1. Introduction

The Korea Atomic Energy Research Institute (KAERI) has been conducting anaerobic long-term corrosion experiments in the YS03 exploration borehole at a depth of 300 m to study the corrosion behavior of candidate materials for disposal canister. The bottom of the YS03 borehole is characterized by water quality conditions with a temperature of 20°C, an Eh of around −300 mV, and a hydrogen sulfide (HS) concentration ranging from 4 to 5 ppm [1]. In the YS03 long-term corrosion test, various metal specimens, such as copper and cast nodular iron, were surrounded by compact bentonite blocks within a rigid corrosion module and placed at the borehole’s bottom. The bentonites used were calcium-type GyeongJu bentonite (GJ-I) and sodium-type Wyoming bentonite (MX80). The objectives of the YS03 experiment were to assess the corrosion behavior of candidate materials for high-level waste (HLW) disposal canister in a hydrogen sulfide-rich environment and to evaluate the alteration behavior of bentonite materials. The corrosion modules have been recovered twice from the YS03 borehole, and the impact of hydrogen sulfide (H₂S) on metal corrosion has been investigated [2, 3]. Recently, upon recovering the corrosion module in which MX80 was used as a buffer material, it was observed that the significant swelling of MX80 had caused the module head to detach, thereby releasing the compressed state of the internal buffer. Analysis of the remaining MX80 in the corrosion module confirmed that montmorillonite, the primary component of MX80, had undergone complete alteration to form a substantial amount of muscovite, goethite, and magnetite.

Bentonite is generally used as a buffer material in HLW repositories because its main component, montmorillonite, develops high swelling pressure and low permeability when it absorbs water. This high swelling pressure and low permeability provide support for the disposal canisters, restrict nuclide transport, and inhibit microbial activity, thereby contributing to the long-term integrity of the disposal canisters. For these reasons, SKB sets the montmorillonite content in its bentonite buffer at a high 75–90% [4]. However, if montmorillonite transforms into non-swellable minerals like illite or muscovite over the long disposal period, it could pose a significant problem to the integrity of the disposal canisters. While extensive research has been conducted on the alteration of montmorillonite to illite in a repository environment, previous studies have not reported severe montmorillonite alterations as observed in this research [5-9]. As a result, it is necessary to report these experimental findings to the broader research community. This paper presents the experimental conditions, analysis results, and subsequent discussions.

2. Experiment

2.1 YS03 Borehole

The YS03 borehole is a 300-meter-deep geological exploration hole excavated in 2002 within the territory of KAERI (Korea Atomic Energy Research Institute). Despite its potential, the borehole remained unused for several years after its initial exploration. In 2002, when it was first explored, the sulfate (SO₄²⁻) concentration in YS03 was approximately 4.4 mg·L⁻¹. Over the years, organic matter such as fallen leaves accumulated at the bottom of the borehole. In 2018, water properties were measured, revealing a dissolved oxygen (DO) value of about 0.1 mg·L⁻¹ at a depth of 250 meters, indicating nearly anaerobic conditions. The presence of hydrogen sulfide (HS) was detected by the characteristic smell of rotting eggs in the groundwater. Using a colorimetric method, the HS concentration was measured and found to be between 4 and 5 ppm.

Table 1 summarizes the analysis results of the dissolved solutes in the YS03 groundwater. The chloride (Cl⁻) concentration, which can affect metal corrosion, was found to be low, at the level of 2–3 ppm. Bicarbonate (HCO₃⁻) was the most abundant ion. Sulfate (SO₄²⁻), which can be converted to hydrogen sulfide (HS^⁻) by sulfate-reducing bacteria (SRB), was around 4.8 ppm. Among the cations, sodium (Na^⁺) was the most abundant at levels of 20 to 30 ppm, and calcium (Ca^²⁺) was also present in significant amounts at around 4 ppm. This detailed characterization of the YS03 borehole environment provides crucial baseline data for understanding the corrosion behavior of metals and alteration processes of minerals exposed in this unique anaerobic and chemically distinct environment.

2.2 Test Module

The cylindrical corrosion module inserted into the YS03 borehole was a stainless steel (SS304) canister of BX size (D60.5 mm × t 2.8 mm), designed to fit within the NX size (D76 mm) YS03 borehole. Inside this canister, nine compact bentonite blocks (D54 mm × t 10 mm) were placed, as depicted in Fig. 1(a). Four corrosion specimens were placed on each block layer, arranged in the following order from top to bottom in the module: SKB forged copper, SS304, cast nodular iron, CSC (cold spray coating) copper, titanium (Gr. 2), and hot-rolled copper, as shown in Fig. 1(b). Both ends of the bentonite blocks were blocked with 2 mm thick metal filters to prevent the bentonite buffer from draining out. In a separate compartment at the bottom of the corrosion module, each corrosion specimen was wrapped in paper towels without any bentonite to compare the protective effect of the compact bentonite. This arrangement allowed us to assess not only the corrosion behavior of various metals in the bentonite-buffered environment but also the effectiveness of bentonite in preventing metal corrosion under the unique conditions of the YS03 borehole.

Fig. 1

(a) Schematic diagram of corrosion module and (b) the compact bentonite blocks with corrosion specimens.

2.3 Microbial Activity

Microbial analysis was conducted on the recovered paper towel from YS03 to assess the activity of sulfatereducing bacteria (SRB) capable of producing hydrogen sulfide (HS). It was found that anaerobic bacteria were the dominant species at the bottom of the YS03 borehole, although aerobic bacteria were also present. The most abundant species at the bottom was the anaerobic bacterium Geofilum rhodophaeum (20%). Six species belonged to SRB, comprising 5.28% of the total microbial community. Desulfobulbus propionicus, which is responsible for the breakdown of sulfur into sulfate and sulfide, accounted for 1.68%. Sulfurisoma sediminicola, which can oxidize sulfur, was 0.22%, and Desulfuromonas acetexigens, which can reduce sulfur into sulfide, was 0.4%. These microbial analyses indicated that a significant amount of sulfur was present in the YS03 underground water. Additionally, four species of nitrate-reducing bacteria (NRB), capable of producing nitrite (NO₂⁻), accounted for 1.88% of the total community, indicating the potential for corrosion of iron materials such as cast nodular iron and stainless steel 304. Notably, Thermostilla marina, which is capable of both sulfur reduction and nitrate reduction, comprised 1.06% of the community. The presence of nitrite in YS03 was inferred from the discovery of Melioribacter roseus (0.87%), a nitrite- reducing bacterium, and Nitrospira lenta (0.10%), a nitrite-oxidizing bacterium. Furthermore, other bacteria capable of reducing ferric (III) ions were identified, including Melioribacter roseus (0.87%), Paludibaculum fermentans (0.37%), and Ferribacterium limneticum (0.86%). Additionally, Sideroxydans lithotrophicus (0.18%), which oxidizes iron, was also detected. In conclusion, the microbial analyses revealed the presence of various anaerobic microorganisms in the YS03 borehole, including a significant number of microorganisms capable of producing hydrogen sulfide.

2.4 Test Module Recovery

The first recovery of the corrosion module was conducted after 178 days. X-ray diffraction (XRD) and X-ray fluorescence (XRF) analyses were performed on the recovered corrosion specimens and bentonite. The XRD analysis detected chalcocite on the copper specimen, indicating corrosion by hydrogen sulfide (HS). However, no apparent changes in the mineral composition were observed in either type of compact bentonite.

The second recovery of the corrosion modules was carried out after 745 days (2.04 years) for the GJ-I module. The MX80 module was not recovered at that time due to the high swelling pressure of MX80, which caused the detachment of the module head. Approximately 6 months later, at 945 days (2.59 years), the MX80 module was successfully retrieved using a specially designed fishing hook. The recovered corrosion modules were dried in a glove box, and bentonite residues were extracted, as shown in Fig. 2. However, a significant amount of MX80 was lost, and only some portions were retrieved. Additionally, the color of the MX80 had changed from its original pale gray to redbrown. The recovered bentonites were analyzed using XRF, XRD, ion chromatography (IC), and inductively coupled plasma optical emission spectrometry (ICP-OES). This comprehensive analysis aimed to understand the extent and nature of the alterations in the bentonites and the impacts on the corrosion behavior of the specimens.

Fig. 2

(a) Inside view of recovered MX80 module and (b) the bentonite residue.

3. Experimental Result

The analysis results for the secondary recovered bentonite from YS03 borehole are described as follows.

3.1 Compact Ca Bentonite

3.1.1 XRD analysis

We investigated whether there was any mineral alteration in compact GJ-I through XRD analysis. The XRD graphs (Fig. 3) showed no noticeable changes between the 2.04-year-old bentonite near cast nodular iron and the original GJ-I bentonite. Therefore, it was not clear if there were any alterations in the recovered GJ-I after 2.04 years under the YS03 environment. The quantitative XRD analysis results are shown in Table 2. The montmorillonite ((Na,Ca)_0.33(Al,Mg)₂(Si₄O₁₀)(OH)₂·nH₂O) content in GJ-I showed no difference compared to the original GJ-I, but the content of magnetite (Fe₃O₄) was noticeably increased in all used bentonite samples compared to the original GJ-I. This suggests that some mineral within the bentonite may have undergone alteration to magnetite, as magnetite was found consistently across all bentonite blocks.

Fig. 3

XRD curves of (a) 2.03-year-old GJ-I and (b) Original GJ-I; M: Montmorillonite, A: Albite, Q: Quartz.

3.1.2 XRF analysis

The XRF analysis results for GJ-I are summarized in Table 3. There was no significant difference in the inorganic oxide composition compared to the original GJ-I, regardless of the corrosion specimen. Therefore, it was difficult to verify the influence of corrosion products on bentonite alteration. Typically, when minerals such as feldspar dissolve, a decrease in SiO₂ is expected in bentonite [10, 11]. However, since no decrease in SiO₂ was observed in this analysis, it is concluded that the high-density bentonite effectively prevented mineral leaching.

3.1.3 CHNS analysis

The trace elements (C, H, N, S) were analyzed using an Elemental Analyzer at the level of 10⁻³at%, and the results are presented in Table 4. Although some amount of sulfur was expected in the compact GJ-I bentonite, given the high concentration of HS in the YS03 borehole, no sulfur was detected. This suggests that little reaction occurred between HS and the bentonite minerals. The content of Carbon (C) and Hydrogen (H) remained nearly unchanged, while Nitrogen (N) showed a slight decrease. From this, it can be concluded that the high-density bentonite effectively prevented solute permeation into the bentonite.

As a result of analyzing the recovered Ca bentonite, which maintained its compact state, no significant alterations were observed. This outcome was attributed to its dense compactness, which suppressed the mass transfer of reactive agents and limited microbial activity.

3.2 Uncompact Na Bentonite

The compactness of the MX80 in the module was compromised due to the detachment of the module head. The Na bentonite residues were categorized into top, middle, and bottom sections based on the module height. Subsequent analysis was conducted to verify mineral alterations in the bentonite.

3.2.1 Anion analysis

Anion analysis was performed using a Metrohm Ion Chromatograph (IC), and the results are presented in Table 5. Analysis of the MX80 anions indicated that chloride and nitrate levels decreased, while fluoride levels increased. The increase in fluoride is likely attributed to the high fluoride concentration (9.1 mg·L⁻¹) in the YS03 groundwater. However, it was observed that most of the sulfate, which was present in high concentrations in the MX80, had disappeared dramatically. This can be considered evidence of active sulfate-reducing bacteria (SRB) in the MX80. Additionally, although not the dominant species, the presence of nitrate-reducing bacteria (NRB) along with SRB likely contributed to the decrease in nitrate levels.

3.2.2 Cation analysis

The concentrations of major cations analyzed by ICPOES are presented in Table 6. The results indicated a significant increase in Fe cations, confirming that all parts of the bentonite were contaminated by these ions. Given that the corrosion cell made of SS304 exhibited no signs of corrosion, it is likely that these Fe cations originated from the corrosion products of cast nodular iron. Additionally, some contamination by copper cations was noted. The lower copper concentration in the upper part suggests that a significant amount of copper ions diffused out into the groundwater. Moreover, the potassium ion content, which plays a role in the illite transformation of montmorillonite, nearly doubled in the upper part that was in direct contact with the groundwater. This suggests that the potassium ions (0.4 mg·L⁻¹) present in the YS03 groundwater likely influenced the alteration of the bentonite to some extent [2]. Furthermore, the ratio of Na to Ca cations indicated that the composition ratio of MX80 was reversed, suggesting that the interlayer Na cations in the bentonite might have been replaced by Ca cations.

3.2.3 XRD analysis

The major mineral components were analyzed by XRD, and the graphs are presented in Fig. 4. In the XRD graphs, the main peak of MX80 after 2.59 years is identified as Muscovite (KAl₂(AlSi₃O₁₀)(F,OH)₂), whereas the main peak of the original MX80 is Montmorillonite. Additionally, the overall graphs show significant differences in peak distribution between the two samples. The quantitative analysis results for the MX80 are summarized in Table 7. The originally present Montmorillonite has completely disappeared, and new minerals such as Muscovite, Goethite (α-FeO(OH)), and Magnetite (Fe₃O₄) have formed. The total amount of these newly formed minerals closely matches the original content of Montmorillonite. From this observation, it can be inferred that Montmorillonite has mostly transformed into these newly formed minerals. Montmorillonite is known to react with K⁺, releasing Na^⁺/Ca^²⁺ and undergoing transformation into Illite (K,H₃O)(Al,Mg,Fe) ₂(Si,Al)₄O₁₀[(OH)₂,(H₂O)], and ultimately converting into Muscovite [12, 13]. The decrease in sodium ions in Table 6 is believed to be related to the formation of Muscovite. However, since no Illite was detected, it appears that Montmorillonite has completely transformed into its final form, Muscovite. Additionally, a significant amount of Goethite and Magnetite was found in the MX80. These Magnetite and Goethite minerals are believed to be major corrosion products of cast nodular iron, thereby explaining the reddish color observed in the recovered MX80. However, the contents of Albite, Quartz, and Calcite, which are stable minerals in groundwater, remained nearly unchanged as shown in Table 7. From the XRD analysis, it is evident that the compactness of the bentonite has a critical impact on the rate of alteration, as there was minimal alteration in the compact GJ-I sample, whereas the uncompact MX80 was severely altered.

Fig. 4

XRD curves of (a) 2.59-year-old MX80 and (b) Original MX80; C: Calcite, G: Goethite, M: Montmorillonite, Mu: Muscovite, Q: Quartz.

Consequently, montmorillonite, which originally comprised about 78wt% of the MX80, was not observed in the XRD analysis. Swelling minerals might have partially escaped through the open head, but the total amount of newly formed minerals was nearly equivalent to the original montmorillonite content. Therefore, it is inferred that montmorillonite was altered into muscovite, goethite, and magnetite. Montmorillonite is known to alter into muscovite under specific natural conditions [14], and in the presence of iron ions, it can also transform into goethite [15], which are consistent with our experimental results.

3.2.4 XRF analysis

The results of the XRF analysis for the MX80 are presented in Table 8. The XRF analysis revealed that iron oxide is the most abundant component. Additionally, the lattice components of montmorillonite, such as alumina (Al₂O₃), silica (SiO₂), and magnesium oxide (MgO), were found to have significantly decreased. Sodium oxide (Na₂O), an interlayer cation of MX80, also significantly decreased. The increase in iron oxide, potentially originating from the corrosion of cast nodular iron, seems to be related to the formation of goethite and magnetite.

As a result of the XRD and XRF analyses of MX80, montmorillonite has completely disappeared, with muscovite emerging as the primary mineral. Additionally, significant amounts of goethite and magnetite have been generated. In the presence of Fe cations, montmorillonite is known to transform into goethite [16], and magnetite (Fe²⁺, Fe³⁺) is considered an intermediate in the transition of montmorillonite to goethite (Fe³⁺). The red coloration of the bentonite is attributed to reddish goethite.

4. Discussions

In the study of bentonite alteration under anaerobic conditions rich in hydrogen sulfide and microorganisms, no other alteration products were found in the compact GJ-I sample except for a small amount of magnetite, and the montmorillonite content had nearly unchanged. However, in contrast to GJ-I, the decompressed MX80 exhibited severe alterations where montmorillonite had completely disappeared, and new minerals such as muscovite, goethite, and magnetite had formed. Muscovite is a layered silicate mineral composed of aluminum and potassium. Natural muscovite is reported to be produced in granite magma at pressures above approximately 1,500 atmospheres [14]. Therefore, the formation of muscovite typically requires high-temperature and high-pressure conditions, which are unlike our experimental conditions. According to Pusch et al. [12], montmorillonite converts to illite as it firmly bonds with K⁺ ions, and illite acts as an intermediate mineral in the transformation of montmorillonite to muscovite. Additionally, they mentioned that this alteration process is associated with the release of silica [17]. This is consistent with the XRF analysis, which showed a decrease in silica. However, these studies are based on geological phenomena occurring under high temperature and high pressure in the deep Earth. It is uncertain whether they can be directly applied to the rapid alteration observed in this study.

In this experiment, the parameters presumed to have influenced the rapid alteration of bentonite are microbial activity and the widespread Fe cations within the bentonite. According to Carolin Podlech et al. [5], who studied the long-term alteration of bentonite under the influence of microorganisms, no microbial effects were observed, nor was the formation of new minerals reported in their twoyear experiment. They used uncompressed European Cabentonites at an ambient temperature of 25°C and included three species of sulfate-reducing bacteria (SRB). Despite these experimental conditions being similar to ours, their two-year study did not conclude that microorganisms affected mineral alteration. The factors they identified for bentonite alteration were the nature of the bentonite, pore water chemistry, and temperature, which differ significantly from our experimental findings.

In this study, MX80 contained a high content of magnetite and goethite, which seemed to be the alteration products of montmorillonite. Therefore, it was suspected that Fe cations might be one of the parameters influencing the alteration of montmorillonite. These Fe cations might have been released from the corroded cast nodular iron by hydrogen sulfide and may have substituted the Al, Si, and Mg cations in the montmorillonite lattice. According to previous studies, microorganisms reduce sulfate to sulfide, which can corrode metals [18]. additionally, sulfide can reduce Fe³⁺ to Fe²⁺ [19], and Fe²⁺ can penetrate the interlayers of smectite, destabilizing its dioctahedral structure [20]. According to the SKB study [6], tetrahedral substitution of Si⁴⁺ by Al³⁺ was observed in the montmorillonite directly in contact with hot iron, even though there was no significant loss of smectite. This phenomenon was presumed to be related to the presence of ferric and Fe cations resulting from the corroded iron pipe [6]. Therefore, it appears that the alteration of montmorillonite is largely influenced by Fe cations, and the increased presence of goethite and magnetite in our experiment might be the evidence of this influence.

SKB reported two long-term iron-bentonite interaction tests conducted by Finland’s VTT and Japan’s JAEA [8]. In the VTT test, after 8 years of contact between compacted MX80 and cast iron, ferrous cation (Fe²⁺) had diffused into the bentonite, resulting in a slight decrease in swelling pressure. Additionally, they observed an increase in the illite fraction and a decrease in SiO₂ in the bentonite. However, they could not identify clear evidence of mineral alteration in the bentonite. JAEA conducted a 10-year long-term test with a mixture of Kunipia F and iron powder [8]. In samples from 0.3−0.6 M NaCl solution with high pH, montmorillonite was altered to non-swelling berthierine, but no alteration occurred in samples from 0.05 M Na₂SO₄ or distilled water. Composition analysis confirmed the presence of magnetite and pyrite in addition to montmorillonite, and silica release was also observed during the alteration process. The SKB concluded, after reviewing the two tests, that the alteration was caused by the high pH and noted that no alteration occurs at a normal pH of approximately 8 [8]. In our study, the pH of the YS03 borehole remained in the range of 9−10 during the long-term test. Therefore, it is concluded that SKB’s claim has a certain degree of validity. Previous studies on long-term corrosion tests of iron in bentonite did not mention any goethite formation. However, certain studies on geologic bentonite layers rich in iron have reported goethite formation as an alteration product of bentonite [21].

Based on a review of previous studies related to bentonite alteration, there is a claim that bentonite alteration is influenced by high pH and Fe cations. However, this assertion seems insufficient to clearly explain the results of this study. It is hypothesized that sulfate was reduced to sulfide, which then corroded the cast nodular iron, creating an environment rich in Fe cations. It is conceived that these Fe cations, along with potassium cations in groundwater, might have altered the montmorillonite lattice. Additionally, it is believed that SiO₂ was released from the montmorillonite during the alteration process.

5. Conclusions

In this study, we extensively analyzed the compressed GJ-I, a calcium bentonite, and decompressed MX80, a sodium bentonite from the second retrieval, both of which had been in the YS03 borehole for two years. The compacted GJ-I did not exhibit noticeable alterations, but a small amount of magnetite was detected via XRD analysis. In contrast, the uncompact MX80 showed severe alteration of montmorillonite into muscovite, goethite, and magnetite. And the previously retrieved samples of GJ-I and MX80 in compressed state from the first retrieval had shown no signs of bentonite alteration. Therefore, we suspected that the compactness of the bentonite has a critical impact on the rate of alteration. XRF analysis of the Na-bentonite indicated a significant increase in iron content, while major components of montmorillonite, such as alumina (Al₂O₃), silica (SiO₂), and magnesium oxide (MgO), decreased. The causes of the bentonite alteration were estimated to be the high content of Fe cations in the bentonite and the high pH in YS03. The high concentration of Fe cations appears to be due to the corrosion of cast nodular iron by hydrogen sulfide (HS) produced by sulfate-reducing bacteria (SRB). It is believed that these Fe cations destabilized the tetrahedral structure of montmorillonite, leading to its alteration. However, the exact cause of the alteration observed in the decompressed MX80 was difficult to determine in this situation. The first hypothesis is that the absence of sulfate in the decompressed bentonite implies that microbial activity, such as that from sulfate-reducing bacteria, may have influenced the bentonite alteration. The second hypothesis is that the presence of alteration products like magnetite and goethite, commonly found in the altered bentonite, suggests that iron ions might have played a role in the alteration. Additionally, it is important to clarify at which degree of bentonite compression these alterations become prominent. To address these uncertainties, it is deemed necessary to conduct systematic bentonite alteration experiments in the YS03 borehole, varying the compression density and the presence or absence of cast iron samples.