Oxide nuclear fuels discharged from light water reactors (LWRs) are converted into their metallic forms via the oxide reduction (OR) process in order to recover metallic U selectively in the subsequent electro-refining process [1-3]. Employing a carbon anode during the OR process at a voltage high enough to decompose LiCl salt causes profound changes during the OR process [4-5]. Normally, platinum is widely employed as the anode material owing to its high stability even under the severe condition of oxygen generation in LiCl salt at 650℃. However, the use of Pt limits the operational voltage of an OR system due to the electrochemical dissolution potential of Pt. Thus, the operating current and rate of reduction are also limited under this condition. It was also revealed that Pt degrades at a noticeable rate when the OR operations are repeated . Although the use of carbon anodes at high potentials can eliminate these concerns, the generation of chlorine was pointed out as a new issue . Under a high-voltage operating condition, LiCl salt decomposes to produce metallic lithium and chlorine gas. Subsequently, Li reacts with used nuclear fuel (UNF) oxides, mainly UO2, to produce Li2O according to the following reactions.
The equipment and materials previously designed for oxygen generating conditions should be revised, because chlorine is far more corrosive than oxygen especially at high temperatures.
In addition to chlorine gas, the simultaneous generation of O2, CO, and CO2 is expected at the anode according to the reactions below. Oxygen comes from Li2O, which is a reaction product between lithium and UNF, as noted above.
Among the gas products at the anode, the effects of chlorine and oxygen are investigated in the present study. This is done because the effects of CO and CO2 on corrosion are likely to be low. Brown et al.  investigated the corrosion behavior of certain commercial alloys under Cl2 or HCl gas flows. In their work, it was verified that Ni and its alloys (such as Inconel and Hastelloy) exhibited corrosion resistance superior to those of carbon steels . Ihara et al. investigated the corrosion behaviors of Fe , Ni , and Cr  under an HCl-O2 mixed gas flow. However, data pertaining to corrosion under a Cl2-O2 mixed gas flow is not available in the literature. As a starting point for a quantitative analysis of the effect of a Cl2-O2 mixed gas flow on commercial alloys, our group investigated the corrosion behavior of stainless steel 316 (SS-316) exposed to various Cl2-O2 mixed gas conditions . This study revealed that the presence of oxygen could reduce the corrosion rate caused by chlorine, although the maximum temperature for SS-316 was at or below 300℃ owing to severe corrosion by chlorine gas. Inconel X-750 was tested under conditions identical to those used with SS-316 . Excellent corrosion resistance of Inconel X-750 (only 1/22~1/41 of the corrosion rates of SS-316) was revealed in that study. As part of a series of such studies, the corrosion behavior of another Ni alloy, Hastelloy C-276, is investigated in this paper, as a possible reactor material for use under the Cl2-O2 mixed gas condition of the carbon-anode-based OR. The compositions of the alloys (Hastelloy C-276, Inconel X-750, and SS-316) discussed above are listed in Table 1 .
A quartz tube reactor (40 mm diameter) horizontally positioned with an electrical furnace in the middle was employed for the experiments. A Hastelloy C-276 plate (1.0 mm thick with the major constituents being Ni, Cr, Mo, Fe, W, and Co) was cut into 30×10 mm pieces, four of which were used in every experiment. The specimens were washed using deionized (DI) water and ethanol before the experiments. The average weight of each of the four specimens employed in each experiment was 10.474 g. The flow rate of each gas feed (argon, chlorine, and oxygen) was individually controlled using mass flow controllers. It should be noted that the specimens used in this work were intentionally not polished in order to gather corrosion rate data under a condition similar to the actual operation of OR equipment.
The effect of the corrosion reaction temperature was identified within a temperature range of 300–700℃ under a flow of 170 mL·min-1 Ar + 30 mL·min-1 Cl2 for 8 h. Various gas compositions (30 mL·min-1 Cl2, 20 mL·min-1 Cl2 + 10 mL·min-1 O2, and 10 mL·min-1 Cl2 + 20 mL·min-1 O2) were employed to determine the effect of the mixed gas feed at 600℃. The argon flow rate was kept constant at 170 mL·min-1 during the gas composition experiments. The effect of the reaction time was also investigated by varying the reaction durations (1, 2, 4, and 8 h) for each gas composition. The effect of the chlorine concentration was investigated by flowing 5 mL·min-1 Cl2 + 195 mL·min-1 Ar, 10 mL·min-1 Cl2 + 190 mL·min-1 Ar, and 20 mL·min-1 Cl2 + 180 mL·min-1 Ar at 600℃ for 8 h. The specimens were thoroughly washed using DI water and ethanol after each experiment to remove any surface chlorides or oxides.
The HSC chemistry software (version 9.5.1) was used for the calculation of the phase stability diagrams of the Ni- Cl-O, Fe-Cl-O, and Mo-Cl-O systems at 600℃ . The software was also employed in deriving Gibbs free energy change (ΔG) and equilibrium constant (K) values of various reaction equations.
Changes in the surface morphology caused by the various gas compositions were recorded and studied using a scanning electron microscope (SEM, Hitachi SU-8020) with an energy dispersive X-ray spectroscope (EDS, Horiba, X-MAX) to analyze the composition.
3. Results and Discussion
The images shown in Fig. 1 should help readers understand what occurred during the reactions. The picture denoted as “Before washing” was taken after reacting the bare specimens (“Before reaction”) at 700℃ under a flow of 30 mL·min-1 Cl2 + 170 mL·min-1 Ar for 8 h. It is clear in the figure that flaky yellow-brown chlorides formed abundantly on the surface. In addition, dense layers of brown chloride also coated the specimens. After the specimens were thoroughly washed using DI water and ethanol, the chlorides were completely removed, and the cleaned samples were labeled “After washing” (Fig. 1). The specimens lost their original shiny, smooth surfaces. Some of the reaction products evaporated to be collected in the cool region of the quartz reactor, as shown in the figure (denoted as “Evaporated products”). The flaky chlorides observed in the “Before washing” image were also observed in the “Evaporated products” photograph. In addition, the dark brown reaction products were also observed in the “Evaporated products,” suggesting that multiple chlorides evaporated during the reaction. These evaporated products were separately analyzed using the SEM-EDS system, and these results are shown in Fig. 2. The yellow-flaky chloride exhibited a thin plate shape, and a high Ni concentration was identified along with chlorine. This outcome suggests that the flaky product is Ni chloride. On the other hand, Cr and Ni were identified as the main components of the dark brown product. Significant amounts of Fe and Mo were also observed in this product. It should be noted that only Ni and Cl were detected in the reaction products of Inconel X-750 . Though it is unclear as to how Cr could have evaporated despite the non-volatile characteristics of Cr chlorides, this result clearly shows that the corrosion reaction mechanism between Hastelloy C-276 and Inconel X-750 is quite different.
In this work, the corrosion rate was employed for a quantitative comparison of the weight loss per unit surface area and the time under various conditions. The corrosion rate was calculated using the following equation,
where Wi and Wf represent the initial weight and the final (after washing) weight of the sample, S represents the surface area measured under the initial condition, and t indicates the reaction time in hours. Figures 3(a) and 3(b) show the effect of the reaction temperature under a flow 30 mL·min-1 Cl2 + 170 mL·min-1 Ar on the corrosion rate of Hastelloy C-276 (compared to the corrosion rates of SS- 316 (3a) and Inconel X-750 (3b), respectively [11-12]). It is clear that the corrosion resistance of Hastelloy C-276 is far better than that of SS-316, as shown in Fig. 3(a). Although the corrosion rate of Hastelloy C-276 was similar to that of Inconel X-750 over the entire range used in the experiments, certain differences were also observed. Previously, the corrosion rate of Inconel X-750 was not measurable at 300 and 400℃ because the weight changes were less than 1 mg in those cases. However, for Hastelloy C-276, corrosion rates of 8.86×10-1 and 9.24×10-1 g·m-2·h-1 were found at 300 and 400℃, respectively (as shown in the inset of Fig. 3(b)). In addition, Hastelloy C-276 exhibited a corrosion rate of 4.70×101 g·m-2·h-1 at 700℃, which is 9.3% higher than that (4.31×101 g·m-2·h-1) of Inconel X-750 . These results suggest that Inconel X-750 is a better choice than Hastelloy C-276 under a mixed gas flow of chlorine and argon. The corrosion rate equation shown below was derived using the data in the temperature of 500–700℃ with the results included in Fig. 3(b).
Fig. 4(a) shows the relationship between the reaction time and the weight change measured under a flow of 30 mL·min-1 Cl2 + 170 mL·min-1 Ar. A slight increase in the specimen weight was observed after 1 h of the reaction. After washing, a significant decrease in the weight was identified owing to the removal of chlorides formed on the surface of the specimen, as shown above. The “before washing” weight-change values turned negative and decreased with an increase in the reaction time. At the same time, the “after washing” weight decreased more rapidly than the “before washing” weight, indicating that the formation rate of chlorides was higher than their evaporation rate. This trend was identical in the cases with 20 mL·min-1 Cl2 + 10 mL·min-1 O2 + 170 mL·min-1 Ar and 10 mL·min-1 Cl2 + 20 mL·min-1 O2 + 170 mL·min-1 Ar, as shown in Fig. 4(b) and 4(c), respectively. A comparison chart was drawn of the “after washing” weight changes, and is presented in Fig. 4(d). Interestingly, the values were close in all three cases, while the presence of oxygen significantly reduced the weight-change values for Inconel X-750 . The relationship between the reaction time and the corrosion rate is shown in Fig. 4(e) as a function of the gas composition. In the 30 mL·min-1 Cl2 and 20 mL·min-1 Cl2 + 10 mL·min-1 O2 cases, the corrosion rate decreased with an increase in the reaction time, while the opposite was observed in the 10 mL·min-1 Cl2 + 20 mL·min-1 O2 case. After 8 h of reaction, the corrosion rate values were 5.339 (20 mL·min-1 Cl2 + 10 mL·min-1 O2) > 4.858 (30 mL·min-1 Cl2) > 4.578 g·m-2·h-1 (10 mL·min-1 Cl2 + 20 mL·min-1 O2). This result is quite different from the data for SS-316 and Inconel X-750, where the corrosion rate was around two times higher in the 30 mL·min-1 Cl2 case, relative to the other cases. This result suggests that Inconel X-750 is a better option than Hastelloy C-276 owing to its superior performance under a chlorine-oxygen mixed gas flow. It should also be noted here that the corrosion rate of Hastelloy C-276 was also measured under a flow of 30 mL·min-1 O2 + 170 mL·min-1 Ar for 8 h at 600℃. Interestingly, the weight of the washed specimens increased after this experiment, providing a corrosion rate of 1.336×10-1 g·m-2·h-1. This is only 2.75% of the outcome of the 30 mL·min-1 Cl2 + 170 mL·min-1 Ar case.
The effects of the chlorine concentration were also investigated, and these results are shown in Fig. 5. It is clear that the corrosion rate is independent of the chlorine flow rate within the range 5–30 mL·min-1, while the presence of oxygen significantly affects the corrosion rate, as shown in the figure. Here, it is interesting to revisit the impact of the oxygen concentration, which was briefly discussed above. The corrosion rate was accelerated in the 20 mL·min-1 Cl2 + 10 mL·min-1 O2 condition, while it was reduced in the 10 mL·min-1 Cl2 + 20 mL·min-1 O2 condition. This outcome is different from the cases of SS-316 and Inconel X-750 with regard to two aspects: 1) only reduced corrosion rates were observed in the presence of oxygen, and 2) the corrosion rate was less than half in the 20 mL·min-1 Cl2 + 10 mL·min-1 O2 condition in these two alloys. Therefore, it can be concluded that the presence of oxygen can accelerate or decelerate the corrosion rate according to the flow rate, while the change is within ± 8% compared to the chlorineonly condition.
This unique behavior of Hastelloy C-276 was investigated via thermodynamic calculations. Phase stability diagrams of the Ni-Cl-O, Fe-Cl-O, and Mo-Cl-O systems at 600℃ calculated using the HSC chemistry code are shown in Fig. 6. The following reaction equations were derived from the phase stability diagram of the Ni-Cl-O system shown in Fig. 6(a) as possible reaction pathways under the condition of this work.
In the above equations, the ΔG and K values are indicators that determine direction of the reactions. Specifically, a negative ΔG (with K > 1) value indicates that the reaction will proceed along the right-hand direction while a positive ΔG (with K < 1) value means that the reverse reaction is preferred. It should be noted that the chlorination reaction of NiO (ΔG = -14.3 kJ) was experimentally shown to occur [15-16]. However, the profound decrease in the corrosion rate of Inconel X-750 in the presence of oxygen  strongly suggests that the reaction rate of the NiO chlorination reaction is significantly slower than that of the Ni chlorination reaction. According to the phase stability diagram of the Fe-Cl-O system shown in Fig. 6(b), the major reaction products in this system are Fe3O4, Fe2O3, FeCl2, and FeCl3. The following equations represent the possible reaction pathways between Fe oxides and chlorine gas in the Fe-Cl-O system.
The thermodynamic values noted in the above suggest that it is very hard to induce these reactions along the righthand direction. Thus, it can be concluded that the presence of oxygen leads to the formation of Fe oxides which are not reactive with chlorine gas resulting in the reduced corrosion rate of SS-316. The reaction behavior of Mo, the second most abundant constituent of Hastelloy C-276, was also investigated via thermodynamic calculations with the phase stability diagram shown in Fig. 6(c). Feasible reaction pathways between Mo oxides and chlorine gas derived from the diagram are listed in the following.
Interestingly, among the reaction equations above, only the reaction between MoO2 and Cl2(g) to produce MoO2Cl2 exhibited a negative ΔG value. As the MoO2Cl2 phase is stable at a certain region of the diagram, we speculate that this reaction pathway is one of the reasons that resulted in the unique behavior of Hastelloy C-276 in the presence of oxygen. Specifically, MoO2 will be oxidized to MoO3 under an oxygen rich condition, whereas MoO2Cl2 will be produced under a chlorine rich condition. Combining the outcomes from the corrosion rate measurements and the thermodynamic calculations, it is suggested that the corrosion rate of Hastelloy C-276 increases via the Mo → MoO2 → MoO2Cl2 reaction pathway under a Cl2 rich condition while it decreases via the Mo → MoO2 → MoO3 reaction pathway under an O2 rich condition. It should be noted that this conclusion includes one assumption that the reaction rate of the Mo → MoO2 → MoO2Cl2 pathway is faster than that of the Mo → MoCl3 → MoCl4 pathway under the condition of this work.
The surface morphology observation results are shown in Fig. 7. A smooth surface was observed for the bare sample, while the growth of an oxide layer yielded a rough surface in the 30 mL·min-1 O2 + 170 mL·min-1 Ar case. However, totally different patterns were observed in the chlorine and chlorine-oxygen mixed gas samples. Under magnification of ×60, the specimen reacted in 30 mL·min-1 Cl2 + 170 mL·min-1 Ar showed a smooth surface. However, rough surfaces with sharp and vertical etching scratches were revealed under magnification of ×3000. The surface roughness was more severe in the 20 mL·min-1 Cl2 + 10 mL·min-1 O2 and 10 mL·min-1 Cl2 + 20 mL·min-1 O2 cases. As shown in the figure, these specimens appear as though they were intentionally etched to produce sharp scratches on the surface. These results clearly prove that the co-existence of chlorine and oxygen induced a different surface reaction mechanism, although the details require further investigation.
The corrosion behavior of Hastelloy C-276 was investigated under various chlorine-oxygen mixed gas compositions, with the results compared to those of SS-316 and Inconel X-750. Although Hastelloy C-276 showed much slower corrosion rates compared to those of SS-316, the outcome was comparable to Inconel X-750 only in the chlorine-argon mixed gas flow. The dramatic decrease in the corrosion rate under a Cl2-O2 mixed gas flow observed in the other alloys was not found in Hastelloy C-276. The corrosion rate values were 5.339 (20 mL·min-1 Cl2 + 10 mL·min-1 O2) > 4.858 (30 mL·min-1 Cl2) > 4.578 g·m-2·h-1 (10 mL·min-1 Cl2 + 20 mL·min-1 O2) after 8 h of reaction at 600℃. The formation of a unique surface with vertical scratches was observed in the chlorine-oxygen mixed gas atmosphere. Overall, the authors recommend Inconel X-750 as the best option thus far for carbon-anode-based OR applications.