The importance of pyrochemistry is being increasingly acknowledged and becomes unavoidable in the nuclear field. Molten salts may be used for fuel processing and spent fuel recycling, for heat transfer, as a homogeneous fuel and as a breeder material in fusion systems. Fluorides that are stable at high temperature and under high neutron flux are especially promising. Analysis of several field cases reveals that corrosion in molten fluorides is essentially due to the oxidation of metals by uranium fluoride and/or oxidizing impurities. The thermodynamics of this process are discussed with an emphasis on understanding the mass transfer in the systems, selecting appropriate metallic materials and designing effective purification methods.

High temperature molten salts based on chloride or fluoride compounds have several applications in the nuclear field. In the front-end nuclear fuel cycle, molten salts are used for the purification and production of zirconium alloy, which is used as fuel cladding. Then, a pyrochemical treatment in NaCl-AlCl₃ molten salt at 350 °C enables the separation of zirconium and hafnium, which is a neutronic poison¹. In the nuclear fuel fabrication process, conversion of uranium oxide ore requires large quantities of fluorine that is obtained by the electrolysis of 2HF-KF molten salt at 95 °C². Several pyrochemical processes based on chloride or fluoride molten salts have also been conceived in the back-end nuclear fuel cycle, to separate actinides from lanthanides during nuclear waste recycling^{[3], [4], [5], [6], [7], [8] and [9]}. Because fluoride mixtures are thermodynamically stable at high temperature, with very high boiling points, these liquids have been considered as heat transfer or cooling fluids, as coolants for thermal energy^{[10] and [11]} and in nuclear fission and fusion systems. Several criteria have to be considered when choosing a structural material: mechanical strength at high temperature, irradiation resistance (in the case of materials under neutron flux) and chemical corrosion resistance (which depends on the material composition and microstructure, and on the physical chemistry of the molten salt). As it will be shown, in order to avoid corrosion the liquid fluoride salt coolant must be thermodynamically stable relative to the chosen materials.

If molten salts are already industrially used in the front-end nuclear fuel cycle or considered for alternative nuclear spent fuel recycling in the back-end fuel cycle, then the material development and the corrosion studies are essentially performed within the frame of the development of future nuclear reactors: Molten Salt Reactors (MSR), Advanced High Temperature Reactors (AHTR) and Tokamak fusion power plants. For all these cases, the selected molten salt is a fluoride salt mixture. Indeed, the material resistance is a key issue in all applications, but especially so in the case of reactor core use; not only because of the irradiation damage, but also because the operating temperature is determined by the fission reaction and cannot be decreased easily e.g. in case of pit formation. In the other applications the reactor can be cooled more easily.

The purpose of this paper is to give several causes of the chemical corrosion of materials in a fluoride salt and to propose, using a thermodynamical approach, some ways to prevent and control the chemical corrosion.

Description of the material/salt systems

Several fluoride molten salts have been investigated for nuclear applications:

(a) LiF-BeF₂ (66–33mol%) (FLiBe) was extensively studied in the 60s at Oak Ridge National Laboratory (USA) for the development of thermal neutron spectrum molten salt reactor systems (MSRE for Molten Salt Reactor Experiment and MSBR for Molten Salt Breeder Reactor)

^[12],

^[13],

^{[14] and}

^[15]. In these designs the main in-core material was graphite, and a Ni-based alloy (Hastelloy-N) was used for the out-of-core structures. The problems associated with graphite use are its low life time under irradiation (5 years)

¹⁶, a lack of recycling processes and no waste management.

(b) LiF-BeF₂ salt is also a candidate for blanket coolants in fusion nuclear power plants. Indeed, tritium can be produced by neutronic reaction with lithium

^[17],

^{[18] and}

^[19] and using LiF salt instead of metallic Li (which is highly reactive with water and oxygen) is more secure. Moreover FLiBe salt has a low tritium solubility and therefore low tritium inventory

¹⁸. Several corrosion studies of structural materials have been performed for this application

^[20],

^[21],

^[22],

^[23],

^[24],

^[25],

^[26],

^[27],

^[28],

^{[29] and}

^[30].

(c) The AHTR (Advanced High Temperature Reactor) concept is currently studied in the nuclear fission field. This concept includes a solid fuel and a solid core associated with a molten salt as coolant. Due to its high thermal stability (the reactor operating temperature is greater than 900°C), the selected salts for primary and secondary coolants are LiF-BeF₂ (66–34 mol%) and LiF-NaF-KF (46.5–11.5–42 mol%) (FLiNaK)

^[31],

^{[32] and}

^[33]. From a neutronic point of view, the use of LiF-BeF₂ is recommended for the primary coolant.

(d) Since 2000, an innovative design of MSR, the Molten Salt Fast Reactor (MSFR) has been studied by CNRS-France

^[34],

^[35],

^[36],

^{[37] and}

^[38]. The liquid fuel considered in this concept is LiF-ThF₄-UF₄ (77–20-3 mol%). Compared to the American MSRE and MSBR concepts, the MSFR operates with a fast neutronic spectrum. Under these conditions, the operating temperature of the reactor ranges between 650 and 850 °C, possibly higher than in the past American MSR systems. In this design, graphite is replaced by a metallic alloy.

In all these nuclear applications, the chemical resistance to the fluoride salts at high temperature is a key element in the choice of structural material. In the case of applications (b) and (d), the neutronic irradiation resistance must also be considered.

For the AHTR reactor system³³, several alloys were studied by immersing metallic samples in static FLiNaK salt at 850 °C for 500 hours: Haynes-230, Inconel-617, Hastelloy-N, Hastelloy-X, Nb-1Zr, Incoloy-800H, Ni-201 (nearly pure Ni). For the four Ni-based alloys considered, Hastelloy-N, Hastelloy-X, Inconel-617 and Haynes-230, the weight loss due to corrosion was found to increase with the Cr-content of the alloy. For Haynes-230, the significant chromium loss triggered dissolution/precipitation phenomena, especially the formation of W-rich precipitates at grain boundaries. The refractory alloy Nb-1Zr exhibited severe corrosion and embrittlement. Due to the remarkably high corrosion resistance of Ni in molten fluoride salts, Ni-201, a predominantly Ni-containing alloy with no Cr, was virtually immune to attack.

Concerning fusion applications^{[20], [21], [22], [23], [24], [25], [26], [27], [28], [29] and [30]}, the material should be chemically compatible with FLiBe, resistant to radiation damage and have inherently low activation to avoid excessive waste management at the end of plant life. No currently available material can meet all these requirements. From the stand point of corrosion resistance at high temperature, Ni-based alloys are the most resistant structural material to the fluoride salts. That is due to both the high value of the redox system NiF₂/Ni and the low solubility of NiF₂ in FLiBe salt³⁹. Nevertheless, high-nickel alloys are highly sensitive to neutron radiation damage and the activation products are longer-lived radioisotopes. The operating temperature of a fusion plant is lower than 650 °C and the chemical corrosion of steels and other metallic materials which meet low activation criteria may be slow enough at this temperature. Fe and V-based alloys containing Cr have been investigated in various molten salts^{[20], [21], [27] and [29]}. 304 and 316 stainless steels were down-selected^{[17], [18] and [24]}. Reactions in the fusion blanket lead to formation of tritium (T), for which the oxidation state depends on the melt redox potential: T^(+I)F or T₂⁽⁰⁾.

Many papers describe material development for nuclear reactor applications^{[40], [41], [42], [43], [44], [45], [46], [47], [48] and [49]}. Graphite presents an excellent compatibility with molten fluorides but cannot be used for structural applications. After preliminary tests, Cu, Mo, Nb, Ni and austenitic alloys were considered good candidates. However, Cu possesses insufficient mechanical resistance at high temperatures, and the forming of high Mo and Nb-alloys is complex. Corrosion tests have shown that Ni-based alloys were more resistant against chemical corrosion than the austenitic stainless alloys. For these two types of material, the chemical corrosion increases with the Cr content. Intensive corrosion of Inconel-600 was observed with void formation in the bulk due to the selective dissolution of Cr. The corrosion resistance of Hastelloy-B, without any Cr, is excellent, but it was ruled out due to inadequate forming possibilities⁴⁶ and poor oxidation resistance in air. Decreasing the amount of Mo helps for forming and machining, and adding a small quantity of Cr limits air oxidation. An optimum Cr content of 6 to 8wt% was retained. Chemical composition of the MSRE structural alloy, Inor 8 or Hastelloy-N, followed these specifications. The dissolution rate of this alloy in fluoride molten salt at 700 °C was observed to be lower than 2.5 µm/year. However, experience from the MSRE has found that Ni-based alloys are embrittled under neutronic irradiation. Helium is produced (essentially) by the nuclear reaction, ⁵⁸Ni (n,α) ⁵⁵Fe, and nucleated as discrete voids within the alloy matrix. Nb and Ti are added into the metallic alloy to form intergranular carbides which are likely to trap helium atoms and prevent their diffusion along grain boundaries. The quantities of Ti and Nb should remain small to prevent the formation of Ni₃(Ti, Nb) phases. The carbides stability depends on the temperature: at 650 °C, 0.5% Ti+Nb is sufficient while at 700 °C, 2% is necessary⁴⁶.

Alloys strengthened by the addition of tungsten, to replace molybdenum, are under development^{[38] and [50]}. These alloys boast an increased mechanical resistance at high temperature, as previously described³⁸, while keeping the appropriate compatibility with molten fluorides. Hence such materials may allow higher operating temperatures (above 850 °C). Fabrication and analysis of W-rich alloys are ongoing in the frame of the European research project EVOL (EURATOM-FP7). Moreover, and as will be observed in the following, tungsten presents a higher chemical resistance against corrosion than molybdenum.

Main causes of chemical corrosion in fluoride molten salts

The chemical corrosion^{[41], [42] and [51]} is due to the oxidation of a metal or an alloy in contact with its environment. Fig. 1 presents a schematic of the interface between the reacting element from the structural material and the molten salt medium.

The oxidizing species contained in the initial melt

In a fluoride melt, the main oxidizing impurities are said to be H₂O (in general this is present in the solid constituents to be fused) and HF (formed in the melt by several reactions):

Hydrolysis reaction during melt fusion

(1)

The hydrolysis temperature (given for a Gibbs energy equal to 0 for Eq. 1) of selected pure salts was calculated from the thermochemical data of pure compounds⁵² (Table 1). Generally, the salt mixture is fused at a lower temperature than the hydrolysis temperature of the pure salt. Therefore, the solvation process increases the metallic fluoride stability and strongly reduces its hydrolysis process (less than 10 ppm for FLiNaK)^{[29] and [30]}.

Dissolution of water

H₂O(g) is easily eliminated under argon flux at temperatures above 300 °C. Moreover, traces of water are not stable in the fluoride salt after melting, as the water reacts with the fluoride ions to produce HF(g) and oxides.

Melt purification towards oxides

Although oxide ions^{[53] and [54]} have no direct influence on the structural material corrosion, because O^2- is not oxidizing towards metals, the elimination of oxides is nevertheless required to prevent the precipitation of metallic oxides (especially UO₂). O^2- ions react with an excess of HF according to the reaction:

(2)O^2-+2HF(g)→H₂O(g)+2F^-

It was suggested that this purification could be performed by bubbling pure gaseous HF or HF/H₂ mixtures into the salt. Due to its high solubility in the fluoride salts, HF(g) is strongly retained in the salt^{[55] and [56]} and is responsible for structural material corrosion as observed in corrosion tests^{[20], [21], [22], [23], [25], [26], [27], [28], [29] and [30]}. Therefore, the use of pure HF(g) is not recommended for salt purification because it will result in the severe corrosion of the structural material. This corrosion could be reduced by using a mixture of HF(g)/H₂(g), because the presence of H₂(g) will decrease the redox potential of the salt, and the H₂(g) can react with the oxidizing impurities of the salt.

Nevertheless, a purification procedure of molten salt with no introduction of pure HF(g) is strongly recommended to prevent corrosion reactions. The fluoride mixture could be dried under high vacuum prior to melting. It was observed that after a vacuum treatment (10⁻⁵ torr) at 400–425°C the cathodic and anodic residual currents (due to O^2- oxidation or HF-H₂O reduction) were smaller than those measured by other authors in the same melt pretreated with HF and H₂^{[42], [57] and [58]}. The elimination and purification of salts for removing oxides and sulfurs can also be performed by gas mixture bubbling. In this case the HF/H₂ gas composition has to be carefully chosen to prevent an excess of HF and an increase of salt redox potential.

The oxidizing species related to the fission reaction

In the case of MSR, the molten salt chemical composition varies with the fission reaction and therefore with the operating time. In particular, it was assessed that the redox potential of the fuel salt will increase while in service. The fissile element in the MSFR reactor system is ²³³U which is present with two oxidation states in the salt, (IV) and (III), as UF₄ and UF₃ respectively. Then, the redox potential of the fuel salt depends on the x(UF₄)/x(UF₃) mole fraction ratio according to the Nernst relation:

(3)

where R is the ideal gas constant (J/mol/K), F the Faraday constant (C), T the temperature (K) and E° the apparent potential of the UF₄/UF₃ redox system which combines the standard potential of UF₄/UF₃ and the activity coefficients of UF₄ and UF₃.

When the fission reactions occur, the fission products are essentially LnF₃ lanthanides (with oxidation state (III)) and gaseous products or noble metals (M) (with oxidation state 0). The result of the fission reactions on the salt chemistry can be schematized by³⁸:

(4)UF₃+n→LnF₃+M

(5)UF₃+n→LnF₃+M+1/2F₂(g)

When the fission reaction (Eq. 5) occurs, gaseous fluorine is produced. Fluorine being the most oxidizing species, the salt redox potential increases according to:

(6)UF₃+1/2F₂(g)→UF₄

Consumption of UF₃ by (4) and (6) leads to an increase of the redox potential according to Eq. 3. The fluorine production, and consequently the fuel potential, strongly enhances the corrosion of the structural materials during the reactor's operating time. As it will be shown below, addition of a reducing agent is required to curb the increase of the salt potential and thus to limit corrosion.

Some fission products such as S, Se and Te are also harmful to the metallic structures due to their oxidizing properties; specifically, due to their negative oxidation states ((-I) and (-II)). In particular, corrosion tests have shown that a low concentration of metallic tellurium is strongly corrosive and causes intergranular attacks on Ni-alloys resulting in severe embrittlement^{[13], [47] and [59]} (Fig. 2). Analysis of stability phase diagrams calculated for tellurium in fluoride salts, combined with experimental results, showed that the deleterious effect of Te can be cancelled by applying a cathodic polarization to the material⁶⁰. In this case, tellurium, electrochemically reduced to Te(-II), reacts with ZrF₄ (salt constituent) to form ZrTe₂.