Molten fluoride salts have been selected for use as primary reactor coolant and liquid fuel in the Molten Salt Reactor (MSR) due to their desirable thermophysical and thermochemical properties [[1], [2], [3]]. A eutectic 46.5%LiF-11.5%NaF-42%KF (mol%) melt, commonly referred as FLiNaK, is emerging as a leading candidate fluoride heat transfer salt [[4], [5], [6]]. However, these molten fluoride salts are intrinsically highly corrosive, particularly at high temperatures, which can trigger the dissolution of structural materials in MSRs [[7], [8], [9]]. Under the requirement of corrosion resistance, Hastelloy N alloy, called GH3535 alloy in China [10,11], was developed as a structural material for the above by Oak Ridge National Laboratory [12].

Some literature have suggested that GH3535 alloy is resistant to corrosion without degrading its mechanical strength at temperature up to about 700 ℃ [13]. Unfortunately, after prolonged immersion, the most active alloying elements, such as Cr and Fe, especially those at grain boundaries [14], were selectively attacked by and dissolved into the molten salts [15,16]. For the common alloying elements, the susceptibility to corrosion in molten fluoride salts increases in the following order: Ni, Fe and Cr [17]. Apparently, in GH3535 alloy, the primary metal element that is most prone to be dissolved is Cr, hence in general, original Cr content in the alloy has pronounced influence on the corrosion rate. On the contrary, Ni is relatively immune to the attack from molten fluoride environment at the temperatures of interest. Therefore, it was deemed to be desirable to plate Ni on side of structural materials to the molten salt to enhance their corrosion resistance. Recently, Olson et al. developed a Ni electroplating on Incoloy 800H and studied its corrosion behavior in molten FLiNaK salt [18]. The Ni sulfamate process was used in this study since it developed lower internal stresses in deposits than that of Watt electroplating process, which was advantageous to deposit Ni coating with better adhesion to the substrate. Besides, it provided for high deposition rates of pure Ni, and had superior throwing power. Corrosion test results showed that Ni-plating significantly improved the corrosion resistance of Incoloy 800H in molten fluoride salt, but Cr diffused from the alloy into and successively through the Ni coating, finally dissolved into the molten salt, causing corrosion of the alloy. Besides, Kirkendall voids were formed because of elements interdiffusion between the Ni coating and the alloy substrate. Similar work was conducted on GH3535 alloy in molten FLiNaK salt at 700 ℃. A Ni-electroplating was deposited on this alloy and its corrosion resistance was compared with the bare GH3535 alloy [19]. The corrosion rate was sharply reduced by 90% at the initial stage own to the Ni plating. However, Cr and Fe from the underlying alloy diffused out immediately at high temperatures, which eventually dissolved into the molten salt after long exposure time. Meanwhile, Kirkendall voids were also observed at the interface between alloy substrate and the coating. As a result, to obtain high corrosion resistance, the key problem of elements interdiffusion should be addressed for Ni-plating.

Recently, it was reported that a CrN barrier layer was applied between the Ni coating and GH3535 alloy to prohibit elements interdiffusion between them [19]. The results indicated that the diffusion of Fe and Cr from underlying alloy substrate to Ni coating was effectively inhibited at 700 ℃. The problem of elements interdiffusion was perfectly solved. However, a non-ignorable problem is that the CrN layer has poor electric conductivity, resulting in difficulty in electroplating Ni coating on it. Hence, a thin Ni transition layer should be previously deposited on the CrN interlayer by magnetron sputtering in order to enhance the electrical conductivity of the material. Obviously, the above process is inefficient in economic cost and preparation rate. Therefore, finding a coating with effective elements interdiffusion resistance and high electric conductivity is necessary for developing applicable protective coating on the structural materials in MSR.

TiN shows high potential for application. On one hand, TiN coating has high electrical conductivity at room temperature [20], which has drawn wide concern and been applied in the electronics industry. For example, electrical conductivity of TiN based cermets for solid oxide fuel cells (SOFCs) interconnect application was investigated by Pang et al. [20]. Results showed that TiN has high electrical conductivity, about 1-4 × 10⁴ S/cm. Besides, it can also be used in an electrical probe memory as an optimum material for bottom electrode due to its high electrical conductivity [21]. Although, there have been many successful applications of TiN coating in electronics, whether Ni coating can be directly electroplated on the TiN coated material to obtain a well adherent Ni/TiN composite coating has not been reported.

On the other hand, TiN as a diffusion barrier with high stability and good adherence to the coating and substrate has been proved in prohibiting elements interdiffusion for the usage on gas turbines [[22], [23], [24]]. Elements interdiffusion is a stubborn problem for high temperature protective coatings. It combines with the high temperature oxidation to rapidly consume the beneficial element Al [25]. Lou et al. [22] investigated the effect of TiN barrier on the diffusion of aluminum from the sputtered CoCrAlY coating to a Ni-based superalloy substrate. It was found that TiN barrier could inhibit the diffusion of aluminum from the coating to the substrate. A similar work was carried out by Coad et al. [23]. A TiN interlayer was deposited between MCrAlY coating and metallic substrate. Results showed that TiN layer could remain perfectly stable at up to 1000 ℃ and reduce the diffusion of aluminum from the coating to substrate significantly. Furthermore, the iron diffusion between the coating and substrate was also prohibited by a TiN barrier layer. Therefore, it is now generally accepted that TiN layer is able to effectively prohibit elements interdiffusion and has excellent high-temperature stability. So far, however, whether this interlayer is compatible to the Ni coating and the GH3535 alloy and whether the Ni/TiN composite coating is corrosive resistant in the molten fluoride salts are still unknown.

In this work, we electroplated a Ni coating on the TiN coated GH3535 alloy without extra conductive transition layer. Its effect on corrosion resistance and elements interdiffusion in molten LiF-NaF-KF salt at 700 ℃ was studied. In addition, the evolution of the coating microstructure at elevated temperature was also performed by transmission electron microscopy (TEM).

GH3535 alloy (the analyzed composition is shown in Table 1) is used as the substrate material. Rectangular specimens of the alloy substrate with dimensions of 20 mm × 10 mm × 4 mm were cut, and then holes with diameter of 2 mm were drilled in the center of the specimens for suspension. Finally, all of them were successively ground to 2000-grit SiC paper, polished with diamond paste (2.5 μm), and ultrasonically cleaned within acetone. The preparation of Ni/TiN composite coating consists of two steps. First step, TiN layer was deposited on the alloy substrates by arc ion plating. The specimens were suspended in a vacuum chamber and rotated continuously in order to ensure the uniformity of the TiN layer on each surfaces. Prior to deposition, the chamber was evacuated to a background vacuum of 0.01 Pa, and the substrates were sputter-cleaned for 3 min to remove contaminant layer and to ensure good adhesion of coatings. Then the deposition of TiN layer was carried out in an Ar/N₂ mixture atmosphere, using a pure Ti target. The detailed deposition parameters of TiN layer are as follows: Ar: 8 sccm, N₂: 6 sccm, current: 70 A, temperature: 200 ℃, time: 1 h. Second step, after coated with a thin TiN layer, samples of high electrical conductivity were directly coated with thick Ni coating by using the sulfamate nickel-plating process. The process parameters of nickel plating are as follows: 300 g/L Ni(NH₂SO₃)₂, 40 g/L H₃BO₃, 20 g/L NiCl₂•6H₂O, pH: 4.0, temperature: 25 ℃, current density: 2 A/dm². The plating power supply is steady direct-current source, the anode is nickel plate, and the cathode is TiN/GH3535 sample.

Thereafter, samples were annealed at 1000 ℃ for 3 h in a vacuum furnace in order to make the finished Ni coating compact and reduce its internal stresses. Then corrosion tests of the Ni/TiN coated GH3535 samples in molten FLiNaK salt were carried out in an isothermal static immersion setup as shown in Fig. 1. The setup consists of an internal graphite crucible and an external 304 stainless steel container. The graphite crucibles holding the samples and fluoride salts (a ternary eutectic mixture of FLiNaK (LiF-NaF-KF: 46.5-11.5-42 mol%)) were encapsulated in 304 stainless steel containers and welded shut. The fluoride salts was dried at 200 ℃ for 48 h in a vacuum oven before moving into graphite crucibles and the whole process of molten salt filling process was carried out in a glove box filled with argon. So, argon as the cover gas was used for the capsule test. The setup was then placed in an electrical resistor furnace and maintained at 700 ℃ for 100 h. After corrosion test, the setup was removed from the furnace, and the corroded coupons were took out and cleaned out of residual FLiNaK using 1 mol/L Al(NO₃)₃ for 48 h. The samples were then weighed by electron balance with a sensitivity of 0.01 mg to obtain the material mass loss during corrosion test.

Surface and cross-sectional morphologies were observed by scanning electron microscope (SEM, Inspect F50, FEI Co., Hillsboro, Oregon) equipped with an energy-dispersive X-ray spectrum (EDS, X-Max, Oxford instruments Co., Oxford, UK). Phase compositions and the scanning transmission electron microscopy (STEM) images were also examed by a Jeol JEM 2010 F transmission electron microscopy (TEM) equipped with a Tracor EDS. Besides, phase constituents were analyzed by X-ray diffraction (XRD, D8 Advance, Bruker Co., Brook, Germany).

Fig. 2 shows surface and cross-sectional morphologies, XRD pattern of the as-prepared TiN layer. It can be seen that some droplets and hollows appear on surface of the TiN layer as shown in Fig. 2(a). The macro droplet is likely caused by accumulation of Ti, which is common for the presence of droplets in arc ion plating on account of the inadequate evaporation of Ti target. A uniform TiN layer, about 5.3 μm in thickness, was deposited on the alloy substrate, as shown in Fig. 2(b). XRD peaks shown in Fig. 2(c) demonstrate the existence of TiN phase.

Fig. 3 displays the STEM image and the diffraction rings of the as-deposited TiN layer. The fragmentary rings indicate that preferential growth of fine TiN grains occurred possibly during the deposition. Columnar grains with preferential crystalline orientations are found frequently in magnetron sputtered film. Besides, the EDS results, as shown in Table 2, suggested that the composition of TiN layer was 51Ti - 49 N (at.%).

After preparing a thin TiN layer on the GH3535 alloy substrate, a Ni coating was directly electroplated on the TiN interlayer, forming a Ni/TiN composite coating to protect the substrate from corrosion in molten salt. In order to eliminate the defects and internal stresses of the Ni coating, samples were annealed at 1000 ℃ for 3 h in a vacuum furnace. Fig. 4 shows SEM morphologies of the surface and cross section of the Ni/TiN coated GH3535 alloy after annealing. Grain boundaries are clearly seen on the surface, and the grain size of the Ni coating is in the range of 20-50 μm as shown in Fig. 4(a). Fig. 4(b) indicates that the Ni coating is compact and well adherent to the TiN diffusion barrier. At the interface of TiN/alloy, it can be seen that the TiN interlayer is also adhered well to the substrate after annealing.

Fig. 5 (a) presents the cross-sectional STEM image of the interface between the substrate and the TiN diffusion barrier, and the electron diffraction rings of the fcc-TiN after annealing at 1000 ℃ for 3 h. The crystalline TiN layer shows a dense columnar structure. The electron diffraction rings of the TiN layer became more fragmental and clear after annealing compared with those of the as deposited sample shown in Fig. 3(b), probably because of the growth of TiN grains. No phase decomposition took place in TiN diffusion barrier, and the fraction of built-in structural defects generally decreased upon annealing [26]. EDS results shown in Table 2 indicate that Cr and Fe in the TiN layer are rare, only about 0.17 and 0.18 at.%, respectively. The diffusion of Ni from the superalloy to the TiN diffusion barrier also occurred and the composition of Ni was about 2.37 at.% in the layer. The initial diffusion between substrate and coating should be responsible for the good adhesive bonding. Besides, large amount of nitrogen was released out of TiN layer upon annealing. The ratio between Ti and N atoms in the annealed TiN layer is about 71:25 with an atomic nitrogen content of 25%, much less than that of the as deposited one (49 at.%). To our knowledge, TiN is a non-stoichiometric compound and can release nitrogen at high temperature without causing change in structure. Therefore, the nitrogen release from TiN layer would not cause the phase decomposition. The STEM image of the interface between the Ni coating and TiN diffusion barrier is presented in Fig. 5(b). It can be found that the interface between Ni coating and TiN layer is clear and wavy. The SAD result shows that the grains remained TiN phase in TiN diffusion barrier after annealing.

Surface morphology of the Ni coating of the Ni/TiN coated GH3535 alloy after the molten salt exposure is shown in Fig. 6. Compared to its original surface before corrosion (Fig. 4(a)), the corroded sample exhibited a bumpy surface. EDS results reveal that only Ni element existed on the Ni coating surface as shown in Table 2. What’s more, the Ni/TiN composite coating was strong enough to resist the corrosion of GH3535 alloy from molten salt at 700 ℃, and its corrosion rate was only about 4.39 × 10^-3 mm/year in this molten salt system, much lower than that of the Ni-coated sample without diffusion barrier (49.06 × 10^-3 mm/year) [19].

Fig. 7 shows cross-sectional SEM and STEM microstructures of the Ni/TiN coated GH3535 alloy after corrosion in the molten FLiNaK. It can be seen that the TiN interlayer remained continuous and well adhered to the substrate as well as the Ni coating after corrosion test. Contrary to the case of Ni coated GH3535 alloy without diffusion barrier [19], no Kirkendall voids were formed at the interface between the substrate alloy and the coatings. No doubt, the STEM result is in excellent agreement with that of the SEM as shown in Fig. 7(b). It is also revealed that TiN diffusion barrier adhered well to the substrate and Ni coating. In addition, the interfaces of substrate/TiN diffusion barrier/Ni coating are also clearly seen (Fig. 7(b)).

Further study was performed by employing STEM to obtain the detailed microstructure information at interface between the substrate and the TiN diffusion barrier after corrosion. Results are shown in Fig. 8(a). It can be found that the TiN layer is well adhered to the substrate after corrosion. The interface between TiN diffusion barrier and substrate is smooth, without noticeable difference from that before corrosion (Fig. 5(a)). And there is only TiN phase in the TiN diffusion barrier, without phase decomposition after corrosion. The diffusion processes within TiN interlayer are significantly determined by their microstructural changes upon corrosion at elevated temperature, especially by the dissociation processes and phase decomposition. In the present investigation, the TiN barrier remained stable after corrosion, implying a high elements interdiffusion resistance. In addition, the displacement reaction can lead to the diffusion of elements. The reactions of several metals with titanium nitride and their Gibbs free energy of formation at 700 ℃ are shown in Table 3. The results showed that Cr and Fe etc. couldn’t spontaneously react with titanium nitride, which revealed that the element diffusion in TiN layer through displacement reaction could be neglected. Quantitative EDS result was in good accordance with the above conclusion. It is revealed that the detailed composition of TiN layer was 69.28Ti - 25.60 N - 4.7Ni - 0.3Mo - 0.1Cr - 0.02Fe (at.%) as shown in Table 2. The amount of the elements (Fe and Cr) diffusing into the Ni coating from substrate was so small that an effective barrier role is confirmed for the TiN interlayer. Moreover, the ratio between Ti and N atoms is about 69:26, which is practically the same as that before corrosion. No further nitrogen release within TiN layer occurred during the corrosion process because TiN layer has high chemical stability at 700 ℃ in the molten fluoride salt environment.

Similarly, interfacial microstructure between the Ni coating and TiN diffusion barrier was also studied after corrosion and is shown in Fig. 8(b). Compared with the microstructure before corrosion (Fig. 5(b)), it had two obvious changes after corrosion: (1) the interface became smooth, indicating that elements interdiffusion between Ni coating and TiN layer occurred; (2) a new thin layer, in light grey, formed at the interface between them. SAD pattern revealed that the new thin layer was composed of Ni₄N, which could be formed by the interaction of N and Ni. The formation of Ni₄N at the interface between the Ni coating and TiN diffusion barrier can enhance adhesion strength between layers by some interdiffusion. The composition of TiN layer near the Ni coating characterized by EDS (Table 2) after corrosion was 54.81Ti - 38.19 N - 6.76Ni - 0.24Mo (at.%), and no Cr or Fe were detected, which illustrate that it was difficult for Cr and Fe to diffuse into the Ni coating through the TiN diffusion barrier. The elements interdiffusion was effectively suppressed.

Element mapping as shown in Fig. 9 was employed to determine the chemical composition at the interface between the underlying substrate and the TiN diffusion barrier. It is found that almost no Cr or Fe diffused from the substrate into TiN layer. Besides, it was also difficult for Mo and Ni to diffuse into the TiN diffusion barrier.

With the goal of determining the diffusion coefficients of the individual elements, i.e., Fe and Cr, EDX line scans of the cross-section of the Ni/TiN coated GH3535 alloy after corrosion were performed and are shown in Fig. 10(a) and (b). It is observed that small amount of Fe and Cr diffused from the alloy substrate to the TiN diffusion barrier. However, in the Ni coating, Cr and Fe were absent after corrosion, illustrating that Cr and Fe did not diffuse into the Ni coating through the inert TiN diffusion barrier. To further explore their diffusion behavior, specific part of EDX line-scans were enlarged and are shown in Fig. 10(b). For the convenience of discussion, the abscissa of the EDX line-scans is designated as X and the ordinate as Y. Obviously, point B (or D) was at the interface between substrate and TiN diffusion barrier. Point A (or C) was located at the TiN diffusion barrier, where the content of Cr (or Fe) was close to zero. From Cr and Fe diffusion profiles shown in Fig. 10(b), their diffusion coefficients can be expressed by using the Boltzmann-Matano relation below:

where C denotes the intensity (Y), x the distance from the alloy/TiN layer interface into the TiN layer, t the diffusion time. The term C_R is the initial concentration in TiN layer and D_C* the diffusion coefficient at a site where the concentration of the related element in the TiN diffusion barrier equals to C*. In this investigated coating/alloy system, C_R is assumed to zero for both Fe and Cr since the original TiN layer contained neither Cr nor Fe. Thereby, the diffusion coefficients D_C* of Fe and Cr from the substrate into the TiN diffusion barrier at 700 ℃ were calculated and are shown in Fig. 10(c) and (d), respectively. Fig. 10(c) shows the diffusion coefficient of Cr from substrate into TiN interlayer at 700 ℃. At X_B, the initial diffusion coefficient of Cr in TiN layer at 700 ℃ was about 3.6 × 10^-17 m²/s. However, at X_A, the diffusion coefficient decreased to only about 2% of that at X_B. It can be found that the diffusion coefficient of Cr at the interface between the substrate and TiN interlayer was higher. However, it decreased sharply in TiN layer with increasing the distance from the interface due to the sharp decrease of Cr. In Fig. 10(d), Fe shows higher initial diffusion coefficient at 700 ℃ in TiN layer than that of Cr. Its diffusion coefficient at 700 ℃ at the interface between the substrate and the TiN interlayer was about 6.5 × 10^-17 m²/s, much higher than that of Cr. And the diffusion coefficient of Fe also decreased rapidly with increase of the distance from the interface, similar to that of Cr. As reported, the diffusion of Fe and Cr mainly occurs along high-diffusivity pathways such as grain and especially column boundaries [27]. However, the grain boundaries of different TiN grains are not completely parallel, the zigzag structure will be formed at the grain boundaries of a large number of grains. Due to this zigzag structure in TiN interlayer, the diffusion distances of Cr and Fe atoms are greatly increased, which raises the difficulty of the diffusion of Fe and Cr in TiN diffusion barrier. In addition, as shown in the above results, the diffusion coefficient at 700 ℃ for Fe in TiN diffusion barrier was much higher than that of Cr, which could be explained by the higher interaction of Cr with N, forming Cr₂N during the corrosion exposure at elevated temperature [26]. The interaction of Cr with N increased the diffusion activation energy of Cr in TiN layer. Although the initial diffusion coefficient at 700 ℃ for Fe in TiN layer was higher than that of Cr, an opposite result for the diffusion lengths of the two elements was found, as shown in Fig. 10(b). Basically, the diffusion length of Fe (L_Fe) in TiN layer was derived by: L_Fe =X_D - X_C. Similarly, that of Cr could be expressed as L_Cr =X_B - X_A. Apparently, L_Cr was longer than L_Fe. It is well known that the elements interdiffusion between coating and substrate is caused by the concentration difference of elements between them. When the driving force produced by the concentration gradient is equal to the energy required for elements diffusion, the diffusion coefficient of elements tends to zero. So, the initial concentration of elements in substrate plays an important role in elements interdiffusion behavior. Therefore, the diffusion length of Fe in TiN layer was shorter than that of Cr due to the content of Fe in the substrate was less than that of Cr.

From the above study, the important results can be summarized as follows:

(1) A Ni coating could be directly deposited on the TiN coated GH3535 alloy, forming a Ni/TiN composite coating to protect the substrate from corrosion in molten salt.

(2) TiN diffusion barrier had excellent thermal and chemical stabilities at elevated temperature in FLiNaK molten salt system, without phase decomposition.

(3) The Ni/TiN composite coating was stable enough to resist corrosion in FLiNaK molten salt at 700 ℃. Elements interdiffusion could be effectively inhibited and the corrosion resistance of GH3535 alloy was greatly enhanced. Its corrosion rate was only about 4.39 × 10^-3 mm/year in the molten salt at 700 ℃.