The effect of the microstructure on the corrosion behavior of N5 superalloy in a molten (Li,Na,K)F eutectic salt

Abstract

The corrosion of structural materials in molten fluorides, influenced by the microstructure of the structural materials, is considered to be a great challenge for the development of molten salt reactors using fluorides as fuel carriers or coolants. In this paper, the corrosion behavior of the monocrystal and polycrystal N5 superalloy in molten (Li,Na,K)F eutectic salt at 700 °C is investigated. The results indicate that the mass loss of the polycrystal is much higher than that of the monocrystal after immersion in molten (Li,Na,K)F for 100 h. The corrosion of the monocrystal and the polycrystal N5 superalloys occurs mainly through the preferential dissolution of Cr and Al, which leads to an Al/Cr-depleted layer and some internal voids. Furthermore, the corrosion rate of the polycrystal N5 superalloy is obviously accelerated by the existence of grain boundaries, which results in a mass loss of about 16.995 mg cm⁻² and a Al/Cr-depleted layer of around 175 μm, compared to 11.373 mg cm⁻² and 125 μm, respectively, for the monocrystal N5 superalloy.

---

1. Introduction

The molten salt reactor (MSR) is considered to be a priority for the development of fourth generation reactors due to its remarkable advantages in safety, economics, and the continual development of fissile resources.¹ Molten fluoride salts, such as LiF–NaF–KF (46.5–11.5–42%, mole percent) and LiF–BeF₂ (67–33%, mole percent), are often used as the coolants and fuel carriers in MSRs due to their excellent thermal physical and chemical properties.^2–4 However, the corrosion of the structural materials caused by the molten fluoride salts at the desired operating temperatures between 650 and 850 °C is an inevitable problem for the application and development of MSR.⁵ In recent years, increasing attention has been focused on the investigation of the material compatibility with various molten fluoride salts by ORNL (Oak Ridge National Laboratory-USA), the University of Wisconsin, and National Institute for Fusion Science (NIFS), etc.^5–9 The results reveal that the Ni-based alloys with a low Cr content are regarded as the most suitable structural materials for MSR, among which the Hastelloy-N (Ni–17%Mo–7%Cr–5%Fe, mass percent) developed by ORNL exhibits a great potential for long-term applications.

However, the intergranular corrosion of the structural materials is the typical corrosion pattern in MSR.¹⁰ The corrosion mechanism of the alloys in molten fluoride salt has already been investigated by many researchers.^8–15 Olson, et al. reported that the alloys always underwent intergranular corrosion and the deepest penetration and the first attack always occured at the grain boundaries of the alloy.⁸ Moreover, it is speculated that the precipitation of Cr carbides due to the higher C content in the alloy at the grain boundaries could result in the intergranular void structure.⁹ You, et al. also discovered the intergranular corrosion of Ni-based alloys in molten (Li,Na,K)F after long time immersion.¹¹ Apart from Ni-based alloys, Kondo, et al. also observed selective corrosion in the grain boundary of SS304 and SS316L during the static corrosion test in molten salt LiF–BeF₂.¹² Hence, the corrosion behavior of structural materials may be influenced by the diffusion paths of elements through the grain boundaries. However, to our knowledge, the effects of microstructure on the corrosion behavior of structural materials in molten fluoride salt are scarcely investigated and are highly desirable.

The objective of this work is to investigate the influence of the microstructure on the corrosion behavior of structural materials in molten fluoride salts at 700 °C through the electrochemical and the oxidation tests. Therefore, some useful information may be obtained by investigation of the corrosion behavior of the Ni-based structural alloy in molten (Li,Na,K)F.

2. Experimental

2.1 Fabrication of the working electrode

The working electrodes used were monocrystal and polycrystal N5 superalloys, which were cut into specimens with the dimension of 8 mm × 9 mm × 2 mm (used for mass weight measurement) and 30 mm × 5 mm × 2 mm (used for electrochemical measurement), respectively, by an electric spark cutting machine, followed by grinding down to 1000 grit SiC paper, rinsing with distilled water and then drying. For the electrochemical measurements, a Fe–Cr wire was spotwelded to one end of the specimen for electrical connection. Then, the samples were sealed in an alumina tube with high-temperature cement, with a length of 15 mm exposure to the molten salt. Finally, the cement was dried at room temperature for 24 h and then further solidified at 200 °C for 12 h. The exposed surface of samples were polished again with 1000 grit SiC paper, rinsed, and dried before tests.

2.2 Electrochemical and oxidation measurements

After drying the analytical grade of LiF, NaF and KF (supplied by Sinopharm Chemical Reagent Co., Ltd, China) at 200 °C for 24 h, respectively, a mixture of (Li,Na,K)F was prepared and then put into a high-density graphite crucible. The mixed fluorides were further dried at 200 °C in the reaction chamber under vacuum for 48 h, and then the furnace was heated to 700 °C.^16,17 It was the fact that KF was hygroscopic and deliquescent and a certain concentration of water vapor would be carried by Ar flow, so the HF was generated at elevated temperatures by eqn (2.1):

KF + H₂O → K₂O + 2HF (2.1)

Therefore, the initial molten (Li,Na,K)F would contain a small amount of HF (in several ppm).

All electrochemical experiments were performed in a closed 304 stainless steel chamber under the protection of the pure Ar by a Princeton Applied Research PARSTAT 2273 potentiostat/Galvanostat system as reported in the previous study.¹⁸ A conventional three-electrode system was utilized, in which the reference electrode was a Pt electrode, the counter electrode was a high-density graphite plate (supplied by Sinosteel Shang-hai Advanced Graphite Materials Company, China) with the dimension of 30 mm × 15 mm, and the working electrodes were the monocrystal and polycrystal N5 superalloys as described in Section 2.1. The potentiodynamic polarization curves were measured at a scan rate of 20 mV min⁻¹ and at the temperature of 700 °C.

For the oxidation kinetics analysis, a high-density graphite crucible containing with 297 g salt was used. The samples were obtained by terminating the immersion test after varied durations from 0 to 100 h, then taken out of the melt, cooled in air down to the room temperature, cleaned in distilled water and alcohol, and weighed by a balance with an accuracy of 10⁻⁴ g. Each mass change was repeated three times, and the average values obtained were used for oxidation kinetics analysis.

2.3 Characterization

The cross-sectional and the surface morphology of the samples were characterized by Optical Microscope (OM) and scanning electron microscope (SEM) coupled with an energy dispersive X-ray spectrometer (EDX).

3. Results and discussion

3.1 Characterization of the pristine N5 superalloys

The material studied in this work is a N5 superalloy, the Ni-based monocrystal and polycrystal superalloys with around 3% Re and about 6.5% Ta. The chemical composition of the N5 superalloy is shown in Table 1. The microstructures of the monocrystal and the polycrystal N5 superalloys are illustrated in Fig. 1. By comparison with Fig. 1a and b, the grain boundary is observed in the polycrystal N5 superalloy. Meanwhile, the results reveal that the monocrystal and polycrystal N5 superalloys are composed of γ/γ′-phase with typical cubic structure as shown in Fig. 1c and d, where the γ phase is the matrix and the γ′-phase is the strengthening phase and formed by (Ni,Co)₃(Al,W,Cr) compound with irregular bulk structure and embedded in the fcc γ-Ni solid solution phase. In addition, it is obvious that the size of the γ/γ′-phase of the polycrystal N5 superalloy is much larger than that of the monocrystal as shown in Fig. 1c and d.

Table 1 The chemical composition of the N5 superalloy

Element	Co	Cr	Mo	W	Ta	Al	Re	Hf	Ni
Content/wt%	7.5	7	1.5	5	6.5	6.2	3	0.15	Bal.

---

Fig. 1 Microstructure with different magnifications of the N5 superalloys. (a) and (c) monocrystal (the insert shows the corresponding image at high magnification); (b) and (d) polycrystal.

3.2 Potentiodynamic polarization

Fig. 2 presents the change of free corrosion potential with exposure time for the monocrystal and the polycrystal N5 superalloys in molten (Li,Na,K)F at 700 °C. The results reveal that the E_corr for both of the monocrystal and the polycrystal maintains a value of about 300 mV vs. Pt, with a small fluctuation.

Fig. 2 Free corrosion potentials of the monocrystal and the polycrystal N5 superalloys in molten (Li,Na,K)F at 700 °C in an Ar gas environment.

The potentiodynamic polarization curves of the monocrystal and the polycrystal N5 superalloys in molten (Li,Na,K)F at 700 °C in an Ar gas environment are shown in Fig. 3. The results reveal that both of the monocrystal and the polycrystal are in active states at the corrosion potential. The kinetic parameters of the monocrystal and the polycrystal N5 superalloys in molten (Li,Na,K)F at 700 °C are obtained by Tafel extrapolation method and the results are shown in Table 2. It can be seen that the self-corrosion potentials of the monocrystal and the polycrystal N5 superalloys are about −327 and −328 mV vs. Pt, respectively. Moreover, the corrosion current density of polycrystal N5 superalloy is approximately 589.3 μA cm⁻², which is slightly higher than that of the monocrystal N5 superalloy as 585.0 μA cm⁻². In addition, the polarization resistance for the monocrystal is about 52.5 Ω cm², a little larger than that for the polycrystal with the value of about 47.2 Ω cm². Therefore, there is no big difference between the monocrystal and the polycrystal N5 superalloys under steady-state polarization condition in molten (Li,Na,K)F.

Fig. 3 Potentiodynamic polarization curves of the monocrystal and the polycrystal N5 superalloys in molten (Li,Na,K)F at 700 °C in an Ar gas environment at a scan rate of 20 mV min⁻¹.

Table 2 The kinetic parameters obtained by Tafel linear fitting of monocrystal and polycrystal N5 superalloys at 700 °C in molten (Li,Na,K)F

Alloy	ba/mV dec^−1	bc/mV dec^−1	Ecorr/mV vs. Pt	Icorr/μA cm^−2	Rp/Ω cm^2
Monocrystal	193.5	131.4	−327	585.0	52.5
Polycrystal	148.8	238.3	−328	589.3	47.2

---

3.3 Oxidation kinetics of the N5 superalloy

The representative oxidation kinetics plots for the corrosion of the monocrystal and the polycrystal N5 superalloys in molten (Li,Na,K)F at 700 °C in an Ar gas environment are presented in Fig. 4. The results reveal that both of the corrosion processes of the monocrystal and the polycrystal N5 superalloys include two stages. In the first stage, about 20 h, the N5 superalloy goes through a fast mass loss, while a steady increasing, approximately linear with the immersion time, of the mass loss is obtained after 20 h. Finally, the mass losses of the polycrystal and the monocrystal N5 superalloys reach to 16.995 and 11.373 mg cm⁻² for 100 h, respectively. This may be due to the polycrystal N5 superalloy possesses a lot of grain boundaries, as shown in Fig. 1b, where the deepest penetration and the first attack always occurs, and finally, the higher mass loss is obtained.

Fig. 4 The representative oxidation kinetics plots for the corrosion of the monocrystal and the polycrystal N5 superalloys in molten (Li,Na,K)F at 700 °C in an Ar gas environment.

3.4 Characterizations of the corroded N5 superalloys

Fig. 5 illustrates the surface morphologies of the monocrystal and the polycrystal N5 superalloys after corrosion in molten (Li,Na,K)F at 700 °C for 100 h in an Ar gas environment. It can be seen that both of the monocrystal and the polycrystal N5 superalloys experience general corrosion and are eroded by the molten (Li,Na,K)F. However, comparison with the monocrystal N5 superalloy, the larger amount of voids caused by the corrosion are observed on the surface of the polycrystal N5 superalloy, as shown in Fig. 5b and a, which means that the polycrystal N5 superalloy experiences more serious corrosion than that of the monocrystal. These results are in good agreement with the oxidation kinetics plots discussed in Section 3.3.

Fig. 5 Surface morphologies of the monocrystal (a) and polycrystal (b) N5 superalloys corroded in molten (Li,Na,K)F at 700 °C for 100 h in an Ar gas environment.

The cross-sectional morphologies of the monocrystal and the polycrystal N5 superalloys after immersion in molten (Li,Na,K)F at 700 °C for 100 h under pure Ar environment are illustrated in Fig. 6 and 7, respectively. It is obvious that the corrosion depth of polycrystal N5 superalloy is larger than that of the monocrystal after immersion in molten (Li,Na,K)F for 100 h (Fig. 6a and 7a). In addition, both of the monocrystal and the polycrystal N5 superalloys are eroded by selective dissolution of Cr and Al into the melt instead of aluminous and chromic oxide scale formation, which result in the non-uniform corrosion. Finally, a 125 μm Cr/Al-depleted zone with many voids is formed in the corrosion layer of the monocrystal N5 superalloy after 100 h immersion in molten (Li,Na,K)F (Fig. 6a and b). However, the polycrystal N5 superalloy experiences much more severe corrosion by forming a much deeper Cr/Al-depleted zone with a depth of around 175 μm with many internal voids (Fig. 7a and b).

Fig. 6 Cross-sectional morphologies (a and c) and EDX analysis (b) of the monocrystal N5 superalloy after immersion in molten (Li,Na,K)F at 700 °C for 100 h immersion in an Ar gas environment (d): the high resolution for the (a) etched by oxalic acid solution.

Fig. 7 Cross-sectional morphologies (a and c) and EDX analysis (b) of the polycrystal N5 superalloy after immersion in molten (Li,Na,K)F at 700 °C for 100 h under Ar environment (d): the high resolution for the (a) etched by oxalic acid solution.

It is clear that both of the monocrystal and the polycrystal N5 superalloys are eroded along the boundaries of the γ/γ′ phase as shown in Fig. 6c and 7c, respectively. And both the cubic γ′-phase in the monocrystal N5 superalloy and irregular bulk γ′-phase in the polycrystal N5 superalloy are eroded into cellular structure with some corrosion products in the voids (Fig. 6d and 7d). Coupled with the results of the EDX, it is found that the Ta-rich phase is still uncorroded (Fig. 6c and 7c), which is due to the fact that W and Ta are more stable than Cr, Al and Ni in molten (Li,Na,K)F.⁴ Moreover, the N5 superalloy undergoes more severe corrosion along the γ/Ta-rich phase boundary and form bigger holes than those in the normal corrosion layer.

In addition, the selective grain boundary corrosion and unconnected voids along the grain boundary are also observed after the polycrystal N5 superalloy immersion in molten (Li,Na,K)F at 700 °C for 20 h under Ar environment as shown in Fig. 8. The preferential corrosion along the grain boundary may be due to that grain boundaries are more chemically reactive than the grains.⁴ The results further confirm that the first attack always occurs at the grain boundary of the alloy.

Fig. 8 Microstructures of the polycrystal N5 superalloy after immersion in molten (Li,Na,K)F at 700 °C for 20 h under Ar environment.

3.5 Discussion

From the above results, it is clear that the corrosion behavior of the N5 superalloy in molten fluorides is significantly influenced by the microstructure of the alloy and the reasons can be discussed as following.

The corrosion of the monocrystal and the polycrystal N5 superalloys is mainly caused via a fluoridation process, which is induced by the dissolved HF in molten (Li,Na,K)F. Therefore, The elements of Cr, and Al in the N5 superalloy can form as fluorides, such as AlF₃ and CrF₂, and then dissolve in the molten (Li,Na,K)F due to these fluorides is more stable than HF according to Gibbs free energy.⁴ In addition, the potential–acidity diagrams for Ni, Co, Cr, and Al in molten (Li,Na,K)F (a(LiF) = 0.465) salt at 700 °C are calculated and shown in Fig. 9, based on the Nernst Equations and the thermodynamic data from HSC Chemistry version 6.0 database.† The calculate principal is referred to the potential–acidity diagrams in LiF–BeF₂ fluoride salt reported by Delpech.¹⁹ It illustrates that the stability domains of elements Ni, Co, Cr, and Al under their different forms are as a function of potential and oxoacidity (related to the oxide concentration in the salt) in a fluoride environment. From Fig. 9, we can deduce that the main elements of N5 superalloy, Ni and Co, possess more positive potential than Cr and Al. Therefore, the elements of Cr, and Al in the N5 superalloy, formed as AlF₃ and CrF₂, can be easily dissolved in the molten (Li,Na,K)F. Finally the elements of Co and Ni are relatively enriched because of the preferential dissolution of Al and Cr via fluoridation process.

Fig. 9 Potential-acidity diagrams calculated for the elements of Ni (a), Co (b), Cr (c), and Al (d) in molten (Li,Na,K)F (a(LiF) = 0.465) salt at 700 °C.

The general corrosion and boundary corrosion mechanisms during the immersion of the monocrystal and the polycrystal N5 superalloys in molten (Li,Na,K)F can be explained as following. For the general corrosion, the CrF₂ and AlF₃, which are caused by the fluoridation of alloying elements of Cr and Al, are dissolved into molten (Li,Na,K)F. At the same time, the phase (γ/Ta-rich phase and γ/γ′ phase) boundaries are selectively eroded because the Cr concentration at the boundary is much higher than the other parts due to the formation of chromic carbides.⁴ However, because of the grain boundary, a deeper depletion of Cr and Al and more voids are formed on the polycrystal N5 superalloy. Therefore, the grain boundary regions can be firstly penetrated by the molten (Li,Na,K)F, and then the Al and Cr are expected to be depleted at a much faster rate than Co and Ni. This may be the reason that the corrosion rate of the polycrystal N5 superalloy is much higher than that of the monocrystal as discussed in Section 3.3.

In addition, all of these results reveal that the phase and grain boundaries in the N5 superalloy are channels for the inward diffusion of the melt and the outward diffusion of Cr and Al ions. The different mass loss values between the monocrystal and the polycrystal implies that the corrosion resistance is definitely affected by the microstructure of the N5 superalloy.

Finally, according to Olson and Stretcher's report,^4,20,21 possible corrosion mechanisms for the severe corrosion in the phase boundary and grain boundary is considered to be the electrochemical process and summarized as following. A circuit of electrochemical cell is formed due to the potential difference between the anodic boundary and the cathodic phases and grains. Therefore, the Cr enriched boundary may be eroded more easily than the normal area which is subjected to the general corrosion. In addition, the fluoridation will be stronger for Al than that for Cr because the AlF₃ is more thermodynamically stable than CrF₂ as shown in Fig. 9c and d, respectively.

The above results clearly show that the corrosion resistance of the alloy is significantly influenced by the grain size of the structure alloy. Actually, the welding process may give rise to the difference in grain size of the welding area, weld heat affected zone and matrix. Their corrosion may be accelerated significantly with the change of grain size. Therefore, the influence of microstructure on corrosion, intergranular corrosion, is a great challenge for MSR using molten fluorides, and thus should be given special attention.

4. Conclusions

The corrosion resistance of Ni-based alloys in molten fluorides is significantly influenced by the microstructure of the alloy. The corrosion current densities of the monocrystal and the polycrystal N5 superalloys are almost the same while the mass loss of the polycrystal, calculated as 16.995 mg cm⁻², is much higher than that of the monocrystal calculated as 11.373 mg cm⁻² after immersion in molten (Li,Na,K)F for 100 h. In addition, the general corrosion and phase boundary corrosion are observed after immersion of the monocrystal N5 superalloy in molten (Li,Na,K)F at 700 °C. However, besides the general corrosion and phase boundary corrosion, the grain boundary corrosion is also observed, which may significantly influence the corrosion rate of the polycrystal N5 superalloy and lead to a faster corrosion rate than the monocrystal. The corrosion mechanism for the severe corrosion in the phase boundary and grain boundary is considered to that a galvanic cell is formed due to the potential difference between the Cr enriched anodic boundary and the cathodic phases and grains.

Acknowledgements

This work is supported by National Natural Science Foundation of China, Grant no. 51271190.

References

1. M. W. Rosenthal, P. R. Kasten and R. B. Briggs, ORNL, 1969.
2. M. Rosenthal, P. Kasten and R. Briggs, Nucl. Appl. Technol., 1970, 8, 107.
3. H. Macpherson, Nucl. Sci. Eng., 1985, 90, 374.
4. L. C. Olson, J. W. Ambrosek, K. Sridharan, M. H. Anderson and T. R. Allen, J. Fluorine Chem., 2009, 130, 67.
5. D. Williams, L. Toth and K. Clarno, ORNL/TM-2006/12, 2006.
6. M. W. Rosenthal, P. N. Haubenreich and R. B. Briggs, ORNL-4812, 1972.
7. J. W. Koger, Corrosion, 1973, 29, 115.
8. L. Olson, K. Sridharan, M. Anderson and T. Allen, Mater. High Temp., 2010, 27, 145.
9. L.C. Olson, U W-Madison, 2009.
10. G. Y. Lai, High-temperature Corrosion and Materials Applications, ASM International, Materials Park, Ohio, 2007.
11. B. You, Study on corrosion behavior of Nickel-based alloy in FliNaK molten salt, NTHU, 2010.
12. M. Kondo, T. Nagasaka, A. Sagara, N. Noda, T. Muroga, Q. Xu, M. Nagura, A. Suzuki and T. Terai, J. Nucl. Mater., 2009, 386–388, 685.
13. F. Y. Ouyang, C. H. Chang, B. C. You, T. K. Yeh and J. J. Kai, J. Nucl. Mater., 2013, 437, 201.
14. M. Kondo, T. Nagasaka, V. Tsisar, A. Sagara, T. Muroga, T. Watanabe, T. Oshima, Y. Yokoyama, H. Miyamoto and E. Nakamura, Fusion Eng. Des., 2010, 85, 1430.
15. L. Olson, K. Sridharan, M. Anderson and T. Allen, J. Nucl. Mater., 2011, 411, 51.
16. F. Motthiesen, E. Christensen, J. H. von Barner and N. J. Bjerrum, J. Electrochem. Soc., 1996, 143, 1793.
17. L. P. Polyakova, G. A. Bukatova, E. G. Polyakov, E. Christensen, J. H. von Barner and N. J. Bjerrum, J. Electrochem. Soc., 1996, 143, 3178.
18. Y. Wang, H. Liu and C. Zeng, J. Fluorine Chem., 2014, 165, 1.
19. S. Delpech, C. Cabet, C. Slim and G. S. Picard, Mater. Today, 2010, 13, 34.
20. M. Streicher, Corrosion, 1976, 32, 79.
21. E. M. Zahrani and A. M. Alfantazi, Corros. Sci., 2012, 65, 340.

---

Footnote

† HSC Chemistry is a commercial software including a thermochemical database, developed and sold by Outukumpu, Finland. See http://www.outokumpu.com/hsc for more details.

---