Electrochemical Properties and Analyses of CeCl 3 in LiCl-KCl Eutectic Salt

Thermodynamic and electrochemical properties of cerium in LiCl-KCl eutectic salt have been measured and studied at different concentrations (0.5 – 4 wt%) and temperatures (698 K – 798 K) via both cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) techniques as a part of developing a fundamental understanding and methodology in materials detection and accountability for pyroprocessing technology. CV experiments were performed to determine the diffusion coefﬁcient and apparent standard potential of CeCl 3 on the tungsten working electrode. The diffusion coefﬁcient was calculated by using Delahay equation, and raging from 0.48 × 10 − 5 to 1.01 × 10 − 5 cm 2 s − 1 . Results reveal that the calculated diffusion coefﬁcient of CeCl 3 in the salt follows the Arrhenius temperature relationship and it is weakly affected by the changes in concentration of CeCl 3 . The apparent standard potentials were calculated from peak potentials showing linear relationship with temperature. Exchange current density values of Ce 3 + /Ce couple in the salt were obtained from EIS experiments, ranging from 0.0076 A cm − 2 to 0.18 A cm − 2 . The results indicate that these values follow Arrhenius temperature dependence and increase when CeCl 3 concentration increases.

Pyroprocessing technology has been proposed as another promising method for the recovery and recycle of uranium and actinide elements from the used nuclear fuel. An essential step in this technology is the electrorefining process in which uranium is selectively recovered by using solid cathodes in chloride-based molten salt at high temperature. 1 Then, co-recovery of uranium and transuranic elements can be accomplished by replacing the solid cathodes with a liquid cadmium cathode because the reduction potentials of the elements become close when the liquid cadmium is used as a cathode electrode. 2,3 Since uranium is the major element in most nuclear fuel cycle paths as well as pyroprocessing technology, the assessment of accurate thermochemical data for the element in the molten salt is extremely important. 4 Many studies on the thermochemical properties of uranium have been done in LiCl-KCl molten eutectic salt in different temperature ranges. Masset et al. 5,6 investigated diffusion coefficients of actinides and lanthanides in LiCl-KCl via cyclic voltammetry (CV) and chronopotentiometry (CP). Kuznetsov et al. 7,8 studied the electrochemical behaviors of actinides and rare-earth metals in LiCl-KCl salt. They performed CP and chronoamperometry (CA), and linear sweep voltammetry to determine the diffusion coefficients. Hoover et al. in 2014 9 extended the uranium concentration in LiCl-KCl molten salt up to 10 wt% and observed the electrochemical and thermodynamic behaviors of uranium using CV, CP, and anodic stripping voltammetry. These data are valuable to a development of kinetic models, which can be useful for understanding the main features of actinide deposition at the electrode surface, and also for prediction of material distribution in an electrorefiner of a safeguarding aspect. Zhang 10 developed a kinetic model for electrorefining system showing that the model is capable of predicting the kinetic features and material fluxes of nuclear materials in the electrorefiner.
However, an exchange current density of uranium, which is essential to the physic-based model, has not been well measured and understood due to challenges of the measurement. Only few studies on the exchange current densities of uranium in LiCl-KCl have been published. [11][12][13] Choi et al. in 2009 11 performed a linear polarization method in LiCl-KCl at 773 K and reported the exchange current density of uranium ranging from 0.3 to 0.5 A cm −2 . Later, Ghosh and co-workers reported that the exchange current density of uranium is 8 ± 2 mA cm −2 , by measurements of Tafel plot. 12 Rose et al. in 2015 13 measured the exchange current density of uranium from Tafel plot, ranging from 69.5 ± 9 to 220 ± 32 mA cm −2 . From these literature z E-mail: yoond2@vcu.edu results, a meaningful comparison of the reported data is not possible owing to the dispersed values of the exchange current density of uranium. This may be due to challenges of measuring reliable slope on Tafel plot and electrode surface area where uranium deposition occurs. Therefore, further studies must be conducted on measurement of the exchange current densities of uranium.
These aforementioned challenges provide a motivation for this study to investigate another element that may exhibit similar uranium characteristic and to establish proper electrochemical techniques that can be applied toward uranium. Cerium is one of the common elements to be used as a surrogate material for uranium. The main reason is that cerium has similar ionic size with uranium, and its reduction potential is closer to uranium than other lanthanide materials. 14 Therefore, cerium was selected for the purpose of developing methods to evaluate the electrochemical and thermodynamic properties of uranium in the molten salt. Several studies on the electrochemical behaviors of cerium in the molten salt at high temperatures have been conducted. In 1998, Iizuka 15 conducted CP to determine diffusion coefficient of CeCl 3 at different temperatures. Marsden and Pesic in 2011 14 measured apparent standard potentials and diffusion coefficients of CeCl 3 by CV. They also determined the exchange current densities of CeCl 3 using the linear polarization method.
Thus, the main goal of this work is to measure and analyze thermodynamic and electrochemical properties of CeCl 3 in LiCl-KCl eutectic salt at different concentrations and temperatures using two methods: 1 CV for measuring the diffusion coefficients and apparent standard potentials and 2 electrochemical impedance spectroscopy (EIS) for determining the exchange current densities. This study will provide useful insight into these properties with a unique feature of EIS technique by reducing uncertainty of electrode area measurement because very small current is applied at around an open circuit potential.

Experimental
The electrochemical experiments were performed in an argonatmosphere glove box, as shown in Figure 1a. The oxygen and water concentrations were monitored and maintained below 0.5 ppm ( Figure  1b) for all the experiments. Within the glove box, a Kerrlab melting furnace (Figure 1c) was used to melt and maintain the electrolyte at the desired temperatures. The electrochemical measurements were performed using a VSP-300 potentiostat/galvanostat from Biologic Science Instrument (Figure 1d) at five different temperatures (698, 723, 748, 773, and 798 K). Figure 2 shows the experimental setup within the furnace. The LiCl-KCl-CeCl 3 electrolyte was loaded in a tapered alumina crucible (Coorstek, 99.8% Al 2 O 3 ). This alumina crucible was placed in a secondary crucible designed to contain any molten salt upon possible failure of the primary crucible. Once the salt was melted, the cathode, anode, and reference electrodes were lowered into the salt through the alumina oxide sheaths. The salt temperature was monitored via an inserted thermocouple (see Figure 2).
Tungsten rod (1.5 mm and 2 mm in diameter) was used as the working electrode. The length of the working electrode submerged into the salt was measured, and the measured surface areas were ranging from 0.32 cm 2 to 0.63 cm 2 depending on experimental runs. Prior to using the counter electrode, an oxide layer on the cerium chips was eliminated using sand paper under argon environment. The cerium chips   were then loaded in a molybdenum basket and lowered into the prepared salt. Silver-silver chloride (Ag/AgCl) reference electrode was prepared by contacting a 1 mm Ag wire with LiCl-KCl-5 mol% AgCl in a 7 mm diameter Pyrex tube. At the tip of the Pyrex tube, the thickness of the wall was made thin enough (less than 0.5 mm in thickness) allowing ionic conduction between the solution and electrolyte. Prior to each experiment, the working electrode was anodically cleaned by stripping at a potential of −0.1 V versus the reference electrode for 3 minutes.
Then, an open circuit potential (OCP) was checked to ensure the equilibrium condition has reached in the system. It should be noted that all salts were dried at around 523 K for 5 hours to remove possible moisture despite having the salt in sealed glass ampoules under argon prior the melting processes. The furnace was heated at 5 K/min to avoid thermal shock on the alumina crucible.

Results and Discussion
Cyclic voltammetry (CV) of the LiCl-KCl-CeCl 3 system.-The CV technique was first applied to the pure LiCl-KCl system to identify that no other reaction occurs in the range between 0 V and −2.5 V versus Ag/AgCl reference. The voltammogram of pure LiCl-KCl (in Figure  3) shows that Li reduction starts at −2.55 V (vs. 5 mol% Ag/AgCl). No red-ox reaction between 0 to −2.4 V was observed and residual current in that region was less than 2 mA. Therefore, it was safe to perform the CV experiments over that potential range without worry from other reactions. Figure 4 shows the cyclic voltammograms of CeCl 3 (0.5 wt%, 2 wt%, and 4 wt%) in LiCl-KCl at 773 K. Cerium reduction and oxidation peaks were observed at around −2.2 and −2.09 V versus the Ag/AgCl reference electrode, respectively. For 0.5wt% CeCl 3 , the peak potentials stay at the same potential under different scan rates representing the reversibility of the reaction in the range of the scan rate. However, the peak potentials move slightly in the negative direction according to the scan rate when the concentration of CeCl 3 was increased to 4 wt%. This may be considered as a quasi-reversible reaction. The difference between peak potential and half peak potential can be used to calculate the number of electron transferred by the following expression, 15 where E p is the peak potential (V), E p/2 is the half peak potential (V), R is the universal gas constant (J mol −1 K −1 ), and T is the absolute temperature (K), F is the Faraday constant (C mol −1 ), and n is the number of electrons transferred. The calculated number of electron transferred, n, was ranging from 2.5 to 3.1 agreeing with the expected value for the reduction process of Ce 3+ /Ce. The cathodic peak currents were plotted with respect to the square root of the scan rate to  calculate diffusion coefficient of CeCl 3 in LiCl-KCl using Delahay equation which is known for soluble-insoluble process: 16 where i p is the peak cathodic current (A), S is the electrode area (cm 2 ), C 0 is the bulk concentration of CeCl 3 (mol cm −3 ), v is the scan rate (V s −1 ), and D is the diffusion coefficient (cm 2 s −1 ). Here, the concentration of CeCl 3 (mol cm −3 ) is a function of temperature. The diffusion coefficient of CeCl 3 was determined at different concentrations, as indicated in Figure 5. Small decrease of the values could be observed by increasing concentration from 0.5 wt% to 2 wt%; however, the diffusion coefficients of CeCl 3 in LiCl-KCl salt were approximately the same between the concentration of 2 and 4 wt%. Present study shows smaller values for the diffusion coefficients comparing with those from Marsden and Pesic 14 and Iizuka. 17 But these values possess a similar trend. The diffusivity generally follows Arrhenius temperature relationship, which can be expressed as where, D 0 is the pre-exponential factor, and E a is then an activation energy (kJ mol −1 ) for the diffusion. Therefore, the activation energy can be calculated from the slope when ln(D) is plotted versus 1/T. The values of R-squared between the fitted regression lines and experimental points were all greater than 0.96 indicating a good fit to the data sets. Table I lists the diffusion coefficients of CeCl 3 and the average activation energies with different temperatures at three different concentrations.
From the cyclic voltammogram, the apparent standard potential of CeCl 3 was calculated from the cathodic peak potentials. For a reversible soluble/insoluble system, the cathodic peak potential can be expressed as 5 where E p is the peak potential (V) obtained from the cathodic side in this case, X is the mole fraction, and E 0 * Ce 3+ /Ce is the apparent standard potential. The apparent standard potentials versus Cl − /Cl 2 reference electrode were calculated by using the potential difference between Ag/AgCl (5 mol%) and Cl − /Cl 2 reference electrode from   the study by Yang and Hudson. 18 The calculated apparent standard potentials are plotted in Figure 6 showing a proportional relationship with respect to an increase in temperature. The apparent standard potentials for the concentration of 0.5 wt% and 2 wt% agree with each other (staying within similar ranges of values), but the apparent standard potential for 4 wt% of CeCl 3 was slightly more negative. It is suspected that this is due to the characteristic of quasi-reversibility in the CV curves at high concentration of CeCl 3 and it might be better to use a soluble/insoluble irreversible expression for this calculation in the future study.
Electrochemical impedance spectroscopy (EIS).-The EIS technique was selected and performed to calculate exchange current density (i 0 ) of Ce 3+ /Ce couple in LiCl-KCl salt. Compared to the linear polarization technique, EIS has an advantage that the electrode surface area is almost maintained the same due to an extremely small current that is being applied at OCP or in that proximity. Therefore, uncertainty of electrode area can be reduced significantly in determining i 0 . For analyzing impedance spectra, a simple equivalent circuit was proposed as shown in Figure 7 where R s is the solution resistance, R ct is the charge transfer resistance on the electrode surface, C dl is the double-layer capacitance and W is the diffusion related resistance (Warburg). A frequency ranging from 50 kHz to 50 mHz was used, and the applied potential amplitude was set at 10 mV. Figure 8 shows impedance spectra for 0.5 wt% of CeCl 3 in which the potential was gradually increased from the equilibrium potential (−2.169 V). In general, an impedance should be measured at an equilibrium potential to properly calculate i 0 . However, at the equilibrium potential, the impedance swiftly increases at the high frequency and downward distortion was observed at low frequency as shown in Figure 8 because no ion transfer can occur between the tungsten electrode and cerium ions. Therefore, minimum overpotentials (η = 1-5 mV) was applied to the cell for the cerium reduction to occur at the electrode surface. For an example, in Figure 8, by increasing η from equilibrium potential, a transition point can be observed at −2.172 V (open circles in Figure 8), where also a diffusion related impedance (Warburg impedance) started to be seen at low frequency region. This indicates that electrons transfer and diffusion from the bulk salt to the electrode surface started to occur at that potential. In this case, current density flows through the EIS experiment was only less than 1.5 mA cm −2 . The electron exchange was confirmed by OCP measured right after the EIS experiments. After performing EIS at the potential of −2.172 V, OCP was maintained at the equilibrium potential for 500 s while OCP was released from the equilibrium potential when the applied potential was lower than −2.172 V. Therefore, minimum η for Ce 3+ /Ce reduction to occur were found and R ct were measured at those voltages by fitting the Nyquist plot to the equivalent circuit.
The measured and fitted impedance spectra of Ce 3+ /Ce for the three different concentrations of CeCl 3 at the different temperatures are indicated in Figure 9. Instead of using capacitance and Warburg impedance, constant phase element (CPE) was introduced, which is 19   where T is a constant in F cm −2 s φ−1 is the constant, φ is the number constant between −1 and 1, and ω is the frequency. CPE is useful in fitting the equivalent circuit because it can represent resistor, inductor, capacitor, Warburg response, and combination of these impedances by changing the value of φ. First, the measured spectra were automatically fitted by using randomize and simplex method in Z-fit software (Bio-Logic), then a manual adjustment was done by changing the values of the equivalent circuit components. As the manual curve fitting was performed, the relative error could be minimized below a fraction of 10 −1 . The measured R ct S and η are summarized in Table II. From the measured R ct , i 0 can be readily computed by using 20 where k 0 is the rate constant for Ce 3+ /Ce, and α is the transfer coefficient of Ce 3+ /Ce. From the measured charge transfer resistance, k • can be calculated by assuming α is 0.5 based on the observation from CV experiments that Ce 3+ /Ce reaction follows reversible behaviors with a weak diffusion effect (at slow scan rates). Table III summaries i 0 and k • calculated from R ct . Figure 10 plots the exchange current densities of Ce 3+ /Ce reaction which can be characterized with concentrations and temperatures. The results indicate that the exchange current densities with 0.5 wt% of CeCl 3 are in between 0.0076 A cm −2 and 0.016 A cm −2 , agreeing well with repeated experimental runs. By increasing the concentration of CeCl 3 to 4 wt%, the exchange current density appears to increase up to 0.18 A cm −2 . Marsden and Pesic 14 reported the exchange current density of CeCl 3 at 4 wt% concentration using the linear polarization method. The values of i 0 from this study are slightly higher, but both studies show similar range of values for the exchange current density of CeCl 3 in LiCl-KCl salt.
Based on the given data sets, Arrhenius temperature dependence form can be applied to further looking into temperature effects on the exchange current density using the expressing i 0 = I 0 exp(-E a /RT) where I 0 is the pre-exponential factors (often referred to as an exchange current density at an infinite temperature). Figure 11 shows a plot of the logarithm of i 0 against the inverse temperature. Here, a straight line can be seen for all three different CeCl 3 concentrations. E a and I 0 were calculated from the slope of the straight lines and the intercept of ln i 0 , respectively. E a values for Ce 3+ /Ce were 34.5, 30.9, and  32.4 kJ mol −1 (R 2 > 0.98) for 0.5 wt%, 2 wt% and 4 wt%, respectively. These values are similar to the activation energy for U 3+ /U measured by Roes et al., 13 which is 34.5 kJ mol −1 . As expected, higher activation energy is required for the charge transfer at lower concentration of CeCl 3 . Interestingly, the activation energy values from the diffusion coefficients shown in Table I are within similar range in comparing to those for the charge transfer, but behave in an opposite trend. That is, the activation energy for the diffusion increases with increasing the concentration of CeCl 3 , suggesting that it would be due to interaction between particles at high concentration. Dimensionless quantities of i 0 /I 0 are plotted versus exp(-E a /RT) in Figure  All exchange current densities from three different concentrations are laid on a single straight line. Although the data point at 773 K in the study by Marsden and Pesic 14 is slightly off from the trend line, it is shown here that the exchange current densities from both studies exhibit a similar trend on temperature effect.
Analysis on practical application.-The results of this work provide the fact that cerium is a good surrogate material for uranium since they show similar electrochemical and thermodynamic behaviors in LiCl-KCl eutectic salt. By comparing the properties of cerium with those of uranium, both are very stable in the trivalent form in LiCl-KCl salt and reduced to metal form by gaining three electrons at certain potentials. However, the standard reduction potential of UCl 3 is ranging from -2.4 to -2.6 V versus Cl 2 /Cl − reference electrode, [5][6][7][8][9] which is about 0.7 V more positive than the standard reduction potential for CeCl 3 . The diffusion coefficients for UCl 3 in LiCl-KCl molten salt have been reported by many researchers, [4][5][6][7][8][9]11 which are shown in the Figure 13. Although the values for UCl 3 are generally higher than Ref. [4] Ref. [5] Ref. [6] Ref. [7] Ref. [8] Ref. [9] Ref. [11] D [×10  Ref. [11] Ref. [12] Ref. [13] i 0 [A cm -2 ] 1000/T [K -1 ] Figure 14. Plot of the exchange current densities for U 3+ /U reaction from other research studies comparing with those of Ce 3+ /Ce measured in this study.
those for CeCl 3 , the diffusion coefficients for both UCl 3 and CeCl 3 are in the same order of magnitude and can be correlated with the temperature. The activation energies for the diffusion of UCl 3 have been reported, ranging from 24.2 to 34.4 kJ mol −1 , 5,7 which is in a good agreement with the activation energy for CeCl 3 . The similarity between both the activation energies for the diffusion may be owing to the similar ionic size of uranium and cerium. The exchange current densities of uranium in the molten salt have been reported with few different methods. [11][12][13] As shown in Figure 14, the reported i 0 for U 3+ /U reaction is scattered in the order from 10 −2 to 10 −1 ; thus, it is difficult to compare each data or observe its trend. The main reason of the deviation could be that the authors performed the different methods to measure i 0 at different temperatures and concentrations, and also be the challenges that have been previously mentioned. Therefore, further study is necessary to determine the exchange density for U 3+ /U reaction, and observe its trend regarding to the concentration and temperature. EIS technique in this study shows an alternative method in determining i 0 of CeCl 3 , and can also be an important tool to investigate the exchange current density of UCl 3 in LiCl-KCl eutectic salt.

Conclusions
Electrochemical properties of CeCl 3 in LiCl-KCl eutectic salt at different concentrations and temperatures have been studied by using CV and EIS techniques. From the CV experiments, it has been shown three-electron exchange for the reduction of Ce 3+ /Ce. The diffusion coefficients of CeCl 3 in LiCl-KCl were measured and calculated to be from 0.48 × 10 −5 to 1.01 × 10 −5 cm 2 s −1 , which can be correlated with temperature using Arrhenius expression. The results reveal that the concentration of CeCl 3 has a weak effect on the diffusion coefficients. Comparing with the values for UCl 3 , the diffusion coefficients of CeCl 3 are slightly smaller than those of UCl 3 . Apparent standard potentials were also computed by using peak potentials from the CV experiments, which were linearly dependent on temperature. EIS ex-periments were performed to determine exchange current density of Ce 3+ /Ce couple in LiCl-KCl molten salt system. Minimum η was applied for Ce reduction to happen and the charge transfer resistance was measured to calculate the exchange current density. The exchange current densities range from 0.0076 A cm −2 and 0.18 A cm −2 , which can be related to temperature and concentration (see Table III). From Arrhenius temperature dependence, the activation energy for Ce 3+ /Ce exchange was determined though EIS data, which is in the same range obtaining through CV data sets and in similar range with the activation energy for U 3+ /U. 13 By plotting dimensionless quantities of the exchange current density, the exchange current densities of Ce 3+ /Ce reaction in this work are in good agreement with those measured by linear polarization method. 14 In comparison with the exchange current densities of U 3+ /U measured by other researchers, the values for Ce 3+ /Ce are in a similar order, but a meaningful comparison is hard to be made due to the dispersed data for uranium.