Elucidating the Corrosion Mechanism of Graphite Anodes in Fluoride Molten Salt for Rare Earth Electrolysis: A Multiscale Structure-Property Investigation

Abstract

Graphite anodes are widely used as consumable electrodes in the high-temperature electrolytic production of rare earth metals within fluoride molten salts. However, their rapid and complex corrosion presents significant economic and operational challenges, including high consumption costs, process instability, greenhouse gas emissions, and product contamination. While the corrosion morphology of specific graphite types has been studied, a systematic investigation linking the intrinsic properties of diverse graphite materials to their microstructural and chemical evolution during corrosion is lacking. This study elucidates the corrosion mechanisms of three distinct graphite anodes—fine-grained, isostatically pressed graphite anodes (#1), medium-coarse-grained, extruded graphite anodes (#2), and recycled, extruded graphite anodes (#3) in industrial PrNdF₃–LiF molten salt electrolytes at 1050 °C. Through a multifaceted analytical approach encompassing SEM, EDS, XRD, Raman, and FT-IR, we investigated the macro- and microscale corrosion behaviors across multiple scales. The results revealed markedly different degradation patterns: the #1 anode exhibited intergranular corrosion with granular exfoliation; the #2 anode developed a protective but cracked resolidified salt layer; and the #3 anode suffered the most severe uniform and pitting corrosion. Postcorrosion analysis confirmed surface enrichment with fluorine, praseodymium, and neodymium, the formation of PrF₃ and NdF₃ phases, and substantial degradation of the graphitic structure. Raman spectroscopy specifically revealed a reduction in the crystallite size, introduction of in-plane point defects, and disruption of the interlayer stacking order. On the base of infrared spectroscopy analysis, all key characteristic absorption peaks of the graphite anodes undergo consistent attenuation after corrosion. This work provides critical insights for the informed selection and optimization of graphite anodes to increase the efficiency and sustainability of rare earth electrolysis.

1. Introduction

Molten salt electrolysis technology has been extensively used in various metal smelting processes, including the Hall-Héroult process for aluminium and the electrolytic production of magnesium. Material selection is critical for this technology, as the structural components and anodes must exhibit excellent corrosion resistance at high operating temperatures. Graphite materials possess excellent electrical conductivity, high-temperature resistance, and ease of processing. In processes such as aluminium smelting, graphite serves as a consumable anode, where it reacts with oxygen ions to form carbon dioxide. This reaction occurs at a lower overpotential than the evolution of oxygen gas, making the process more energy efficient than the use of inert anodes. This capability enables the electrolysis process to proceed smoothly and continuously.

Various graphite materials have been extensively studied and utilized on the basis of the characteristics of different molten salt systems. In molten salt reactors, nuclear graphite serves as both a neutron reflector and moderator, making direct contact with molten salt. Its compatibility and corrosion behavior are crucial. Zhang et al. [1] studied the microstructure evolution and molten salt penetration characteristics of the fine-grained isotropic graphite ZXF-5Q produced by Poco Inc. in molten eutectic fluoride salts. Additionally, research shows that the molten fluoride salt FLiBe can react with graphite in both liquid and gaseous environments, resulting in the formation of covalent bonds and semi-ionic carbon-fluorine bonds on the graphite surface, which alters the surface microstructure [2]. In contrast, research on graphite compatibility extends to other high-temperature fluids. For example, He et al. [3] compared the permeation behavior and compatibility of six different commercial nuclear graphite structures at 800 °C in molten lead-bismuth eutectic (LBE). Pyrolytic graphite, a highly oriented, dense, and crystalline material, exhibits excellent oxidation and corrosion resistance, along with favorable mechanical, thermal, and electrical properties. Consequently, it is recommended for use as a structural material in high-temperature molten LiCl-KCl environments and is suitable for the high-temperature chemical posttreatment of metal fuels [4]. Hareesh et al. [5] further assessed the corrosion resistance of high-density oriented pyrolytic graphite as an electrode for the direct electrochemical reduction of metal oxides in molten LiCl with 1% Li₂O. In a different context, flake graphite offers several advantages, including high thermal conductivity, low thermal expansion, excellent thermal stability, and cost-effectiveness, making it suitable as a high thermal conductivity additive in molten salt composites. For example, partially exfoliating flake graphite into graphene nanosheets within a molten salt matrix can significantly enhance the thermal conductivity of a composite system [6]. Graphite materials have also been extensively studied for their applications in solar thermal storage. Xu et al. [7] systematically examined the corrosion behavior and failure mechanism of graphite produced from four distinct raw materials and preparation processes in molten nitrate.

Rare earth elements are vital but limited mineral resources and are often referred to as “the vitamins of modern industry” [8,9]. Rare earth metals possess unique optical, electrical, magnetic, and catalytic properties, making them essential for modern high-tech industries and the enhancement of traditional sectors. Specifically, they are key constituents of neodymium-iron-boron permanent magnets and are extensively utilized in advanced applications, including new energy vehicle motors, wind turbines, and consumer electronics. Owing to their roles as phosphors and laser materials, rare earth metals are broadly employed in energy-efficient lighting, display technologies, medical devices, and optoelectronic communication systems. Within the petrochemical industry, they function as highly efficient molecular sieve catalysts, greatly improving fuel efficiency and assisting in the purification of vehicle emissions. Additionally, they are vital for enhancing the properties of special alloys, such as heat and corrosion resistance. As additives and polishing powders, they are essential in the glass, ceramics, agriculture, and electronics industries [10,11,12]. They are essential for transforming and upgrading traditional industries, fostering strategic emerging sectors, and advancing national science and technology. Among the methods for preparing rare earth metals, fluoride molten salt electrolysis is considered the most effective. For example, more than 90% of metallic neodymium is produced via this process [13,14].

A typical rare earth electrolysis cell features a graphite crucible that serves as the cell lining, multiple graphite anodes, a vertically suspended tungsten cathode, and a metal collection tray. The crucible is filled with a fluoride-based molten salt electrolyte. The graphite anodes are immersed in this electrolyte. The cell’s operating temperature is maintained primarily by the resistive heat generated from the direct current passing between the anodes and the cathode. The molten salt acts not only as the reaction medium but also as the source of rare earth ions, which are formed through the dissolution of rare earth oxides [15,16]. Industrial data indicates that the consumption of graphite is 200 kg per ton of rare earth metals produced [17]. Therefore, a thorough understanding of the corrosion behavior of graphite materials in these fluoride molten salt electrolytes is crucial for improving process efficiency and material sustainability.

The widespread use of consumable graphite anodes in rare earth molten salt electrolysis presents a multifaceted challenge, impacting economic viability, operational stability, environmental compliance, and final product quality. Economically, graphite consumption constitutes a major direct cost, severely eroding corporate profits. Operationally, the rapid consumption of anodes necessitates frequent replacement, which requires a full cell shutdown. This initiates a costly cycle of cooling, physical replacement, and subsequent reheating to operational temperatures (typically >1050 °C). This “shutdown-replacement-restart” process consumes substantial energy and time, drastically increases overall energy consumption, requires intensive manual labor, and introduces significant safety hazards. From an environmental standpoint, the anode reactions are a primary concern. The oxidation of carbon by oxygen ions generates a mixture of CO and CO₂ [18]. Moreover, at elevated temperatures, graphite can react with fluoride ions in the electrolyte, generating potent greenhouse gases such as perfluorocarbons [19,20]. The anode off-gas may also contain volatile fluorides, posing air pollution and health risks within the workshop. In the context of global “dual carbon” goals (carbon peak and carbon neutrality), this high carbon footprint places considerable environmental compliance pressure and policy risk on producers [21,22]. The dynamic consumption of the anode also severely complicates process control. The continuous shrinkage of the anode alters the effective reaction area and current density, whereas increasing the anode-cathode distance (electrode pitch) destabilizes the cell voltage and thermal balance. This erratic consumption makes stable process control challenging and necessitates constant voltage and electrode pitch adjustments, which in turn lead to fluctuations in product composition and challenges in maintaining consistency across production batches [23,24]. Finally, anode degradation directly compromises product quality. The anodes are prone to spalling and corrosion due to weakened structural integrity at high temperatures, causing carbon particles to flake off into the electrolyte. These carbon impurities, introduced via anode corrosion, contaminate the molten rare earth metal. This is particularly detrimental for high-end applications such as neodymium-iron-boron permanent magnets, where even minor impurities can severely degrade magnetic performance [25,26,27]. Therefore, a fundamental study on the corrosion behavior of graphite anodes in fluoride molten salt electrolytes is essential to mitigate these interconnected issues, enhance product quality, and achieve superior process control.

Although previous studies have utilized experimental and finite element simulations to investigate the corrosion of graphite anodes in rare earth electrolysis systems, the majority have focused primarily on morphological changes in a specific type of graphite material [15,27]. Traditional single-scale studies frequently yield incomplete insights. A thorough understanding of the corrosion process in fluoride molten salt electrolytes requires a multiscale structure-property approach. Corrosion initiation and propagation occur over multiple length scales, ranging from macroscopic material loss and crack formation (millimeter to micrometer scale) to microscopic chemical erosion, crystal disorder, and alterations in functional groups. Studies of materials demonstrate that geometric multiscale analysis is essential for integrating observations across different scales, thus linking macroscopic properties with microscopic structural evolution [28,29].

Therefore, we employ a multiscale structure-property investigation that combines macroscopic imaging (e.g., SEM for micrometer-scale morphology), crystallographic analysis (e.g., XRD for assessing long-range order at the nanoscale), and molecular-level spectroscopy (e.g., Raman and FT-IR for analyzing atomic-scale defects and chemical bonds) to elucidate the complex corrosion mechanisms. To date, the evolution of macro- and microscale morphology in structurally diverse graphite materials, as well as the phase composition of corrosion products and alterations in surface functional groups during corrosion, has not been systematically reported. Additionally, it remains unclear whether the degradation of graphite anodes in fluoride molten salt occurs uniformly across the entire open-pore surface through gas phase reactions or is predominantly localized at the salt-graphite interface. To address these questions, this study investigates three distinct graphite anodes—fine-structured, medium-coarse, and recycled—after their exposure to fluoride molten salt at 1050 °C. By employing scanning electron microscopy, X-ray diffraction (XRD), Raman spectroscopy, and infrared spectroscopy, we evaluated both morphological changes and the evolution of microstructures and functional groups. These analyses elucidate the underlying corrosion mechanism.

2. Materials and Methods

2.1. Materials

Three types of graphite anodes prepared from different raw materials were provided by Fangda Carbon New Materials Technology Co., Ltd. (Lanzhou, China). The anode was processed into an arc-shaped structure, with an inner ring radius of approximately 125 mm, an outer ring radius of approximately 215 mm, a height of approximately 500 mm, and a single-piece weight of approximately 20 kg. The basic performance parameters of the graphite anodes are listed in

Table 1. Basic properties of the graphite anodes.

Anodes	Bulk Density (g/cm3)	Compressive Strength (MPa)	Thermal Conductivity (Room Temperature) (W/(m·k))	Electrical Resistivity (μΩ·m)	Apparent Porosity (%)	Thermal Expansion Coefficient (×10−6/°C)
#1- Fine-grained, isostatically pressing graphite anode	1.73	34.87	122.42	11.02	17.12	1.55
#2- Medium to coarse particle, extruded graphite anode	1.74	33.26	150.89	8.17	14.24	1.56
#3- Recycled graphite, extruded graphite anode	1.64	43.10	33.55	14.33	20.03	2.98

.

2.2. Corrosion Test

The experiment was performed in an industrial 8 kA open-type electrolytic cell operating continuously with an inserted cathode. The key components of the setup consisted of a cylindrical graphite crucible, a four-piece arc-shaped graphite anode assembly, a tungsten cathode rod aligned along the central axis, and a metal collector positioned at the crucible base.

Electrolysis was carried out in a molten salt system comprising praseodymium-neodymium fluoride and lithium fluoride at a mass ratio of 9:1. The process maintained a current of 6900 ± 100 A and a molten salt temperature of 1050 ± 30 °C. The current density was set from 5–6 A/cm² at the cathode and 1.7–1.9 A/cm² at the anode. The corrosion experiment was conducted on ten samples of each anode type, with a duration of 48 h per sample.

2.3. Analysis and Testing Methods

Due to the strong adhesion of the resolidified salt layer to the corroded graphite substrate, which poses a significant risk of damaging the fragile corroded structure during mechanical removal, accurate post-experiment mass measurements for corrosion rate calculation are not feasible. Therefore, the corrosion assessment primarily relies on morphological, microstructural, and chemical evolution.

Macroscopic and microscopic characterization: Photographs of the entire graphite anode were taken before and after corrosion to document changes in appearance and shape. Photographs were captured via a cell phone (Oppo Reno5, Dongguan, China).

The microstructure and corrosion layer of the graphite anode before and after corrosion were observed via SEM, and the surface elements were analyzed via EDS. The working voltage of the SEM-EDS instrument (SEM5000, Guoyi Quantum Technology (Hefei) Co., Ltd., Hefei, China) was 5 kV, the working distance was approximately 13–15 mm, and the exposure time was 4 µs. Owing to the inherent microstructural heterogeneity of the graphite samples, the SEM and EDS results from specific locations are representative rather than quantitative.

XRD analysis: XRD patterns of the powdered graphite anodes were obtained via a powder X-ray diffractometer (D8 ADVANCE, Bruker, Rheinstetten, Germany) with a step size of 0.02°, a scanning rate of 10°/min, and a scan range from 20° to 80°. The XRD patterns were segmented and fitted via MDI Jade 6.5 software to determine the peak position and peak height.

Raman Spectroscopy Analysis: Raman spectra of the powdered graphite were obtained via a Raman spectrometer (Renishaw in Via Reflex, Gloucestershire, UK). The test parameters were as follows: 532 nm semiconductor laser, spectral scanning range of 1000–3500 cm⁻¹, and resolution of 1 cm⁻¹.

FT-IR analysis: The infrared characteristic peaks of the powdered graphite anodes before and after the experiment were determined via a Fourier transform infrared absorption spectrometer (Perkin Elmer, Inc., Waltham, MA, USA). The test conditions were as follows: attenuated total reflection (ATR) mode, scanning range 1000–4000 cm⁻¹, resolution 1 cm⁻¹.

For comparison, the graphite powder of the precorrosion sample originated from the intact graphite anode, whereas that of the postcorrosion sample was derived from the corrosion layer.

2.4. Statistical Analysis

All the data are presented as the mean ± standard deviation (SD) of more than three experiments, and the number of corrosion tests for each type of graphite anode was 10 times. All the statistical results presented were analyzed via one-way analysis of variance via Origin 2021 software.

3. Results

3.1. Changes in Corrosion Morphology

Figure 1 illustrates the markedly different corrosion behaviors of the three graphite anodes in fluoride molten salt, which are attributable to their distinct structures, grain sizes, preparation processes, and origins (virgin or recycled). The overall surface of the original graphite anode is relatively flat and intact, with clear corner contours and a complete structure. After corrosion analysis, a spectrum of corrosion morphologies was observed, ranging from slight surface roughening to complete structural disintegration.

The #1 anode, with its original relatively flat and intact surface and sharp edges, was partially retained, indicating a certain degree of resistance to the corrosive melt. However, higher magnification images revealed the onset of nonuniform structural degradation at the microscale. This manifests initially as granular exfoliation, where fine particles detach from the bulk, leading to measurable increases in surface looseness and porosity. Furthermore, the formation of interstitial pores and corrosive debris wedged within the initially compact grain boundaries signifies the penetration of the molten salt along the grain boundaries, selectively attacking the weaker interfaces and initiating a granular disintegration process. In stark contrast, the #2 anode developed a radically distinct surface morphology, being entirely encapsulated by a continuous, monolithic layer that EDS analysis identified as resolidified fluoride salt. This layer formed through intense wetting and subsequent cooling, where the thermal contraction coefficient mismatch between the salt and the graphite substrate generated significant internal stresses. These stresses are relieved through a network of numerous thermal-stress-induced microcracks that traverse the salt layer. The presence of this brittle, adherent salt layer has a complex dual effect on the morphological evolution of the underlying anode. It could act as a physical barrier, temporarily shielding the graphite surface from direct contact with the molten salt. Conversely, the pervasive and interconnected cracking network likely serves as direct capillaries for the continued infiltration of the corrosive melt, posing a significant risk of localized subsurface corrosion and eventual spallation of the layer under operational thermal cycling, which would repeatedly open and close these microcracks. The most severe and pervasive corrosion was observed on the #3 anode, which exhibited a morphology characterized by extensive material removal superimposed with aggressive localized pitting. Macroscopically, the surface transitioned from relatively flat to the naked eye to heavily roughened, becoming extensively covered with a layer of corrosion products that altered its visual appearance and texture. At the microstructural level, SEM revealed a severely compromised surface topography characterized by a high density of irregularly shaped pits of varying depths, suggesting an uneven corrosion rate across the surface. Concurrently, the prominent exposure of the underlying graphite aggregates, which protrude from the surface, points towards a selective corrosion process where the matrix material surrounding these aggregates has been preferentially etched away. This complex morphology indicates that the corrosion mechanism was not confined to preferential attack at the grain boundaries but also involved direct and pervasive degradation of the graphite grains themselves, leading to a general surface recession and a pronounced increase in surface area and structural fragility.

The surface and cross-sectional morphologies of the graphite anodes before and after corrosion were characterized by SEM. Figure 2 presents comparative microstructural features of the three anodes, including pre corrosion surface views at 50×, postcorrosion surface views at 200× and 5000×, and cross-sectional views at 100× after exposure.

Graphite anode #1 exhibited characteristics typical of intergranular corrosion. Low-magnification (200×) observation revealed extensive granular spalling and microcracks on the surface, suggesting a uniform corrosion pattern. Higher-magnification (5000×) SEM images further revealed a loose, honeycomb-like surface structure scattered with granular debris. This morphology results from the destruction of the intergranular bonding forces in graphite, leading to grain detachment. Cross-sectional analysis indicated that the corrosion had progressed inwards, forming a distinct corrosion layer with a diffuse boundary with the underlying matrix, and the thickness of the corrosion layer was approximately 200–300 µm. This suggests that the molten salt penetrated along the grain boundaries, not only attacking the surface but also advancing into the interior of the material, thereby compromising the overall structural integrity. This corrosion mode is expected to significantly degrade the mechanical strength and electrical conductivity of the anode.

A notable feature of graphite anode #2 is the presence of a continuous and dense adherent layer covering its surface. At low magnification (200×), a network of cracks is visible on this layer, which is attributable to shrinkage stresses generated by the thermal expansion mismatch between the adherent layer and the graphite substrate during cooling. High-magnification (5000×) imaging confirmed that the layer was tightly bonded to the substrate but was internally cracked. While this adherent layer may partially act as a barrier against direct molten salt erosion, the crack network provides pathways for salt infiltration, potentially inducing localized corrosion. The cross-sectional view clearly reveals a well-defined interface with a certain layer thickness and strong bonding, indicating significant high-temperature interactions between the anode material and the molten salt.

Graphite anode #3 displays the most severe corrosion damage. At low magnification (200×), the surface appeared highly rough and structurally compromised and was covered with corrosion products and macroscopic defects. High-magnification (5000×) images show numerous irregular pits and fragmented, uplifted graphite layers, which is consistent with uniform corrosion superimposed with localized pitting. This indicates that corrosion attacked not only the grain boundaries but also the graphite grains directly and aggressively. Cross-sectional observations revealed a coarse corrosion layer containing abundant pores and internal defects. These defects facilitate the rapid penetration of molten salt deeper into the material, severely impairing the structural integrity of the cross-section. This morphology reflects the poorest tolerance of the #3 anode to fluoride molten salt corrosion.

Table 2. The characteristics of the corrosion layer.

Anodes	Morphology of the Corrosion Layer	Thickness of the Corrosion Coating/µm
#1	The surface exhibited a loose, honeycomb-like structure, with granular debris scattered across it.	200–300
#2	The surface was covered by a continuous, dense adherent layer that was tightly bonded to the substrate but exhibited internal cracking.	100–130
#3	The surface was highly rough and covered with corrosion products and macroscopic defects, indicating a structurally compromised state.	180–400

summarizes the morphology and thickness of the corrosion layer of various graphite anodes. The characteristics of the corrosion layers reveal distinct degradation levels among the anodes. Anode #2 demonstrated superior corrosion resistance, evidenced by having the thinnest (100–130 µm) and most continuous layer, which effectively shielded the substrate despite minor internal cracking. In contrast, the loose, honeycomb-like structure of the anode #1 thicker layer (200–300 µm) suggests poorer cohesion and easier pathways for corrosive agents. Anode #3 experienced the most severe degradation, indicated by the thickest and most variable layer (180–400 µm) with a rough, defective morphology, highlighting a structurally compromised state.

3.2. Changes in Corrosion Elements

To examine the surface elemental composition of the graphite anode in the fluoride molten salt, EDS analysis of the #2 graphite anode was performed, and the results are presented in Figure 3. The SEM image in Figure 3a shows the local morphology of the graphite anode surface, which has a relatively rough structure with visible pores, providing a basis for interpreting the subsequent elemental distribution. As shown in the carbon distribution map in Figure 3b, carbon is the predominant element, accounting for approximately 97.7% of the material, and is uniformly distributed across the surface, which is consistent with the inherent carbon-based nature of graphite. With respect to other elements, the chlorine distribution map in Figure 3c reveals a sparse presence of chlorine, indicating a low surface concentration prior to corrosion of approximately 0.9%. The EDS spectrum in Figure 3d,e quantitatively summarizes the relative content of each element: carbon constitutes the vast majority, whereas other elements—such as Si, Cl, K, and Ca—are present in extremely low quantities.

In the context of graphite corrosion in fluoride molten salt, graphite is primarily composed of carbon, and the corrosion process generally involves interactions between the carbon and molten salt components. The low concentrations of elements such as chlorine—which may originate from chloride impurities in the molten salt or the environment—along with trace amounts of calcium and silicon, likely derived from impurities in the graphite or external sources, indicate that the graphite anode surface is relatively “clean” before corrosion, dominated by carbon with only minor impurities. This initial elemental profile serves as a reference for tracking changes in surface composition during the subsequent corrosion process in fluoride molten salt. Both the interaction of the original trace impurities with the molten salt and the direct reaction between the carbon and the molten salt collectively govern the corrosion behavior and evolution of the surface elemental composition.

To elucidate the surface elemental characteristics and corrosion behavior of the graphite anode after exposure to fluoride molten salt, EDS analysis of #2 was conducted, and the results are presented in Figure 4. The SEM image in Figure 4a displays the local topography of the graphite anode surface after corrosion. A distinct reticular white shell structure, or aggregated white deposits, is observed covering the surface, with white substances filling cracks that extend into the graphite matrix. These features indicate substantial morphological changes induced by corrosion.

In terms of the elemental distribution, the F element map in Figure 4b shows a widespread and relatively uniform distribution of fluorine across the surface. Given the corrosive nature of the fluoride molten salt environment, the abundant presence of F suggests strong interactions between the molten salt and the graphite anode. The distributions of Pr in Figure 4c and Nd in Figure 4d are also relatively uniform, which is consistent with the formation of deposition products enriched in these rare earth elements—supporting the macroscopic observation of white substances. The O distribution in Figure 4e shows a moderate presence, which may originate from residual oxygen-containing impurities or oxidation during the corrosion process. Although the C element in Figure 4f remains detectable, its distribution is less uniform than that in the pre correction state, reflecting the consumption or structural alteration of carbon due to corrosion. The Na distributions in Figure 4g and Ca distributions in Figure 4h are present only in trace amounts, likely originating from impurities in the molten salt or the graphite itself.

The EDS spectrum in Figure 4i,j quantitatively summarizes the relative contents of each element. F, at approximately 18.5%, constituted a significant proportion, and Nd (approximately 59.1%) and Pr (approximately 9.2%) were also present in considerable amounts, whereas C, O, Na, Si, S, and Ca were detected at relatively low concentrations. Upon completion of the experiment, the graphite anode surface was covered with a white layer of solidified salt. EDS analysis confirmed the predominant presence of F, Nd, and Pr in this layer. The corrosion process not only altered the surface morphology but also created the composition. The originally carbon-dominated surface is supplemented with elements derived from the molten salt. The presence of O may be attributed to trace oxygen involvement in the reaction system, while the trace levels of Na and Ca further confirm the role of impurities in influencing the corrosion process.

In summary, combined morphological and elemental analyses demonstrate that the graphite anode undergoes a complex corrosion reaction in fluoride molten salt, resulting in the formation of new deposition products and notable changes in the surface elemental distribution.

3.3. Phase Changes

To comprehensively investigate the corrosion behavior and phase evolution of graphite anodes in a fluoride molten salt system at 1050 °C, XRD analysis was performed, and the results are presented in Figure 5.

For the original graphite anode, within the characteristic diffraction angle (2θ) range, the diffraction peak of graphite—particularly the (002) peak corresponding to its basal plane—shows high intensity and a sharp profile (as seen in the black reference curves in all subfigures of Figure 5). This indicates a typical and well-ordered crystalline structure. In graphite, carbon atoms are arranged in an ordered hexagonal layered structure, and the strong, sharp (002) peak reflects the high crystallinity of the graphite material, which is a fundamental characteristic of its structure.

After the graphite anode is subjected to molten salt corrosion at 1050 °C, significant phase changes occur, as clearly observed in the XRD patterns. The most obvious change is that the characteristic diffraction peaks of graphite shifted and weakened significantly. In the detailed view on the left (covering the 24–30° 2θ range) and the subfigures (#1, #2, #3) on the right, the intensity of the graphite peaks in the corroded samples (red curves) is substantially lower than that of the original graphite (black curves). In some regions, the graphite peaks nearly disappear.

The peak height of the graphite (002) crystal plane directly reflects the integrity of the long-range ordered structure of graphite. The data for the crystal plane position, crystal plane spacing, peak height, and grain size corresponding to the (002) plane are shown in

Table 3. Changes in the (002) crystal plane parameters of the graphite anodes before and after corrosion.

Samples		2θ/°	Shift/°	Peak Height/a.u.	Change Value/a.u.	Grain Size/nm
#1	Original	26.53	0.03	15692	−9453	70.4 ± 5
	After corrosion	26.56		6239		49.8 ± 5
#2	Original	26.47	0.06	8686	−1422	88.3 ± 5
	After corrosion	26.53		7264		66.5 ± 5
#3	Original	26.45	0.13	12252	−8193	50.1 ± 5
	After corrosion	26.58		4059		27.2 ± 5

.

After corrosion, all anodes exhibited a positive shift in 2θ, corresponding to a contraction in d₀₀₂. The extent of this shift, however, was strongly dependent on the initial microstructure of the graphite. Anode #1 showed the smallest shift (Δ2θ = +0.03°). This can be attributed to its high initial crystallinity and structural density, as evidenced by its sharp and intense original (002) peak. This robust structure provided superior resistance to the corrosive attack, minimizing lattice distortion. Anode #2 exhibited a moderate shift (Δ2θ = +0.06°). The most critical factor for #2 appears to be the protective, continuously solidified salt layer formed on its surface. This layer acted as an effective barrier that significantly impeded the penetration of molten salt, thereby limiting the interlayer contraction. Anode #3 demonstrated the most pronounced changes (Δ2θ = +0.13°). Its high porosity (20.03%, Table 1) and pre-existing microcracks from regeneration facilitated extensive molten salt infiltration. Severe corrosion led to significant disruption of the graphitic lattice.

The significant attenuation of the graphite (002) diffraction peak after corrosion is a direct consequence of the degradation of long-range crystallographic order, which diminishes the volume of material capable of coherent X-ray scattering. This phenomenon is driven by several interconnected mechanisms: First, the corrosive attack by F⁻ ions preferentially erode defective regions, such as grain boundaries and edge planes, thereby reducing the number of coherent scattering domains. This is evidenced by the substantial decrease in peak height, particularly in Sample #3 (66.9% reduction), where the disintegration of larger crystalline domains into smaller, misoriented fragments disrupts constructive interference. Second, the corrosion induces lattice distortion. This distortion, stemming from point defects and layer misalignment, reduces the periodicity of the atomic planes and introduces phase shifts in the scattered X-rays, leading to partial destructive interference and a net decrease in peak intensity. The extent of degradation is further governed by the initial microstructure. Samples with high inherent defect densities (e.g., the porous, regenerated Sample #3) or abundant grain boundaries (e.g., the fine-grained Sample #1) offer numerous pathways for corrosive infiltration, resulting in more severe destruction of ordered domains. In contrast, the formation of a protective salt layer on Sample #2 effectively mitigated these processes, accounting for its minimal peak intensity loss (16.4%) and lattice parameter change. Collectively, the attenuation of the (002) peak intensity unequivocally reflects a corrosion-induced decrease in crystallinity, governed by the decrease in the scattering domain size and number, coupled with an increase in lattice disorder.

This observation directly reflects the disruption of the layered crystal structure of graphite under the combined effects of high temperature (1050 °C) and reactive ions in the fluoride molten salt. The ordered arrangement of carbon atoms in the hexagonal lattice is disturbed, defects are introduced, and parts of the ordered structure are destroyed. From a structural perspective, this degradation explains the deterioration in the physical properties of graphite anodes during corrosion, such as reduced mechanical strength and electrical conductivity, due to the loss of structural integrity.

In addition to the attenuation of the graphite peaks, new diffraction features emerge in the corroded samples. Specifically, characteristic peaks corresponding to PrF₃ and NdF₃ are observed, indicating the formation of these rare earth fluorides. For NdF₃, peaks corresponding to the (002), (110), and (111) planes are identified, whereas for PrF₃, the (101) plane is detected (as labeled in the left-hand detailed view and the right-hand subfigures).

The weakening of the graphite diffraction peaks, together with the emergence of PrF₃ and NdF₃ peaks, provides direct evidence of the corrosion of the graphite anodes in the fluoride molten salt. The attenuation of the graphite signals reflects the degradation of the bulk graphite structure, which is consistent with the corrosion behavior of carbon materials reported in molten fluoride salts, including disruption of the lattice order and increased defect density [5,30]. In contrast, the appearance of PrF₃ and NdF₃ peaks indicates the formation of corrosion products resulting from interactions between the graphite anode and the molten salt medium.

In summary, the XRD results in Figure 5 clearly demonstrate the corrosion behavior of the graphite anodes in the fluoride molten salt at 1050 °C. The originally well-ordered graphite structure underwent significant degradation, while new fluoride phases formed as corrosion products. This phase evolution not only alters the chemical composition of the graphite anode surface but also affects its physical and chemical properties, providing important insight into the service life and performance degradation mechanism of graphite anodes in high-temperature fluoride molten salt environments.

3.4. Raman Analysis

Raman spectroscopy is a non-destructive and highly sensitive technique for structural characterization, offering unique advantages in the microstructural analysis of graphite materials. It is particularly sensitive to changes in the crystalline order, defect types, and layer stacking [31]. Long-term contact between graphite and fluoride molten salts (used as the anode) leads to salt penetration, chemical reactions, and thermal stress at elevated temperatures. The combined effects of these factors significantly alter the graphite structure, ultimately affecting its electrochemical performance and service life.

The first-order Raman spectrum of graphite is dominated by three characteristic peaks whose assignments are based on group theory and the double resonance mechanism. The G peak (~1582 cm⁻¹) corresponds to the E_2g phonon mode and arises from the in-plane stretching vibration of sp²-hybridized carbon atoms. This peak serves as a hallmark of the graphitic structure, with its full half-height width (FWHM) being a sensitive indicator of in-plane crystallinity and its position being influenced by stress and doping. The D peak (~1350 cm⁻¹) is a defect-induced peak activated by the intervalley double resonance process. It is associated with defects that break the in-plane symmetry, such as edges, grain boundaries, and vacancies. The intensity ratio of the D to G band (I_D/I_G) is commonly used to quantify the defect density. The D’ band (~1620 cm⁻¹) is a weak peak arising from the intravalley double resonance process, which is also activated by defects, particularly in-plane defects such as point defects. In the second-order Raman spectrum, the G’ band (~2700 cm⁻¹), also known as the 2D band, is an overtone of the D band. This peak is highly sensitive to the electronic structure and thus the stacking order of the graphitic layers. For example, single-layer graphene has a single, sharp 2D peak, whereas the Bernal-stacked bilayer and few-layer graphite have a characteristic multipeak structure due to interlayer coupling. The 2D band line shape is, therefore, a key fingerprint for identifying the number of layers and the stacking configuration [1,32,33,34].

Evolution of in-plane Crystallinity and Defects (G, D, and D’ Bands). The Raman spectra of the three types of graphite anodes before and after corrosion are shown in Figure 6.

The key Raman spectral data of the three types of graphite anodes before and after corrosion are summarized in

Table 4. Table of Raman spectrum data changes before and after graphite anode corrosion.

Samples		FWHM of G Band/cm−1	Change Values/cm−1	ID/a.u.	ID/IG	ID’/a.u.	Position of the 2D Band Peak/cm−1	Shift of 2D Band/cm−1	FWHM of 2D Band/cm−1	FWHM Change Values/cm−1
#1	Original	19 ± 2	≈5	234	0.09	0	2711	11	42 ± 2	≈21
	After corrosion	24 ± 2		300	0.10	2821	2722		63 ± 2
#2	Original	18 ± 2	≈7	179	0.06	0	2701	1	42 ± 2	≈20
	After corrosion	25 ± 2		426	0.13	3434	2702		62 ± 2
#3	Original	20 ± 2	≈6	562	0.17	0	2690	29	42 ± 2	≈22
	After corrosion	26 ± 2		719	0.22	3979	2719		64 ± 2

, including the positions of the D peak, 2D peak, G peak, D’ peak, intensity, and I_D/I_G transformation.

The first-order Raman spectrum provides critical insights into the in-plane structural integrity of graphite. Before corrosion, the G band of the graphite anode was located at 1581.8–1582.2 cm⁻¹ with a narrow FWHM of 18–20 cm⁻¹. This is consistent with the characteristics of highly crystalline graphite reported in the literature, indicating a well-ordered in-plane lattice structure with minimal disturbance to the sp² carbon stretching vibrations [35]. After corrosion, the FWHM of the G band substantially widened from 18–20 to 24–26 cm⁻¹, as shown in Table 4. The FWHM is generally linked to the amorphization of the sp² carbon network. The FWHM broadening from the original narrow range directly indicates the amorphization of the sp² carbon network, likely induced by the interactions between F and the carbon lattice at elevated temperatures [33]. This interaction, along with the propagation of microcracks due to thermal stress, disrupts the periodicity of the in-plane lattice, leading to peak broadening and a consequent decrease in crystallinity [36].

The defect evolution was further quantified by the D and G bands. Prior to corrosion, the D band was weak, and the intensity variation of peak D is shown in Figure 6b. The ratio changes in the I_D/I_G are shown in Figure 6c, and the original graphite anodes with I_D/I_G ratios in the range of 0.10–0.17 have a low initial defect density. After corrosion, the I_D/I_G ratio clearly increased to 0.11–0.24. The relationship between the I_D/I_G ratio and the in-plane crystallite size (La) was first established by Tuinstra and Koenig, who demonstrated that the I_D/I_G ratio is inversely proportional to La [33,34,37]. Importantly, this inverse relationship is primarily valid for the low-defect regime. The observed increase in I_D/I_G after corrosion unequivocally indicates a reduction in La and a greater density of defects. This grain refinement is attributed to the penetration of fluoride molten salt through open pores, which not only creates fresh edge defects on the pore walls but also promotes the fragmentation of large graphite grains into smaller domains, thereby increasing the D band intensity via the intervalley double resonance process.

The appearance of the D’ band at 1619–1621 cm⁻¹ after corrosion offers direct evidence for the generation of in-plane point defects, as shown in Figure 6a. This band, activated by the intravalley double resonance mechanism, confirms that the high-temperature exposure to F⁻ ions introduces atomic-scale disorder [38]. We propose that F⁻ ions temporarily adsorb onto dangling bonds at the graphite surface, disrupting the periodic arrangement of carbon atoms and creating vacancy-like defects, even in the absence of stable fluorocarbon formation, as detected by XRD.

Disruption of the interlayer stacking order (2D Band). The second-order 2D band serves as a sensitive probe for interlayer stacking. The initial graphite exhibited a well-resolved doublet at 2690 cm⁻¹ and 2727 cm⁻¹, characteristic of Bernal (AB) stacking in three-dimensional ordered graphite. After corrosion, this characteristic doublet merged into a single, broad peak. As quantified in Table 4, the 2D band position shifted substantially (e.g., by 29 cm⁻¹ for sample #3), and its FWHM increased significantly from approximately 42 ± 2 cm⁻¹ to over 63 ± 2 cm⁻¹ for all samples, with #3 exhibiting the most severe broadening, to 66 cm⁻¹. This transformation aligns perfectly with the 2D band profile of turbotrain graphite, which features random layer rotations and translations. This finding confirms that the molten salt penetrates the graphite interlayers, screening the van der Waals forces and disrupting the ordered AB stacking. Concurrently, the refinement of in-plane crystallites reduces the interlayer correlation length, collectively manifesting as the collapse of the bimodal 2D structure into a unimodal one.

Based on the Raman spectral evolution summarized in Table 4 and the comprehensive multi characterization results, the corrosion mechanism of graphite anodes in fluoride molten salt can be delineated into three sequential stages, with distinct quantitative differences observed across the three graphite types (#1, #2, and #3) that correlate strongly with their initial microstructural properties.

In the molten salt penetration stage, fluoride salts infiltrate the graphite bulk through open pores, preferentially attacking pore walls and generating edge defects, which is reflected by the initial increase in the D band intensity. As shown in Table 4, the original D band intensity of #3 graphite is notably greater than those of #1 and #2, which is consistent with its highest initial porosity, facilitating more extensive salt infiltration. After corrosion, the D band intensity increases across all samples, with #3 exhibiting the most significant increase (from 562 to 719 a.u., +27.9%), followed by #2 (from 179 to 426 a.u., +137.9%) and #1 (from 234 to 300 a.u., +28.2%). This variation indicates that while salt penetration initiates defect formation in all samples, the extent is modulated by the initial pore connectivity and salt-accessible surface area.

The defect generation and grain refinement stage involves high-temperature chemical interactions between F⁻ ions and basal plane carbon, introducing in-plane point defects (evidenced by the appearance of the D’ band in corroded samples) and thermal-stress-induced microcrack propagation, leading to pronounced grain refinement. The I_D/I_G ratio of all graphite anodes has increased, corresponding to a decrease in grain size. Concurrently, the G band FWHM broadens from 19 ± 1 to 24 ± 1 cm⁻¹ (#1), 18 ± 1 to 25 ± 1 cm⁻¹ (#2), and 20 ± 1 to 26 ± 1 cm⁻¹ (#3), directly reflecting the loss of in-plane crystallinity. Notably, #2 graphite, which has a protective cooled fluoride salt layer on its surface, exhibits a moderate increase in I_D/I_G and G band broadening, indicating that the salt layer mitigates but does not entirely suppress defect generation.

In the stacking disorder stage, molten salt molecules intercalate into the graphite galleries, weakening the interlayer van der Waals interactions and disrupting the AB stacking order. This is unambiguously reflected by the fusion of the 2D band doublet into a single peak. The 2D peak position shifts from 2711 to 2722 cm⁻¹ (#1), 2701 to 2702 cm⁻¹ (#2), and 2690 to 2719 cm⁻¹ (#3), accompanied by a significant increase in the FWHM (from ~40 to 55, 50, and 65 cm⁻¹ for #1, #2, and #3, respectively). The largest shift and broadening in the #3 graphite confirmed that regenerated graphite underwent the most severe interlayer disorder, likely due to its high initial defect density and extensive salt intercalation.

Collectively, these quantitative Raman, XRD, and SEM results validate the tri-phasic corrosion sequence while highlighting the critical role of the initial graphite microstructure (porosity, grain size, and defect density) and surface salt layer formation (in #2) in dictating the extent of each corrosion stage, providing a mechanistic framework for optimizing graphite anode durability in fluoride molten salt systems.

3.5. Infrared Spectroscopy Analysis

Infrared spectroscopy serves as a powerful analytical technique for probing molecular vibrations and identifying functional groups, rendering it highly valuable for characterizing the structural properties of carbon materials. Owing to its sensitivity to chemical bonding and surface chemistry, this method is particularly effective for investigating redox behavior, surface functionalization, doping, and other modification processes in graphite [39]. When graphite anodes are exposed to fluoride molten salt environments, their chemical structure and surface functional groups undergo significant alterations as a result of high-temperature conditions, salt penetration, and associated electrochemical reactions. These molecular-level changes can be effectively monitored via infrared spectroscopy. For example, in the case of graphite oxide, a broad absorption band at approximately 3435 cm⁻¹ is common, which is attributed to O–H stretching vibrations, suggesting the presence of adsorbed water or surface hydroxyl groups. A characteristic peak near 1632 cm⁻¹ corresponds to C=O stretching vibrations from carboxyl groups, confirming the introduction of oxygen-containing functionalities. The absorption at approximately 1590 cm⁻¹ is associated with the C=C stretching vibration of sp²-hybridized carbon within the graphitic lattice (E₂ symmetry mode), reflecting the degree of graphitic crystallinity. Moreover, the peak at approximately 1351 cm⁻¹ is generally assigned to structural defects in graphite and may also include contributions from bending vibrations of surface hydrogen-containing moieties such as CH₂ and CH₃ [40].

By comparing the infrared spectra of various graphite anodes before and after corrosion in fluoride molten salts, it is possible to trace their structural evolution, including the formation and cleavage of chemical bonds. This analysis offers essential molecular-level insights into the performance degradation and durability of graphite anodes under aggressive molten salt conditions. As shown in Figure 7, noticeable changes in the infrared spectra of different graphite anodes after corrosion further illustrate the structural alterations discussed above.

Figure 7a,b clearly shows a consistent attenuation of the infrared absorption peaks at 1351, 1384, 1590, 1632, and 3435 cm⁻¹ across all three types of graphite anodes after corrosion in the fluoride molten salt environment. These spectral changes provide direct evidence of microstructural and chemical transformations resulting from the combined effects of high temperature, molten salt infiltration, and fluoride-ion attack. A detailed interpretation of these spectral changes, organized by wavenumber region, is presented below.

In the high-wavenumber region, the pronounced decrease in the broad absorption band near 3435 cm⁻¹, which is associated with O–H stretching vibrations, indicates a substantial loss of surface hydroxyl groups and adsorbed water. This decrease is attributed to two main processes: the thermal desorption of water molecules at elevated temperatures and the chemical reaction between F⁻ and hydroxyl groups, which likely results in the production of volatile hydrogen fluoride. The variation in the extent of this decrease among the three graphite samples can be correlated with differences in their pore structure and surface-active-site density. Graphite with higher porosity and more active sites experiences greater interfacial contact with the molten salt, leading to more extensive removal of hydroxyl species.

In the medium-wavenumber region, the spectral changes offer key insights into the degradation of the graphitic framework. The decrease in the peak at 1632 cm⁻¹, assigned to C=O stretching vibrations in carbonyl or carboxyl groups, reflects a net decrease in the number of oxygen-containing functional groups. This can be explained by two factors: first, pre-existing oxygenated functional groups are likely etched away through reactions with F⁻, forming volatile fluorine-containing compounds; second, the oxygen-free molten salt environment inhibits the formation of new oxygen-containing groups during corrosion. More importantly, the decrease in intensity at 1590 cm⁻¹, corresponding to the in-plane C=C stretching vibration of the sp²-carbon lattice (G band), signals a decrease in graphitic crystallinity. The combined action of thermal stress and chemical attack by fluoride ions disrupts the long-range order of the graphitic layers, induces lattice distortion, and partially damages the conjugated π-system, thereby weakening the C=C stretching vibration.

In the low-wavenumber region, the spectral changes reveal complex behavior related to defect evolution and removal of hydrocarbon species. The absorption near 1351 cm⁻¹, known as the D band, is associated with structural defects in sp² carbon materials. The decrease in its intensity suggests a reorganization of the defect structure, possibly due to the disruption of local C–C bonds by infiltrating F⁻ ions at high temperatures. Additionally, the peak at 1384 cm⁻¹, which is commonly linked to C–H bending vibrations in aliphatic hydrocarbons (e.g., –CH₂ and –CH₃), also has a reduced intensity, indicating the desorption of hydrocarbon impurities under high-temperature conditions. Concurrently, F⁻ may undergo substitution reactions with edge C–H groups, forming C–F bonds and further reducing the content of hydrogen-containing groups, contributing to the overall decline in absorbance in this region.

A comprehensive comparative analysis of the FTIR spectra acquired before and after corrosion revealed a multifaceted degradation mechanism for graphite anodes in fluoride-based molten salt environments. This mechanism encompasses the depletion of surface functional groups coupled with a fundamental deterioration of the graphitic crystalline order. Variations in corrosion severity across the three graphite variants underscore the critical role of initial physicochemical properties—including porous architecture and surface-active site density—in governing their electrochemical durability within such corrosive operational environments.

4. Conclusions

This study presents a comprehensive investigation into the corrosion behavior and underlying mechanisms of three distinct graphite anodes in a high-temperature fluoride molten salt environment simulating industrial rare earth electrolysis. The findings conclusively demonstrate that the corrosion resistance and service life of graphite anodes are not dictated by a single factor but are the result of a complex interplay between intrinsic material properties, including the crystal structure, physical characteristics, and microstructural morphology. The following conclusions are drawn from the multi technique analysis:

(1)

The three graphite anodes exhibited fundamentally different corrosion modes directly linked to their initial structure. The fine-grained, isotropic #1 anode underwent intergranular corrosion, characterized by granular exfoliation and the formation of a honeycomb-like surface structure, indicating preferential attack along the grain boundaries by the penetrating molten salt. In contrast, the medium-coarseness #2 anode was covered by a continuous, adherent layer of resolidified fluoride salt. While this layer could act as a temporary physical barrier, its network of thermal-stress-induced microcracks provided pathways for continued salt infiltration and potential localized subsurface attack. The recycled #3 anode displayed the most severe degradation, with a morphology of extensive material removal, aggressive pitting, and exposure of underlying aggregates, indicative of a combined uniform and localized corrosion mechanism that attacked both the matrix and the grains.

(2)

After corrosion, EDS and XRD analyses confirmed significant phase alteration of the anode surfaces. The original carbon-dominated surface was enriched with fluorine, praseodymium, and neodymium. The formation of different PrF₃ and NdF₃ phases was identified via XRD, providing direct evidence of the molten salt composition deposited on the surface of the graphite anode. The presence of PrF₃ and NdF₃ phases in the graphite anode is attributed primarily to surface deposition and infiltration of the molten salt electrolyte during the electrolysis process. Upon cooling, the infiltrated salt solidifies within the pores of the graphite matrix. XRD and EDS analyses did not reveal any evidence for the direct formation of rare earth carbides (e.g., PrC₂, NdC₂) or a direct reaction between the rare earth metals (Pr/Nd) and the carbon lattice to form these fluorides. Therefore, the formation mechanism of PrF₃ and NdF₃ is a physical deposition and encapsulation process rather than a chemical reaction with the carbon substrate.

(3)

Raman spectroscopy offers profound insights into the microstructural corrosion mechanism, which can be delineated into three sequential stages. Penetration and initial attack: Molten salt infiltration through open pores creates fresh edge defects, corresponding to the initial increase in the D-band intensity. Defect Generation and Grain Refinement: Chemical interaction of F⁻ ions with the basal plane carbon introduces in-plane point defects (evidenced by the emergence of the D’ band). Concurrently, thermal stress and chemical attack lead to pronounced grain refinement, marked by a significant increase in the I_D/I_G ratio and broadening of the G band FWHM, indicating a reduction in the in-plane crystallite size. Stacking Disorder: The collapse of the characteristic bimodal 2D band into a single, broad peak unambiguously confirms the disruption of the interlayer stacking order. This is attributed to the intercalation of molten salt species into the graphite galleries, which screens van der Waals forces and randomizes the layer orientation, transforming the material towards a turbotrain structure.

(4)

FT-IR analysis revealed a consistent attenuation of absorption bands associated with O–H, C=O, and C–H functional groups after corrosion. These findings indicate that the thermal desorption and chemical etching of surface functional groups by fluoride ions further contribute to surface degradation and the loss of structural integrity.

(5)

The superior corrosion resistance of the #2 anode is attributed to its optimal combination of properties: large grain size, high bulk density, and low apparent porosity. The #1 anode, with its finer grain structure and higher porosity, offered more pathways for corrosive attack. The #3 anode, despite a potentially high degree of graphitization, was compromised by its inferior physical properties, likely stemming from its recycled nature, leading to the most severe corrosion.

In summary, this work systematically elucidates the multistage corrosion mechanism of graphite anodes in fluoride molten salts, linking macroscale damage to microscale structural and chemical evolution. The key to enhancing anode longevity lies in selecting or engineering graphite materials with high crystallographic order (high graphitization, large grain size), dense and low-porosity microstructures, and high chemical purity. These findings provide a robust theoretical foundation for the selection, optimization, and potential development of advanced graphite anodes, paving the way for improved economic efficiency, operational stability, and environmental performance in the industrial electrolytic production of rare earth metals.