Introduction

A comprehensive body of literature exists on the study of corrosion of alloys in molten chlorides [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19], The focus of corrosion studies has been primarily concerned with Fe-based and Ni-based alloys, i.e., stainless steels versus nickel super alloys. These studies often experimentally evaluate depletion of individual elements in super alloys. However, Ellingham diagrams, often overlooked in the literature, are a useful tool for predicting behavior of alloying elements. In combination with chloride-oxide stability diagrams, Ellingham diagrams can explain much of the experimentally observed behavior of alloys is chloride salts. Typically, super alloys exhibit high corrosion resistance due to the presence of a passivating oxide layer such as chromium oxide. However, in molten chloride salts Cl− ions challenge this oxide layer, and expose the alloying constituents to further oxidation, creating what has been termed the ”chlorine-oxidation cycle” [1,2,3,4,5,6,7,8,9,10], For all alloys corrosion follows the same stages of (1) oxidation of elements in alloy, followed by (2) dissolution or vaporization of oxidized elements. The rate of corrosion has therefore been strongly correlated to the presence of oxidizing impurities in the chloride salt [3, 5,6,7,8,9,10,11, 13,14,15,16, 18,19,20,21,22,23]. Understanding these studies using Ellingham diagrams and chloride-oxide stability diagrams enables predictive insight into alloy performance. Furthermore, it underscores the importance of limiting corrosive impurities via control of salt exposure to water, oxygen, and other oxidizing agents.

Results and discussion

Construction of Ellingham diagrams

The utility of Ellingham diagrams in evaluating molten chloride salts systems can be illustrated in the example case of MgCl2. The formation of corrosive impurities primarily stem from the hygroscopic nature of MgCl2 resulting in the formation of oxide/hydroxide species in the presence of oxygen and moisture via the following reactions:

$${\text{MgCl}}_{{{2}({\text{l}})}} + {\text{H}}_{{2}} {\text{O}}_{{({\text{g}})}} \leftrightarrow {\text{MgOHCl}}_{{({\text{l}})}} + {\text{HCl}}_{{({\text{g}})}}$$
(1)
$${\text{O}}_{{{2}({\text{g}})}} + {\text{4Cl}}_{{({\text{l}})}}^{ - } \leftrightarrow {\text{2O}}_{{({\text{l}})}}^{2 - } + {\text{2Cl}}_{{{2}({\text{g}})}}$$
(2)

Furthermore, MgOHCl has been found to decompose above 550°C according

$${\text{MgOHCl}}_{{({\text{l}})}} \leftrightarrow {\text{MgO}}_{{({\text{l}})}} + {\text{HCl}}_{{({\text{g}})}}$$
(3)

nb

Several studies have found a direct correlation between corrosion rates and concentration of MgOHCl present in the chloride salt [20, 23], Furthermore, the formation of HCl(g) and Cl2(g) via leads to corrosion of alloys in the headspace [7, 23, 24]. The corrosive impurities lead to degradation reactions with alloying constituents M (e.g., Cr, Fe, Ni) according to:

$$x{\text{HCl}}_{{({\text{g}})}} + {\text{M}} \leftrightarrow {\text{MCl}}_{x} + {1}/{2}x{\text{H}}_{{{2}({\text{g}})}}$$
(4)
$$x{\text{MgOHCl}}_{{({\text{l}})}} + {\text{M}} \leftrightarrow x{\text{MgO}}_{{({\text{l}})}} + {\text{MCl}}_{x} + {1}/{2}x{\text{H}}_{{{2}({\text{g}})}}$$
(5)
$${1}/{2}x{\text{O}}_{{{2}({\text{g}})}} + y{\text{M}} \leftrightarrow {\text{M}}_{y} {\text{O}}_{x}$$
(6)

This degradation is an electrochemical process that can be explained via half-cell reactions as follows [9, 25]:

Anodic oxidation of alloying element M:

$${\text{M}} \to {\text{M}}^{n + } + ne^{ - }$$
(7)

where n is the number of electrons exchanged.

Cathodic reduction of corrosive species:

$$ox + ne^{ - } \to red$$
(8)

where Ox is the oxidizing impurity in this case (e.g. MgOHCl), and Red is the reduced form of the oxidizing impurity.

The complete redox couple reaction:

$${\text{M}} + ox \, \leftarrow \to {\text{M}}^{n + } + red$$
(9)

For the electrochemical process to occur spontaneously the change in Gibbs-free energy needs to be negative, which can be calculated via:

$$\Delta G_{rxn} = - nF\Delta E_{rxn}$$
(10)

where ∆Grxn is the change in Gibbs-free energy of reaction (J mol−1), F is Faraday’s constant (96,485 C mol−1), and ∆Erxn is the redox potential of reaction (V ).

Equation 10 allows for the construction of an Ellingham diagram, shown in Figure 1 (constructed for various alloying constituents using the thermodynamic software HSC v8).

Fig. 1
figure 1

Gibbs-free energy of reaction of metal to metal-chloride as a function of temperature

Full size image

According to Figure 1, the most stable chlorides are K, Na, and Mg. Therefore, common alloying constituents (e.g. Fe, Cr, Ni) should theoretically remain stable within the chloride salt. However, the influence of oxidizing impurities such as MgOHCl is not readily captured by simply looking at Equation 10. To assess the effect of oxidizing impurities, ∆Erxn can be further expressed into the anodic and cathodic potentials via the Nernst Equation:

$$\Delta {E}_{rxn}={E}_{c}-{E}_{a}$$
(11)
$${E}_{a}={E}_{a}^{0}+\frac{RT}{nF}\mathrm{ln}\left(\frac{{a}_{{M}^{n+}}}{{a}_{M}}\right)$$
(12)
$${E}_{c}={E}_{c}^{0}+\frac{RT}{nF}\mathrm{ln}\left(\frac{{a}_{Ox}}{{a}_{red}}\right)$$
(13)

where E0 is the potential under standard conditions (V ), R is the ideal gas law constant (8.314 J mol−1 K−1), T is temperature (K), and a is the activity, which for a pure solid is unity, i.e., aM = 1. From equation 13, the effect of increasing the concentration of oxidizing impurities (increasing aOx) leads to increasing Ec — thereby increasing ∆Erxn, and results in a more negative ∆Grxn. So far, the mathematical treatment around the Ellingham diagram is useful in predicting some experimental trends, such as the rate of depletion of certain alloying constituents. The diagram correctly predicts the rate of depletion for Mn > Cr > Fe > Ni [5, 11]. However, the diagram incorrectly predicts the rate of depletion of Nb, Mo, and W. Several studies have suggested the presence of these alloying constituents to slow down the rate of corrosion [10, 26].

Construction of chloride-oxide stability diagrams

The construction of a chloride-oxide stability diagram elucidates more information regarding the thermodynamic behavior of alloying constituents in molten chloride salts [10]. Such a diagram was constructed by calculating the ∆Grxn of the oxides versus chlorides of various alloying constituents via HSC v8 (Fig. 2.).

Fig. 2
figure 2

Chloride-oxide stability diagram of various alloying constituents at 500°C

Full size image

Figure 2 has three primary regions:

  1. i.

    Lower half representing oxide species are more stable than chloride species.

  2. ii.

    Upper half representing chloride species are more stable than oxide species.

  3. iii.

    Along the parity line, representing both oxide and chloride species in equilibrium with each other.

Several experimental observations can be explained via the chloride-oxide stability diagram. For example, K and Na chlorides are highly resistant to oxidation and can be considered stable in the presence of oxygen and moisture, whereas Mg is not [25]. Additionally, upon oxidation of Mn, Cr, Fe, Co, and Ni, the oxide will equilibrate with chloride ions and form chlorides as suggested by the chlorine-oxidation cycle. Lastly, alloys containing W, Mo, Al and Nb have enhanced corrosion resistance due to the formation of a relatively stable oxide that can serve as a passivation layer [10]. The utility of the chloride-oxide stability diagram elucidates several experimental observations that the typical Ellingham diagram overlooks. The diagram highlights that even Ni itself falls victim to the chlorine-oxidation cycle, as was observed by Liu et al. [5]. Even commercially pure Ni (e.g. Ni-201), which is expected to provide superior corrosion resistance, has been observed to corrode in chloride salt to a point of failure within days [26].

Combined use of Ellingham diagrams and chloride-oxide stability diagrams as predictive tools

Chloride-oxide stability diagrams and Ellingham diagrams are useful tools in evaluating corrosion and interpreting results, even in less well studied systems such as convective molten chloride systems. Though studies under these conditions are limited, notable examples include forced convection studies dating back to 1960s at the Brookhaven National Laboratory [12] and natural convection studies that were conducted in early 2010 between Idaho National Laboratory and the University of Wisconsin Madison [26]. We find that these examples are well explained by Chloride-oxide stability diagrams and Ellingham diagrams.

Natural convection corrosion studies elucidated time-dependent corrosion mechanisms [26, 27]. In the initial stages, corrosion is primarily driven by oxidative impurities, which can be understood using chloride-oxide stability analysis. Once the concentration of these impurities diminished, the corrosion was dominated by active dissolution of selective alloying constituents (e.g., Cr) from the hot side, and deposited on the cold site. The studies highlighted the effect of temperature-dependent metal solubilities of species such as chromium chloride. Similar observations were observed in the static corrosion study conducted by Gong et al. [20]. Under static conditions, the first 250 hours was primarily impurity-driven corrosion, after which the dominant mechanism became thermal effects resulting in metal solubility differences.

Surprisingly, forced convection corrosion studies conducted at Brookhaven National Laboratory observed no appreciable change in corrosion rates compared to static conditions [12]. The work highlighted the importance of salt purification and designing a leak tight system with an inert atmosphere. Therefore, an effective purification strategy can minimize corrosion by limiting the impurity-driven corrosion mechanism.

Conclusion

Corrosion remains a significant problem in molten chloride salts systems, but we propose that corrosion behavior of specific alloys can be predicted, and that experimental corrosion evaluation results can be understood, using a combination of Ellingham diagrams and chloride-oxide stability analysis. This method correlates corrosion behavior to fundamental thermodynamics and can be used to identify and explain performance of specific promising alloying elements. Thus, it can be used to identify high-performance alloys and to guide materials development toward appropriate alloys for use in molten chloride salts, which could in turn enable advances in chloride salt based CSP, nuclear, thermal energy storage, and other applications.