Figure 1 shows a real picture of the welded joint of the pipe bellow adaptor in vessel 03 where molten fluoride salt was processed with HF/H₂, showing that this pipe is almost completely blocked. Figure 2A is the cross-sectional SEM image of the blocked pipe. The EDS elemental mapping image in Figure 2A shows that the blockage is composed of dense depositions and cracks. The elemental mapping images in Figures 2B–I indicate that the blockage is composed of the elements Ni, Cr, Mo, F, K, Na, C, and O, which is different from molten FLiNaK salt. The XRD pattern in Figure 3 reveals that the blockage is composed of KNiF₃, Ni, K₂NaNiF₆, Na₅Al₃F₁₄, LiF, K₂NiF₄, Na₂Mo₃O₆, and Cr₉Mo₂₁Ni₂₀ phases. KNiF₃, Ni, and K₂NaNiF₆ are the main components of the blockage, where Ni should be from Ni vessel. Ni-containing fluorides should be corrosion products of Ni vessel induced by impurities in molten FLiNaK salt. KNiF₃, K₂NiF₄, and K₂NaNiF₆ should be complexes of NiF₂ with KF and NaF. The binary diagrams of NiF₂ with alkali fluorides reveal that NiF₂ can react with LiF, NaF, and KF to form (Li_{1 - 2x} Ni _x )F, NaNiF₃, K₂NiF₄, and KNiF₃ (Ocadiz-Flores et al., 2018). The results of the XRD pattern in Figure 3 indicate that there are KNiF₃, K₂NaNiF₆, and K₂NiF₄ phases in the blockage. Lack of (Li_{1 - 2x} Ni _x )F and Li₂NiF₄ in the blockage should be attributed to the smaller ionic radius of Li⁺ than K⁺ and Na⁺ because the alkali cation with a smaller ionic radius is closer to the F⁻surrounding the Ni²⁺ to form a weakened complex (Ocadiz-Flores et al., 2018). KNiF₃, K₂NaNiF₆, and K₂NiF₄ in the blockage should be complexes of the corrosion product of Ni with KF and NaF. The KF–NiF₂ phase diagram reveals that NiF₂ firstly reacts with KF to form K₂NiF₄ and then forms KNiF₃ with the increase in the concentration of NiF₂ above 33 mol% (Ocadiz-Flores et al., 2018). The XRD pattern in Figure 3 shows that KNiF₃ is the main Ni-containing fluoride. The existence of KNiF₃ indicates that there should be a large amount of NiF₂ in the terminal salt during transportation between different vessels. The saturation concentration of NiF₂ in the FLiNaK salt at 550°C is only 0.42 wt% (Jardan et al., 1956). The amount of NiF₂ in molten FLiNaK salt which is more than the saturation concentration can result in the turbidity of molten salt (Jardan et al., 1956). In addition, the melting points of KNiF₃, K₂NiF₄, and NaNiF₃ (Ocadiz-Flores et al., 2018) are higher than the processing temperature (i.e., 580°C). Therefore, the extra KNiF₃, K₂NiF₄, and NaNiF₃ in the terminal salt should be in the form of precipitants. The Ni-containing fluorides can be reduced by H₂ to form metallic Ni, which is also in the form of a precipitant.

X-ray can penetrate into the matter. The difference in density and atomic composition exhibits a different X-ray absorption coefficient, which results in the difference of the gray value in the X-ray radiographies at different scanning points. The shaded gray or black color is for low gray value, while the white color is for high gray value. The gray value is linearly related to the density of materials. X-ray CT is a non-destructive technique to image the internal feature of one object in three dimensions (Otani et al., 2010). Therefore, industrial 3D CT was employed to study the internal structure of pipes in the FLiNaK processing system. The 2D CT images in Figures 4A,B show that there is a layer-like structure in the blockage. The layered structure is consistent with the SEM image in Figure 2A. It appears that the alternate layers of depositions and cracks are like a layered structure. The alternate layers of depositions and cracks should be related to the periodic transportation of molten salt between different vessels. The 3D CT image in Figure 4C shows that the welded pipe is almost fully blocked at the inner of the weld. Therefore, the mixed H₂/HF sparging gas cannot come into the vessel, and molten salt cannot be transferred between the different vessels. To unveil the puzzle, the blockage and the salt that adhered to the blocked pipe were removed by using image processing, and the result is shown in Figure 4D. It can be clearly seen that there is a welding defect at the backside of the weld. It is a typical incomplete root penetration which is a failure of the weld metal to penetrate into the root of a joint (Hughes and Matthews, 2009; Mandal and Mandal, 2017). It is worth noting that this defect occurs at all the backside for the half-girth weld. The incomplete root penetration defect can stop the transfer of precipitants in the terminal salt and then deposit on the welding defect area to result in the blocked pipe. Figure 5 shows the internal structure of another welding pipe above the adaptor in vessel 02 which was scanned by 3D CT as well. It is noteworthy that the internal surface of the girth weld is very smooth, and no welding defects and depositions were found on the inner surface of the pipe. A comparison of Figure 4 and Figure 5 indicates that the incomplete root penetration defect plays an important role in the blockage of the pipe. Thus, this kind of welding defect should be avoided in molten salt purification processing system and MSRs.

Several reasons can be usually ascribed to the incomplete root penetration defect (Hughes and Matthews, 2009; Mandal and Mandal, 2017). The first one is low welding current, which normally results in shallow penetration. The second one is the unreasonable design of the welding groove and gap, such as thick root face and narrow or too wide root gap, which results in the root being hard to penetrate. The third one is poor surface cleaning before welding or fast heat dissipation rate. A better solution to prevent the incomplete root penetration defect is employing a higher welding current and reasonable welding preparation. There should be filters in the molten salt processing system to remove the precipitants.

Figure 6 is a 3D CT image of the pipe which is immersed in salt in vessel 04, with the end of the pipe shown to be completely blocked. Therefore, the H₂ sparging gas cannot come into the vessel to purify the molten fluoride salt, and the molten fluoride salt cannot be transferred between vessels. The origin of the blockage in the pipe immersed in salt in vessel 04 should be different from that in the pipe bellow adaptor in vessel 03. The structure of the blocked pipes is different. The blocked pipe immersed in salt in vessel 04 is a straight pipe, while the pipe bellow adaptor in vessel 03 is blocked at the welding site with inadequate root penetration. The environments in the pipe in vessel 04 and 03 are different. The pipe in vessel 03 was used to bubble with H₂/HF to remove oxygen-containing and sulfur-containing impurities in molten FLiNaK salt and transfer the molten salt between vessels. The pipe in vessel 04 was for H₂ to remove metallic ions in the molten FLiNaK salt. The vessels and pipes in the processing system are composed of metallic Ni. Impurities in raw fluoride salts and residual HF in molten salt can result in the corrosion of the Ni vessel to form NiF₂. The results in Figure 3 indicate that NiF₂ can react with KF and NaF to form K₂NiF₄, KNiF₃, and K₂NaNiF₆. Ni-containing impurities in molten FLiNaK salt can affect the corrosion of nickel-based alloy by the following reaction: Ni²⁺ + Cr ⇄ Cr²⁺ + Ni (Briggs, 1964; Pavlík et al., 2015). Therefore, Ni²⁺ in molten fluoride salts should be removed by processing. It is known that H₂ can reduce Ni²⁺ into metallic Ni. Therefore, the molten salt transferred into vessel 04 was purged with H₂ to remove residual Ni²⁺ in it. The XRD pattern in Figure 7 shows that the blockage is mainly composed of metallic Ni, indicating that the blockage was induced by the deposition of Ni that was reduced by H₂ in the molten salt.

Figure 8 is a diagram of the reactions in vessel 04. The salt in vessel 04 was processed with H₂ by reaction 1). H₂ and Ni²⁺ in the molten FLiNaK salt make up an electrochemical corrosion cell, where H₂ at the end of the pipe in vessel 04 can be referred to as the anode and Ni²⁺ in the salt can be referred to as the cathode. The oxidation of H₂ generates electrons [reaction (2)], while the reduction of Ni²⁺ consumes electrons [reaction (3)]. The metallic outlet pipe provides an electronic path for the movement of electrons from the anode to the cathode. The flow of current is from the cathode to the anode, which is opposite from the flow of electrons. Then, the current flows from the anode to the cathode by the movement of positively charged ions in the molten FLiNaK salt. At the H₂ gas outlet, Ni²⁺ is reduced by H₂ to form metallic Ni and deposited on the surface of the pipe. Ni²⁺ at the H₂ gas outlet is consumed to form a concentration gradient of Ni²⁺ in the molten FLiNaK salt, which then results in the diffusion of Ni²⁺ in the molten FLiNaK salt towards the H₂ gas outlet. As time goes on, Ni²⁺ in the molten FLiNaK salt will be removed little by little, and the deposition of metallic Ni results in the mouth of H₂ gas outlet becoming narrower and narrower until the pipe is finally blocked. H 2 + N i 2 + ⇄ Ni +2 H + Δ G = − 6.843 kcal ( at 550 0 C ) (1) H 2 = 2 H + + 2 e - (2) N i 2+ + 2 e - = Ni (3)

The results shown above indicate that the blockage located at the end of the pipe immersed in salt in vessel 04 is mainly related to the reduction of Ni²⁺ by H₂. Therefore, this kind of blockage should be avoided by decreasing the concentration of Ni²⁺ in salt and the amount of Ni²⁺-containing precipitants. The ratio of H₂ and HF in the sparging gas should be controlled, and residual HF in molten salt should be removed by sparging with anhydrous helium to decrease the corrosion of the Ni vessel and then decrease the amount of Ni²⁺-containing corrosion products. Meanwhile, filters should be added in the processing system to remove Ni-containing precipitants from the molten salt.