater.

Fig. 1

Schematic diagram of the LCE process. (a) Eletroreduction, (b) Matallothermic Reduction, (c) Electrorefining.

3. Results and Discussion

3.1 Development of Direct Electroreduction by Inert Anode

Currently, a consumable graphite anode is used for the direct reduction of metal oxides, and CO₂ gas is continuously generated as a reaction by-product during reduction. This process deviates from the current carbon-neutral policy stance and requires replacement. For this reason, the development of inert anodes has recently been in the spotlight, and research on various anode materials, such as ceramics, alloys, and cermets, is underway [19-22]. As inert anodes must satisfy various conditions, such as a low corrosion rate (degradation rate), high electrical conductivity, resistance to thermal shock, and mechanical strength, there are numerous difficulties in their development, which must be addressed in the future.

As shown in Fig. 2, our laboratory focuses on securing long-term durability in inert anodes. This is achieved by reducing reactivity with the molten salt through grain size control in the case of SOM [23] and by improving the electrochemical durability by forming a protective oxide layer for cermet anodes. Initially, an electrolytic reduction experiment was performed using a SOM to evaluate its oxygen-ion conductivity. Notably, a decrease in oxygen ion conductivity due to the reaction with the molten salt is observed. Therefore, first, the effect of grain size on the reactivity with molten salt was confirmed through a hot corrosion test. The grain boundary acts as the preferred reaction pathway upon exposure to molten salt because of its high energy level. In the case of ‘SOM with a larger grain size (Fig. 2(a)),’ the reaction rate of the molten salt decreased because the penetration path of the molten salt was limited owing to the small grain boundary area.

Fig. 2

Schematic diagram of the degradation behavior of YSZ oxygen ion-conducting ceramic membrane in molten salt according to larger grain (a), smaller grain (b) [22].

Conversely, in the case of ‘SOM with a smaller grain size (Fig. 2(b))’, the molten salt penetration path increases because of the larger grain boundary area, indicating a fast reaction rate. This means that the stability of the positive electrode can be secured by controlling the grain size and changing the content of phase stabilizers such as Y₂O₃ and MgO. In the next step, an electrolytic reduction experiment was performed using SOM with oxygen ion conductivity.

Fig. 3 shows the CV data of the Cu working electrode in the temperature range of 1,083–1,383 K in the CaCl₂- CaF₂-CaO molten salt. At the lowest temperature of 1,083 K, the reduction starts from −1.25 V, and the peak oxidation potential was observed at approximately −1.2 V. It is noteworthy that as the temperature was increased from 1,133 to 1,183 K, the reduction initiation potential remained almost the same, whereas the peak oxidation potential shifted in the positive direction to −1.05 V and −0.8 V, respectively. This suggests that Ca is electrodeposited on the Cu working electrode to form a CuCa alloy. Hence, at a relatively low temperature, it was electrodeposited as a pure Ca phase, and alloy formation became dominant as the temperature increased. This caused a higher anodic potential during oxidation. At the process temperature of 1,383 K, the Ca reduction potential shifts from R₃: −1.25 V to R₂: −1.2 V. Since this temperature is near the melting point of the Cu cathode, alloying is accelerated and underpotential deposition occurred around R₁: −0.6 V. Therefore, the intensity of the Ca oxidation peak was reduced, and the oxidation peak of CaCu was observed at O₁. As shown in Fig. 4(a), a high current was applied to ensure sufficient oxygen ion conductivity owing to the SOM. At this time, the anode potential was confirmed to be approximately 3 V, and the cathode potential was approximately −2 V. The amount of generated oxygen gradually increased and reached a plateau at 0.55% after 35 min. This suggests that oxygen ion conduction through the SOM and the oxygen oxidation reaction in the liquid Ag inside the SOM are maintained at a steady state. In addition, the intermittent anode and cathode potential drop can be explained through the formation of the CaCu liquid alloy, which is schematically presented in Fig. 4(b). As shown in Fig. 4(b), an alloying reaction occurs from the surface as Ca metal is electrodeposited on the Cu cathode during the electrolytic reduction process. A low-melting-point CaCu alloy is formed and melted. A cone shape rupture of the Cu electrode in Fig. 4(b) confirms melting rupture.

Fig. 3

CV results using Cu working electrode system at various temperatures in CaCl₂-CaF₂-CaO molten salt.

Fig. 4

O₂ gas concentration in off-gas and anode/cathode potentials during CaO electroreduction using a copper rod cathode (a) and a schematic showing the occurrence of an intermittent voltage drop (b).

Fig. 5 shows the XRD and SEM/EDS analyses results of the CaCu alloy produced after electrolytic reduction. The generated CaCu was polished in a glove box and then XRD analysis was performed in an airtight holder filled with Ar. All molten salts on the surface were removed during the polishing process. The XRD analysis was used to observe the CaCu peaks after removing the molten salts. SEM/EDS analyses show the oxidation of some Ca, the CaO phase, and a trace amount of CaCl₂ molten salt. The Ca content in CaCu was confirmed to be approximately 40at% or higher.

Fig. 5

XRD pattern (a), cross-sectional SEM image (b), and EDS analysis results (c) of the CaCu alloy reduced through SOM inert anode electroreduction.

3.2 Metallothermic Production of CuTi

The CuTi alloy was prepared by reducing natural TiO₂ using CaCu prepared in the electroreduction step (Step 1 in Fig. 1) of the LCE process. First, after melting CaCl₂ in a Ti crucible, CaCu was charged before TiO₂ to prepare a CuTi alloy through the reaction shown in Eq. (1).

As it is difficult to obtain thermodynamic data on intermetallic compounds, such as CaCu and CuTi, from traditional thermodynamic databases, the formation energy change (ΔE_f) was calculated using the formation energy supplied from DFT analysis [26].

For the formation energy in Eq. (1), the forward reaction prevails in the standard state. The equation for calculating the Gibbs free energy change under the actual thermal reduction reaction conditions can be expressed as Eq. (2). The formation energy in Eq. (1) was obtained using the internal energy differences of the elements for the solid system; therefore, it can be regarded as standard Gibbs free energy and calculated as ΔG⁰ = 302.54 kJ·mol⁻¹ by multiplying Avogadro’s number (6.02 × 10²³/mol) and 1.6 × 10⁻²2 kJ·eV⁻¹.

In Eq. (2), a_CaO is the activity of CaO in the CaCl₂ molten salt, a_CuTi is the activity of the CuTi product, a_CaCu is the activity of the CaCu reducing agent, and a T i O 2 is the activity of the TiO₂ feedstock. Except for CaO, all intermetallic compounds showed very low solubility in CaCl₂; thus, the activity can be regarded as unity. For TiO₂, the activity changed while reacting with CaO in the molten salt to form CaTiO₃; therefore, the Gibbs free energy change is a function related to the CaO and TiO₂ activity in CaCl₂.

Fig. 6 shows the calculated Gibbs free energy change according to the CaO and TiO₂ activity in the CaCl₂ molten salt. In most TiO₂ activity ranges, the Gibbs free energy has a negative value regardless of CaO activity, indicating that the forward reaction is dominant. However, when the TiO₂ activity converges to 0 (5×10⁻¹² in this calculation), its value ranges from the lowest −854.2 kJ·mol⁻¹ to 4.2 kJ·mol⁻¹ depending on the CaO activity; therefore, excessively high activity of CaO alters the reaction direction. TiO₂ added to CaCl₂ molten salt with dissolved CaO is known to form CaTiO₃. The exact value is unknown, but TiO₂ activity can be expected to be maintained at a very low level in this reaction system.

Fig. 6

Calculated Gibbs free energy (kJ·mol⁻¹) change according to the activity of CaO and TiO₂ in CaCl₂ molten salt.

For producing CuTi, Gibbs free energy must remain negative; therefore, the region where CuTi can be generated regardless of the TiO₂ activity being limited to the region where the CaO activity is less than approximately < 0.1. In particular, the CaO activity coefficient in CaCl₂ is known to be 3.371 [25, 26], which indicates a tendency to increase rapidly with increasing CaO concentration. Through this thermodynamic calculation, it can be seen that CaO activity is a decisive process variable in the metallothermic reduction process of TiO₂ using a CaCu reducing agent, and that CaO activity should be maintained as low as possible for a continuous reduction reaction.

Fig. 7 shows the cross-sectional SEM image and EDS analysis results for CuTi generated by CaCu metallothermic reduction. As shown in Fig. 7(a), when the concentration of CaO in the molten salt was 28.46mol% (a_CaO = 1), the CuTi alloy was not formed as a metal pool but locally formed in the molten salt, indicating that oxygen remained in the CuTi alloy and part of the titanium oxide remained in the unreduced form. Fig. 7(b) shows the EDS analysis results of CuTi after reducing the CaO concentration to 7.5mol% through salt exchange to decrease the CaO activity, wherein the final activity was maintained at a_CaO = 0.25. It was confirmed that when CuTi formed a metal pool at the bottom of the crucible, it separated into two phases with different concentrations, where the composition of each phase was confirmed as 29.11wt% Cu and 70.89wt% Ti for phase 1 and 42.63wt% Cu and 57.37wt% Ti for phase 2. It was found that when the CaO concentration was controlled to 7.5mol% or below, a CuTi alloy with low oxygen contamination was formed, as was thermodynamically confirmed.

Fig. 7

Cross-sectional SEM images and EDS analysis results of CuTi produced after the reaction of CaCu with TiO₂ at 1,050℃ in 28.46mol% CaO (a_CaO = 1) (a) and 7.5mol% CaO (a_CaO = 0.25) (b).

3.3 Electrorefining of Group IV Metals

Electrorefining (Fig. 1, Step 3) of the LCE process involves recovering high-purity Ti, Zr, and Hf by refining CuTi, CuZr, and CuHf prepared using a CaCu reducing agent. Currently, Ti and Zr are mostly studied using fluorine-based molten salts, but it is difficult to separate the electrodeposits from the molten salt due to low solubility and vapor pressure [27, 28]. In addition, a critical problem arises with (Zr, Hf, Ti)Cl₄ when the experiment is conducted in a chlorine-based molten salt because of the disproportionation reaction [29, 30]. In addition to the undesirable disproportionation reaction, a chlorine-based molten salt can easily separate the molten salt from the electrodeposited material, and its operating temperature is relatively lower than that of the fluoride molten salt. Subchlorides of Group IV metals, such as MeCl_x (Me = Zr, Ti and x = 1, 2, or 3), suppress the disproportionation reaction, and TiCl₂ is effective in recovering high-quality metallic deposits [16]. The CuMe alloy rods were fabricated by the CaCu metallothermic reduction following injection casting (Fig. 8(a)) and were used as an anode for electrorefining. Chlorine-based molten salts were used for Zr and Ti experiments to synthesize ZrCl₂ and TiCl₂, respectively, and Hf was tested in fluorine-based molten salts. As shown in Fig. 8(b), the CuMe alloy is composed of two phases, Cu-rich and metal-rich, and the oxygen concentrations in CuMe are 245 ppm, < 1 ppm, and 328 ppm for CuZr, CuHf, and CuTi, respectively. The oxygen concentrations were analysed by Eltra ONH-2000. Fig. 8(c) shows a graph comparing the results of the actual dissolution distance with the calculated distance of when Zr dissolved and escaped from the anode surface after 20 h in the range of possible diffusion coefficients. As a result of the calculation, the largest diffusion coefficient showed a dissolution distance of approximately 1 mm for 20 h when only simple diffusion was considered. However, from the experimental observations, 4 mm was dissolved in 2.5 h, and the dissolution distance after 13 h reached 9 mm. This fast anodic dissolution rate is possibly attributed to the Zr-rich island phase (Fig. 8(b)) dissolving, leaving pores, and contributing to the migration of metal ions through this space. This result was confirmed by the formation of sponge-like Cu when the microstructure of the final anode residue after anodic dissolution was completed, as shown in Fig. 8(d) [17].

Fig. 8

CuMe anode fabricated by CaCu metallothermic reduction (a), SEM microstructure with EDS composition analysis (b), comparison of anodic dissolution distance between experimental and calculated data (c), and cross-sectional macrostructure change with electrorefining time (d) [18].

Fig. 9 shows metal electrodeposition through electrorefining of the CuMe alloy. According to the EDS analysis, no impurities were detected within the detection limits. For Ti and Hf, it was confirmed that the average particle size was greater than 100 μm, while the dendritic Zr size was 10 μm. For Group IV metals, an oxide film of approximately 10 nm is formed when exposed to the atmosphere. Therefore, when the particle size decreased, the specific surface area and oxygen concentration increased. An oxygen concentration calculation method based on particle size has already been previously reported [31].

Fig. 9

SEM images and EDS analysis results of the Ti, Zr, and Hf electrodeposits.

Fig. 10 shows the oxygen concentration calculation results according to the particle size. Comparing the industrystandard oxygen concentration limits for the target metals in this study, Ti, Zr, and Hf must be greater than 15 μm, 12 μm, and 70 μm, respectively.

Fig. 10

Oxygen concentration as a function of metal deposit particle size and comparison with industry-standard oxygen concentration limits.

4. Conclusions

The extensive application of pyroprocessing technology to the general metal industry and the development of economical inert anode materials, such as high-priced noble metal anode materials that are stable in molten salt systems, were investigated. Research on various inert anode materials, such as SOM and cermet, has been conducted to develop the LCE process, which is similar to pyroprocessing in that it does not generate greenhouse gases and is an ecofriendly process concept. A CaCu alloy was produced by CaO electroreduction using an inert anode selected according to the research results, and a CuMe alloy was produced for electrorefining by reducing natural TiO₂ with the produced CaCu.

CuTi was produced by reacting the prepared CaCu with TiO₂ in a Ti crucible. When the concentration of CaO was controlled at 7.5mol%, CuTi was normally produced. During the CaCu manufacturing experiment as used-salt recycling, the Cu cathode was consumed due to the generation of CaCu in the cathode, and CaCu was prepared with a Ca-to-Cu composition ratio of 46.3 to 53.7. Finally, high-purity Ti was produced by electrorefining the produced CuTi alloy.

Electrorefining experiments were performed using the CuTi, CuZr, and CuHf anodes prepared using CaCu. For Ti and Zr, a chlorine-based molten salt was used and highquality metal deposits were recovered, while high-purity Hf deposits were recovered using a fluorine-based molten salt. A pilot-scale test was conducted on Ti, which showed optimal results on a laboratory scale, satisfying the ASTM B299 standard.

The results confirmed that the liquid-phase reduction process can effectively increase the productivity and reduction rate and can be used as a reference for the oxide-spent nuclear fuel recycling process.