1. Introduction

Pyroprocessing is a highly non-proliferate technology that can be used to reduce the amount of high-level waste and enhance the utilization of energy sources by separating highly radioactive fission products and recycling uranium (U) and transuranic (TRU) elements from used nuclear fuel (UNF) using high-temperature molten salt electrochemistry [1, 2]. In head-end pyroprocessing, cladding hulls and hardware are supposedly generated as metal wastes for the treatment of pressurized light-water reactors (PWRs). Among the metal wastes, cladding hull wastes are composed of Zr-based alloys and is produced at the rate of ~2.6 tons per 10 tons of UNF [3]. Particularly, since trace amount of UNF may remain inside the cladding tubes after decladding, the cladding hull waste is assumed to be an intermediatelevel waste (ILW) in the domestic regulation for radioactive wastes or a “greater than class C” (GTCC) waste, based on the U.S. regulations [4, 5]. Therefore, recycling Zr, which is the main component of cladding hull, is substantially advantageous for the reduction of metal waste.

Representative Zr recovery processes include a chlorination reaction using Cl₂ gas [6] and electrorefining using molten salt electrochemistry [3, 7]. The electrorefining process involves the deposition of highly pure Zr on cathode by applying a current or voltage between the anode basket (filled with cladding wastes) and cathode in high-temperature molten salt electrolytes. Although the chlorination reaction effectively processes a large quantity of Zr-based cladding hull wastes due to its fast reaction kinetics, toxic Cl₂ gas is used, and an additional Kroll process is required to covert ZrCl₄ to metallic Zr for further applications. On the other hand, the electrorefining process is advantageous for directly recovering Zr in a metallic form. However, the electrochemical reaction mechanism of Zr is complicated and the amount of residual salt in the Zr deposit is considerably high because Zr is recovered as a powder-type deposit in chloride-based molten salts. To resolve this issue, many researchers have studied fluoride-based molten salts like FLiNaK for Zr recovery and acquired coherent Zr deposits [8-10]. However, fluoride-based salts require higher operation temperatures ranging from 700℃ to 900℃ due to their higher melting points and their applications are limited to materials such as crucibles, electrodes, etc.

In a previous study, we examined a chloride-fluoride mixed salt for overcoming the disadvantages of all-chloride and all-fluoride salts [11]. Cyclic voltammograms showed that Zr complex ions of various oxidation states are stabilized as Zr⁴⁺ in LiCl-KCl-ZrCl₄ salts by the addition of the fluoride compound, LiF, which strongly supports the substitution of complexing ions. In addition, through the electrorefining process for Zirlo tubes, better-quality dendrite-type Zr deposit was grown with significant reduction in residual salt content in the mixed chloridefluoride salt. There has been limited research on the effect of additives on the electrodeposition of Zr for improving its morphology in molten salts. G. W. Mellors et al. examined purified ZrO₂, SiO₂, K₂SiF₆, and KBF₄ in chloride-fluoride mixed salt baths or all-fluoride salt baths and found that the addition of KBF₄ and Al₂O₃ produced Zr alloys as smooth coherent plates [10].

Herein we investigate the effect of AlF₃ addition on the morphological feature of Zr deposits during Zr electrorefining using chloride-fluoride mixed salts. Particularly, a consecutive process has been suggested for the enhancement of recovery yield of Zr.

2. Experimental

The base electrolyte used herein for Zr electrorefining was a chloride-fluoride mixed salt, consisting of anhydrous LiCl-KCl eutectic (99.99wt% purity, Sigma-Aldrich), 0.84 M LiF, and 0.07 M ZrCl₄ as an initiator. The effect of adding AlF₃ over the concentration range 0~38.6 mM in the mixed salt on Zr electrorefining was investigated. All the experiments were performed in an Ar-purging glove box where oxygen and moisture are maintained under 5 ppm. Cyclic voltammetry and chronoamperometry were performed using a potentiostat/galvanostat (BioLogics Model SP-150), which has a three-electrode cell including a 30 mm-long tungsten wire (1Ø) shielded within an alumina tube (3Ø) as the working electrode, a Zr rod or a half-cut Zirlo tube as the counter electrode, and a Ag/AgCl reference electrode to examine the effect of AlF₃ addition in LiCl-KCl-LiF-ZrCl₄ salt. The reference electrode was made by inserting a Ag wire in a 6Ø mullite tube partially filled with LiCl-KCl-1wt% AgCl. The operating temperature was fixed at 600℃ and the experiments were performed under a quiescent condition.

Quantitative chemical analysis of the Zr deposit was performed using inductively coupled plasm-atomic emission spectroscopy (ICP-AES) and ion chromatography (IC). Field-emission scanning electron microscopy (FESEM, Hitachi SU8010) and energy-dispersive X-ray analysis (EDX) were used to examine the microstructure of the electrorefined Zr deposit surface and its compositional distribution, respectively. X-ray diffraction (XRD, Bruker D8 Advance, Cu Kα = 1.5418 Å) was used to examine the crystallographic texture of the Zr deposit in the presence of AlF₃. In addition, X-ray photoelectron spectroscopy (XPS, Kratos AXIS Nova) was employed for chemical analysis of the electrorefined Zr surface. Monochromated Al Kα (1486.69 eV) was used to collect the spectra at an angle of 0° between the sample surface normal and the analyzer lens. A pass energy of 20 eV and step size of 0.1 eV were used for the high-resolution region scans. The binding energy scale was referenced with respect to C 1s at 284.6 eV.

3. Results and Discussion

Figure 1a shows the cyclic voltammetric curves for Zr reduction/oxidation in the presence of various concentrations of AlF₃ LiCl-KCl-LiF-ZrCl₄ salt at 600℃. In the absence of AlF₃, unlike the chloride-based salt in which a multi-step reduction of Zr ions is involved [7, 12, 13], a single-step reduction of Zr is evident at about −1.15 V for the chloride-fluoride mixed salt, attributed to a complexation of Zr ions with F ions as follows [14].

With increasing AlF₃ concentration, additional shoulder peaks in the cathodic scans become relevant from about −1.26 V for 3.8 mM AlF₃ and shift slightly to −1.22 V and −1.2 V for 15.4 mM and 38.6 mM AlF₃, respectively, which should be associated with the reduction of Zr-Al alloys. For higher concentration of AlF₃, the peak potential approaches the reduction peak of Al (Fig. 1b). The formation of Zr-Al can also be evidenced from the peak shift for the oxidation of Zr at about −0.84 V and the appearance of additional peaks at a less negative potential regime between −0.75 V and −0.65 V. The Zr-Al alloy is known as a line compound formed in a specific chemical ratio according to the concentration of each component at a certain temperature [15]. Therefore, the peak shifts for the reduction and oxidation of Zr and Al, depending on the concentration of AlF₃, can be analogized as the formation of Zr-Al alloys of different stoichiometries.

Zr electrorefining was performed at a constant potential of −1.2 V in the presence of various AlF₃ concentrations to examine the morphologies of Zr deposits (Fig. 2). To exclude the effect of area-dependent growth of Zr at a constant potential, the processing time was controlled by the same amount of electric charge (−540 A·s). For comparison, Zr deposit electrorefined from a Zirlo tube without AlF₃ in a previous study is displayed in Fig. 2a [11]. Although the total charge was less than a half of that for the AlF₃-free salt, a plate-like Zr deposit with a diameter of ~20 mm is evident in the presence of AlF₃, formed at the salt interface with a limited amount of deposit on the rest of the W rod.

Dependency of the growth mode was examined by applying a constant potential and current, as shown in Fig. 3. The morphologies of the Zr deposit for 15.4 mM and 38.6 mM AlF₃ were compared and the total electric charge passed during the electrorefining process was fixed at −540 A·s for both modes. For electrorefining at a constant current of −50 mA, thicker Zr deposits were formed on the W rods, with only small radial deposits at the salt interface due to a relatively homogenous deposition of Zr on the W rod. This is because the electrode potential is assumed to be become less negative towards the reduction of pure Zr, thereby forming relatively Zr-rich Zr-Al alloys with increasing electrode surface area during deposition. On the other hand, at a constant potential of −1.2 V, where Zr is deposited with a relatively higher Al content (Al-rich Zr-Al alloys), larger currents over −50 mA were passed (not shown) and the current continued to increase with expansion of surface area.

The amount of residual salt in the as-deposited Zr product recovered from the all-chloride salt reaches almost 90% because of the powder-type morphological feature. The residual salt content and current efficiency were compared for various molten salt electrolytes by removing the residual salt using vacuum distillation at 900℃ for 2 h, as shown in Table 1. The current efficiency was estimated with the ratio of actual mass of Zr deposit measured after distillation to the equivalent mass (W_eq) calculated from the total electric charge passed through the experimental setup, according to the following equation,

where, Q_tot is the total charge, M_Zr is the molar mass of Zr, n is the valency of Zr ions, and F is the Faraday constant (96,485 C·mol^-1). The addition of LiF significantly reduced the residual salt content from 92% to 68% of the total amount of Zr deposit by changing the morphological feature from powder to dendritic. Even more reduction of the salt down to 50% was confirmed by the addition of AlF₃ due to the formation of plate-like deposits. The current efficiency was also enhanced from 57.1% for the AlF₃-free case to 88.3% for molten salt with 3.8 mM AlF₃.

Identical experiments were performed using a Zirlo tube-cut specimen as an anode to simulate the treatment of cladding hull waste. Figure 4 shows the cyclic voltammetric curves in the presence of various AlF₃ concentrations in LiCl-KCl-ZrCl₄-LiF salt. The voltammetric responses depending on the AlF₃ concentration for Zr rod are similar to those for Zirlo tubes, the onset and the peak potentials for Zr rod being only slightly greater than those for Zirlo. This may be related to the almost similar composition of Zirlo, i.e. Zr alloyed with only about 1wt% of Nb, 1wt% Sn, and 0.1wt% of Fe, whose electrochemical behavior should be analogous to that of pure Zr. Zr electrorefining experiments were carried out by applying a constant potential of 1.2 V, at which Al can be co-deposited with Zr in the respective molten salt electrolyte containing various AlF₃ concentrations (Fig. 5). The current transient becomes more negative faster for higher AlF₃ concentrations and the anodic potential increases correspondingly. In terms of the morphological features, the size of the radial metallic plate at the top of the salt interface gradually increases, while the diameter of the deposit on the rod decreases with increasing AlF₃ concentration.

To characterize the plate-type deposit formed at the top surface of the salt, the deposit grown at 15.4 mM AlF₃ was subjected to XRD analysis (Fig. 6). The specimen was simply rinsed with ethanol to remove the incorporated salt and placed in an XRD holder with an air-tight polymer seal for residual salt, which consequently produced the higher background signal at lower 2θ values. However, metallic Zr peaks are evident, unlike the diffraction patterns of Zr deposit grown in a chloride-based salt, which exhibited strong Zr oxide and weak Zr metal peaks [7]. Due to the solubility difference between LiCl and KCl in ethanol, only KCl peaks are observed for the salt composition [16]. To confirm the existence and distribution of Al in the Zr deposit, SEM-EDX compositional mapping analysis was performed for the top surface of the same specimen (Fig. 7). The major X-ray signals emitted from the surface were due to Zr Lα and Cl Kα, corresponding to the deposit and the residual salt, respectively. A faint signal of Al Kα, which could also be attributed to the salt component from the residual salt, is shown to have uniform distribution over the surface. Evenly distributed Al signal supports the fact that the added Al was not locally reduced or extracted.

The plate-type Zr product formed at the salt interface appears to have different colors between the top and bottom surface of the product. XPS was utilized to examine the difference in chemical composition between the top and bottom surfaces (Fig. 8). After the deconvolution of the XPS spectra, the more silvery top surface of the deposit, electrorefined in the presence of 15.4 mM AlF₃, yields well defined Zr(3d_5/2) and Al(2p) peaks corresponding to ZrO₂ at 182.8 eV and Al oxide at 74.4 eV, respectively, which were natively formed during sample preparation. The additional Zr(3d_5/2) and Al(2p) peaks at 177.7 eV and 70.4 eV, respectively, are thought to be associated with slightly shifted metallic forms and/or Zr-Al alloys. In comparison with the top surface, the intensity of photoelectron for the bottom surface was lower due to the scattering effect by the higher surface roughness, as indicated by its darker color. The Zr(3d_5/2) peaks that are split into ZrO₂ (183.1 eV) and Zr sub-oxide (182.0 eV) and Al(2p) peaks, corresponding to almost the same positions, are evidently attributed to relatively reduced Al and Zr that might be alloyed with Al at lower binding energies. From the measured intensities of Zr(3d) and Al(2p) spectra, the approximate atomic ratios (N_Zr /N_Al) for the top and bottom surfaces were derived as 2.46 and 5.47, respectively, based on the following equation,

where I is the area intensity of photoelectron peak, σ is the photoelectron cross-section [Scofield factors [17], σ_Al = 0.234, σ_Zr = 1.546], λ is the mean free path [λ_Al = 40.10 Å, λ_Zr = 37.77 Å], and KE the kinetic energy of the respective photoelectrons. The stoichiometries do not exactly match those of the Zr-Al compounds in the phase diagram because metallic Zr can also be deposited with Zr-Al alloys; however, the Zr-Al alloys are approximately positioned between Zr₂Al and Zr. Particularly, relatively Al-rich and Zr-rich Zr-Al alloys are detected at the top and bottom surfaces, respectively, of the plate-type deposit. This shows that lighter Al that is reduced and floats around the cathode at temperatures slightly lower than its melting point (T_{m, Al} = 660℃) can form a seed of radial shape for heterogenous growth of Zr and Zr-Al alloys at the salt interface.

A consecutive Zr recovery process in the plate-type deposit at the salt surface was confirmed by repeating the chronoamperometry experiment with an intermittent break (Fig. 9). After applying a potential of −1.2 V for 60 min in the same LiCl-KCl-ZrCl₄-LiF electrolyte with 15.4 mM AlF₃, the W cathode was lifted up by ~10 mm during the intermittent break and chronoamperometry was continued at the same applied voltage of −1.2 V for 37 min (Fig. 9a). An instant charging current was flown up to −400 mA for few subseconds and the cathodic current began from about −220 mA, lower than the current before the break because the electrode surface area contact with the salt decreased, followed by a gradual increase in current with time. As observed in Fig. 9b, the morphology of the Zr deposit reveals two radial plates at intervals of ~10 mm from the top of the W cathode that meets the alumina shield. At the same time, thick dendritic Zr deposits were formed on the peripheral region of the W rod.

4. Conclusion

Powder-type Zr morphology is the main obstacle to recovering Zr using electrorefining in chloride-based molten salts with high recovery yield. Herein we investigated the effect of adding AlF₃ on Zr electrorefining in chloridefluoride mixed molten salt, LiCl-KCl-ZrCl₄-LiF, at 600℃. Voltammetric experiments in the presence of various concentrations of AlF₃ revealed that a similar monotonic shift of Zr reduction peaks and additional peaks occurred for both Zr rods and Zirlo tubes as anodes, attributed to the formation of Zr-Al alloys. During Zr electrorefining at a constant potential of −1.2 V, AlF₃ addition led to a preferential recovery of Zr around the top of electrode adjacent to the salt surface by forming a radial plate-type Zr deposit, whose size was increased with increasing AlF₃ concentration. The amount of residual salt in the deposit was significantly decreased in the chloride-fluoride mixed salts both in the presence and the absence of AlF₃, compared to the all-chloride based salt. The calculated current efficiency was increased by the addition of AlF₃ in the chloridefluoride mixed salt. Surface analysis of the radial deposit using XPS identified relatively higher Al content on the top surface than on the bottom for the recovered Zr-Al alloys, which supports the assumption that the floating Al acts as a seed for the heterogeneous growth of Zr deposit at the salt surface. The Zr recovery yield should be enhanced in the chloride-fluoride mixed salt by using AlF₃ as an additive at a relatively low temperature of 600℃.