1. Introduction

The study of the chemical behavior of radionuclides under geochemical conditions is essential for long-term safety assessment of the deep geological disposal of spent nuclear fuel (SNF) [1]. The most probable migration path of radionuclides is through groundwater flow [2]. Radionuclides interact with various organic and inorganic components in groundwater as well as with mineral surfaces and colloidal particles [3-5]. The concentration of dissolved radionuclides in groundwater is the critical initial data that determines the maximum quantity to be considered for radionuclide interactions with environmental materials, including surrounding rocks. For a reliable evaluation of radionuclide migration through the geosphere to the biosphere, different local geochemical conditions as well as temporal variations due to the long-term evolution of the disposal system should be considered. These diverse spatial and temporal variations make it difficult to experimentally determine radionuclide solubilities and interactions with environmental materials.

International efforts have been devoted to establishing a reliable thermochemical database (TDB) for geochemical modeling of radionuclide behavior under diverse conditions. The selected data in the NEA-TDB [6] are renowned as reliable international standard data, but the data listed in the NEA-TDB are limited. The ThermoChimie TDB by Andra accepted thermodynamic data from the NEA-TDB and amended more data, even with low reliability, to better describe the systems of interest [7-8]. The ThermoChimie TDB was used for safety assessment of the Olkiluoto repository in Finland [9]. The ThermoChimie_11a version was updated in 2021, accommodating the NEA-TDB 2020 update. The reliability of the geochemical modeling results depends on the reliability of the thermodynamic data used and the comprehensiveness of the database. In addition, geochemical modeling results should be evaluated in conjunction with the experimental results.

Americium is one of the major radionuclides responsible for the long-term radiotoxicity of SNF. The ²⁴¹Am (half-life = 432.7 years) isotope is a relevant element in the radiotoxicity of SNF, as it is continuously generated and accumulated in SNF by the β-decay of ²⁴¹Pu [10-11]. ²⁴¹Am emits alpha (5.49, 5.44 MeV) and gamma rays (59.5 keV), and is converted to ²³⁷Np (α-emitting nuclide) [12]. It is a highly radioactive element with a specific activity of 126.9 GBq·g⁻¹ and can be easily quantified in the sub-nanomolar range using radio-analytical methods. Americium is stable in the +3 oxidation state under a wide range of aqueous solution conditions [2]. Thermochemical data for Am(III), a representative of +3 actinides and lanthanides, are relatively well described in the NEA-TDB and ThermoChimie. Recently, we reported detailed information on the spectroscopic properties of Am(III) species [13-15]. UV-Vis absorption and time-resolved laser-induced fluorescence spectroscopy are useful for Am(III) speciation.

In this study, dissolved americium concentrations and species distributions were examined under synthetic groundwater conditions (Syn-DB3), simulating groundwater from the DB3 site in the KAERI Underground Research Tunnel (KURT). Geochemical modeling was performed using the TheroChimie_11a TDB. Absorption and luminescence spectroscopy were used to monitor the evolution of species and colloidal particles in the samples. The solubility of americium in Syn-DB3 groundwater was measured using liquid scintillation counting (LSC) under oversaturated conditions. The experimentally measured values were compared with those predicted using geochemical modeling. Probable americium solid phases formed in the Syn-DB3 groundwater are discussed.

2. Methods

2.1 Geochemical Modeling

The solubility curves of americium were calculated using the Act2 module of the commercial GWB program (2022 Standard). A pH range of 6–11 was set to include the general groundwater pH range. As summarized in Table 1, the major ions contained in the Syn-DB3 groundwater were considered for modeling. The background ionic strength of the Syn-DB3 groundwater was calculated as 1.5 mmol·kg⁻¹, which has an insignificant effect on the activity of ions; therefore, the molarity itself was used as the activity. In the case of silicate, carbonate, and sulfate ions, ‘speciate over x-y’ was set to consider the degree of dehydrogenation according to pH change. The temperature and pressure were set at 25℃ and 1.0 bar, respectively. The aqueous species distribution of Am in the Syn-DB3 groundwater was simulated using the GWB React module. When the pH was varied from 6 to 11, the redox potential (Eh) value was fixed at 0.2 V.

Table 1

Chemical composition of Syn-DB3 groundwater

2.2 Sample Preparation

Syn-DB3 was prepared to simulate the domestic groundwater sampled at the DB3 site in KURT (Table 1). Only the major components of the KURT DB3 groundwater were considered. Chemicals used for the preparation were NaF (99.99%), Ca(OH)₂ (99.995%), KCl (> 99.0%), Mg(OH)₂ (> 99.0%), Na₂SiO₃∙5H₂O (> 95%), NaCl (99.5%), CaSO₄ (99%), HF (99.99%) from Sigma-Aldrich. Syn-DB3 was prepared in air, indicating that the conditions were oxidative and the pH was 9.03.

²⁴¹Am (Amersham, 1986, YD8886, AMS 1LY) was purified using a cation exchange resin (AG 50W×8, Bio-rad, 200–400 mesh, 1.7 meq·ml⁻¹ resin bed), as described previously [13-14]. The ²⁴¹Am stock solution was prepared in 0.1 M HClO₄ solution, and the concentration was determined to be 5.7 mM by LSC and gamma- and alpha-spectrometry [13].

Americium samples in Syn-DB3 groundwater were prepared by diluting the Am(III) stock solution with Syn- DB3 in two steps. In the first step, 5.7 mM Am(III) stock solution in 0.1 M HClO₄ was diluted to 100 μM Am(III) in Syn-DB3. In the second step, 10 mL of 11.8 μM Am (III) in Syn-DB3 was prepared from the diluted solution. After the second dilution, the pH of Am(III) in the Syn-DB3 solution was 7.5. To vary the pH condition of the samples, the second dilutions were performed with addition of small volume (0.5–4.5 μL) of 0.2 M HClO₄ (ULTREX II, Ultrapure reagent, JT Baker) or 0.1 M NaOH (Semiconductor grade, 99.99% trace metals basis, Sigma-Aldrich). The pH of the samples was measured using a glass combination electrode (Orion^TM, Ross Ultra) when preparing them. Am(III) samples in Syn-DB3 were prepared under five different pH conditions: 6.4, 7.5, 8.4, 8.7, and 10.5. Am(III) solution (11.8 μM) in 0.1 M HClO₄ (pH 1) was also prepared from the first diluted solution as a reference sample. Two milliliters of each sample were transferred into a quartz cell with a screw cap for absorption measurements, and the remaining 8 mL of the sample was transferred into a centrifuge tube (Oak Ridge high-speed PPCO centrifuge tube, 10 mL, Nalgene) for quantification of Am(III) in supernatant after centrifugation. All the sample preparation procedures were performed in an argon-filled glove box to avoid CO₂ dissolution.

Precaution!²⁴¹Am is highly radioactive (specific activity = 126.9 GBq·g⁻¹). This should be handled carefully under proper protection in radiation-controlled areas. The concentrated ²⁴¹Am solution was handled in a fume hood or glovebox equipped with high-efficiency particulate air filters. During the sample purification and preparation process, the probable surrounding contamination was frequently checked using a gamma meter and a surface contamination detection method. The stability of the storage container of the ²⁴¹Am stock solution was an issue for long-term storage, as the erosion of plastic containers was observed over several years. This is likely due to the radicals generated by the alpha particles emitted from ²⁴¹Am. In particular, Teflon is weak against alpha particles, making it unsuitable for longterm storage. Currently, a quartz cell with a screw cap made of polyether ether ketone is used as a storage container for purified stock ²⁴¹Am.

2.3 UV-Vis Spectrophotometry

The absorption spectra of Am(III) were acquired using a spectrophotometer (Cary 5000, Agilent) equipped with a temperature controller. Quartz cells with beam path lengths of 1 cm were used. The absorption was scanned in the range of 525–495 nm with a spectral resolution of 0.1 nm, a spectral bandwidth of 0.5 nm, and an acquisition time of 1 s. Samples in the quartz cell were stored at room temperature (20–30℃ due to seasonal climate change) for aging-dependent measurements over 69 d, and the absorption measurements were conducted at 25℃ (deviation < ± 0.2℃).

2.4 TRLFS

The luminescence spectra of Am(III) samples were acquired using a TRLFS system [14, 16]. An Nd-YAG pumped OPO system (Vibrant B, OPOTEK Inc.) was employed as a wavelength-tunable nanosecond pulsed laser source. The excitation wavelength was chosen between 503.0 and 506.3 nm (2.3 mJ) as per the maximum absorption wavelength (λ_max) of each sample. Emissions were detected using an intensified CCD (ICCD) (iStar, Andor Technology) after a spectrograph (SR-303i, Andor Technology). All measurements were performed at 25℃ and controlled using a thermostat (Q-pod, Quantum Northwest).

2.5 Quantification of Am(III) in Supernatants

A liquid scintillation counter (LSC, TriCarb4910TR, PerkinElmer) was used to measure the Am(III) concentration in the solution. To quantify the Am(III) concentration in the supernatant of Am(III)_Syn-DB3 samples, the samples in centrifuge tubes were centrifuged at 18,000 rpm for 1 h. Ten microliters of each sample was taken from the upper part of the supernatant and mixed with 300 μL of 0.1 M HNO₃ in 20 mL polyethylene LSC vials to acidify Am(III), because LSC measurements for the solutions at high pH tend to be erroneous. The acidified Am(III) samples were mixed with 15 mL of LSC cocktail (Ultima Gold AB, PerkinElmer). The LSC was measured for 10 min using an alpha and beta discriminator.

The concentration of Am(III) was calculated using Equation (1) from the radioactivity (counts/s = Bq) obtained from LSC measurements.

where A is the measured radioactivity value, N_A is the Avogadro number (6.02×10²³/mole), and T_1/2 is the half-life (s) of ²⁴¹Am. The first daughter of ²⁴¹Am is ²³⁷Np (half-life = 2.14×10⁶ y), which undergoes alpha decay, but its half-life is 10⁴ times longer than that of ²⁴¹Am, so the radioactivity value emitted from the daughter was not taken into account.

3. Results and Discussion

3.1 Geochemical Modeling Results

Geochemical modeling was conducted to estimate the solubility curves and dissolved species distributions of americium under the Syn-DB3 condition in the pH range of 6–11 (Figs. 1 and 2). ThermoChimie_11a TDB from Andra was used as the input DB. In Tables 2 and 3, the thermodynamic data of the aqueous and solid americium species relevant to the Syn-DB3 groundwater conditions as in Figs. 1 and 2 are listed. Except for AmF₃, AmSiO(OH)₃²⁺ and AmCO₃OH(cr), all other thermodynamic data were adapted from OECD NEA Chemical Thermodynamic Data Vol. 5 [2] and Vol. 14 [6]. The crystalline phases of AmCO₃OH:0.5H₂O(s), AmCO₃OH(s), and Am₂(CO₃)₃(s) were not defined in ThermoChimie_11a, whereas they were defined as hydrated crystalline AmCO₃OH:0.5H₂O (cr, hyd), amorphous AmCO₃OH(am), and amorphous Am₂(CO₃)₃(am) in the NEA-TDB. In this study, we used the NEA-TDB notation. In the NEA-TDB project, Am(III) was exclusively reviewed as a representative actinide in the +3 oxidation state. They categorized Am(III), Cm(III), Nd(III), and Eu(III) in an analogue element in evaluating the chemical thermodynamic data, because their chemical behaviors are similar due to their close ionic radii (Am³⁺:1.10 Å, Cm³⁺:1.09 Å, Nd³⁺:1.11 Å, Eu³⁺:1.07 Å) [13]. However, the redox behavior and solubility of isostructural solid compounds can be considerably different for each element; thus, they must be evaluated separately [2].

Fig. 1

Geochemical modeling results of americium solubility in Syn-DB3 groundwater condition. ThermoChimie_11a TDB from Andra was used as input DB for the modeling in (a). Solubility curves evaluated excluding the formation of (b) crystalline AmCO₃OH(cr); (c) crystalline AmCO₃OH(cr), AmCO₃OH:0.5H₂O(cr, hyd), and Am(OH)₃(cr); and (d) crystalline AmCO₃OH(cr), AmCO₃OH:0.5H₂O(cr, hyd), Am(OH)₃(cr), and Am(OH)₃(am). In (c) and (d), experimentally determined americium concentrations in the supernatants of Am_Syn-DB3 are denoted with blue rectangles and red circles, which were measured after 5 and 70 d of sample preparations, respectively. The vertical dashed lines represent the dehydrogenation reactions of the carbonate ion (H₂CO₃ ⇆ HCO₃⁻ + H⁺ ⇆ CO₃²⁻ + 2H⁺).

Fig. 2

Geochemical modeling result of aqueous Am(III) species distribution in Syn-DB3 condition. Concentration of americium was set as 11.8 μM. As in Fig. 1(c), formation of crystalline AmCO₃OH(cr), AmCO₃OH:0.5H₂O(s), and Am(OH)₃(cr) was excluded.

Table 2

Thermodynamic data of aqueous Am(III) complexes relevant in KURT DB3 groundwater listed in ThermoChimie_11a

Table 3

Thermodynamic data of solid americium species relevant to KURT DB3 groundwater condition listed in ThermoChimie_11a

Experimentally determined solubilities tend to deviate considerably from each other and are highly dependent on the experimental conditions as well as the crystallinity of the solid compounds [2]. The dissolution of solid compounds at under-saturation conditions accompanies hydration of the surface in contact with the solution, by which the crystalline property is altered and finally becomes amorphous at the interface with the solution prior to dissolution. Thus, modeling results based solely on thermodynamic data may not accurately reflect reality. To consider crystallinitydependent solubility, modeling was conducted stepwise, excluding crystalline phases with low solubility in the order shown in Figs. 1(a)–1(d).

The modeling results under oxidizing and reducing conditions (Eh −0.5~0.5 V) were the same because americium is stable in the +3 oxidation state under these conditions. Under the Syn-DB3 condition at pH 9, the AmCO₃OH solid compound appeared as the major solubility-limiting phase, with crystallinity-dependent dissolved americium concentrations in the order of AmCO₃OH(cr) (~10⁻¹² M) < AmCO₃OH:0.5H₂O(cr, hyd) (~10⁻⁹ M) < AmCO₃OH(am) (~10⁻⁷ M) (Fig. 1). In the pH range of 7–10, the main dissolved chemical species were predicted to be Am(CO₃)⁺ and Am(CO₃)₂⁻. AmSiO(OH)₃²⁺ was distributed within 10% of the total (Fig. 2). Under weakly acidic conditions, the major chemical species were predicted to be Am³⁺, AmF₃, and AmF₂⁺, whereas Am(OH)₂⁺ species appeared under alkaline conditions.

3.2 Absorption Spectra of Am (III) in Synthetic Groundwater

The absorption spectrum of Am³⁺ under acidic conditions is well characterized by two major peaks at 503 nm (ε = 424 ± 8 cm⁻¹·M⁻¹) [13-14] and 813 nm (ε = 68 ± 4 cm⁻¹·M⁻¹) [10, 17-18]. The absorption spectra of Am(III) in the Syn-DB3 groundwater were monitored over 69 d (Fig. 3). The first measurement was performed 3 h after sample preparation. The Am(III) sample in 0.1 M HClO₄ (pH 1) showed the characteristic absorption spectra of aqua Am³⁺ centered at 503 nm, and there was no change throughout the observation period, confirming that our experimental conditions were stable for the duration. All the Am(III) samples in Syn-DB3 showed gradually red-shifted absorption spectra as pH increased; absorption peaks at 505.1 (pH 6.4), 506.0 (pH 7.5 and 8.4), 506.1 (pH 8.7), and 506.3 nm (pH 10.5). The absorption peak positions of the samples in the pH range of 7.5–8.7 were almost identical, and pH 10.5 also showed a very close peak position despite large differences in the pH values. These results suggest that the absorption peaks at approximately 506 nm originate from Am(III) colloidal particles. This is consistent with our previous findings on Am(III)-OH [13] and -CO₃ systems [19], where Am(III) colloidal particles formed under neutral to alkaline pH conditions showed absorption peaks at approximately 506 nm, but were not luminescent. Similarly, the absorption spectra of Am(III)-humic colloidal systems have also been reported to exhibit absorption peaks at ~506 nm [20-21].

Fig. 3

Aging time-dependent absorption spectra of the Am(III) samples in Syn-DB3 groundwater at various pH of 6.5–10.5. For the reference absorption spectrum, the pH 1 sample in 0.1 M HClO₄ was used.

All Am(III)_Syn-DB3 samples showed a rapid decrease in absorbance at 506 nm in 1 d, followed by a continuous decrease over 69 d (Fig. 4), but there was no change in the peak position of each sample (Fig. 3). These observations indicated that Am(III)-colloidal particles were initially formed when the samples were prepared, and the decreased absorbance was due to the precipitation of colloidal particles. It is noteworthy that pH 6.4 showed an absorption peak at a different position (505.1 nm) and residual absorbance was still observed after 69 d (Figs. 3 and 4). According to our previous work, ~10 μM Am(III) in 0.1 M NaClO₄ at pH ~6.4 showed the same absorption spectrum as in pH 1 solution [13-14]. Thus, the observed absorption at 505.1 nm in the sample at pH 6.4 suggests the formation of aqueous Am(III) complexes with ligands in Syn-DB3, such as AmCO₃⁺, as predicted by geochemical modeling (Fig. 1).

Fig. 4

Aging time-dependent absorption changes (absorption area between 505–515 nm) of the Am(III) in Syn-DB3 groundwater.

3.3 Luminescence Properties

The luminescence spectra of Am(III) in Syn-DB3 were acquired 1 d after the preparation. The Am(III) sample at pH 1 showed a luminescence peak at 688 nm, which is well known for aqueous Am³⁺ (Fig. 5) [22-23]. Am(III) in Syn-DB3 samples at pH ≥ 7.5 showed no detectable luminescence signals (Fig. 5), while they still showed detectable absorbance on the same day (1 d), as shown in Figs. 3 and 4. Generally, aqueous Am(III) complexes exhibit red-shifted luminescence with considerably increased intensity [14-15, 24-25], whereas colloidal particles do not exhibit luminescence properties [13-14]. Thus, the absence of luminescence from Am(III) in Syn-DB3 samples at pH ≥ 7.5, also indicates that the absorbance at 506 nm is due to the colloidal particles. However, the sample at pH 6.4 showed an additional peak at approximately 694 nm, although the overall luminescence intensity was almost half of that of the pH 1 sample. Luminescence centered at 694 nm has been reported for aqueous Am(III) complexes [14- 15]. Thus, the result additionally supports that a small portion of aqueous Am(III) complexes, possibly AmCO₃⁺, was distributed along with solid phases in the pH 6.4 sample. Further studies are needed to identify these aqueous species.

Fig. 5

Luminescence spectra of Am(III) in Syn-DB3 groundwater acquired 1 d after sample preparation.

3.4 Solubility of Americium in Syn-DB3 Groundwater

Am(III) concentrations in the supernatants of the samples were quantified by LSC measurements after centrifugation of the samples at 18,000 rpm for 1 h. Colloidal particles of small sizes might not have completely settled by centrifugation and contributed to concentrations in the supernatant. Therefore, the experimentally measured concentrations of dissolved americium in this study are slightly overestimated. On the day of preparation, the first measurements were conducted in 4.5 h. As shown in Fig. 6, dissolved americium concentrations in Syn-DB3 groundwater at pH range of 6.4–10.5 were measured to be significantly lower than that in americium in 0.1 M HClO₄ (1.18×10⁻⁵ M). Samples at pH 6.4 and 8.7, showed approximately 1/10^th concentrated americium in the supernatants, and samples at pH 7.5, 8.4, and 10.5, contained only 1/100^th concentrated americium in the supernatants. Five days after preparation, the americium concentrations in the supernatants decreased to 3×10⁻⁷ M for the pH 6.4 sample, 1×10⁻⁷ M for pH 7.5–8.7 samples, and 4×10⁻⁸ M for the pH 10.5 sample. Interestingly, americium concentration increased back to 3×10⁻⁶ M for the pH 6.4 sample at 70 d. The concentrations of samples at pH 7.5–10.5 also increased and converged to 2×10⁻⁷ M in 70 d. The increasing concentrations in the supernatant after 70 d suggest that initially colloidal particles were over-formed owing to the abrupt introduction of americium into the groundwater, and a slow equilibrium process occurred between the aqueous species and the precipitates as sample aging progressed.

Fig. 6

Sample aging time-dependent americium concentrations in supernatant.

In comparisons to modeling results, the measured dissolved americium concentrations at pH 8.4 and 8.7 were significantly higher than those predicted by formation of crystalline AmCO₃OH(cr) and AmCO₃OH:5H₂O(cr, hyd) (Figs. 1(a) and 1(b)). These values were comparable to those of AmCO₃OH(am) (Figs. 1(c) and 1(d)). Unfortunately, we were not able to set up experiments for characterization of the precipitates because the amount of americium in the stock was not sufficient for solid characterization. This is also the reason for performing the over-saturation experiments.

Under over-saturation conditions, fresh precipitates are initially amorphous, which can transform into more crystalline compounds with aging. Thus, it is reasonable that our experimentally determined solubility values follow the solubility of amorphous AmCO₃OH(am) rather than that of the crystalline phases (AmCO₃OH(cr) and AmCO₃OH:0.5H₂O(cr, hyd)). Interestingly, the solubility showed an increasing tendency over time rather than a decreasing trend (5 vs. 70 d), which implies that the amorphous solid state did not transform into a less soluble crystalline form, at least in our observation time window. This suggests that the transformation of the amorphous state to the crystalline state may be required to overcome the high energy barrier, which is too high to be overcome at ambient temperatures. However, temperature-dependent studies are required to test this hypothesis. Our results also underline that kinetic information on solid-state formation and maturation processes is needed for a more reliable prediction of the dissolved concentration and mobility of radionuclides.

The measured dissolved americium concentrations were lower at pH 6.4 and 7.5, than those for the simulation results, while it was higher at pH 10.5 than that for the simulation result. Furthermore, pH 6.4 and 10.5 samples showed increase in the dissolved americium concentrations by almost one order of magnitude in 70 d, by which the pH 6.4 sample approached the solubility line of Am₂(CO₃)₃(s) and the pH 10.5 sample to that of AmCO₃OH(am) instead of Am(OH)₃(am) (Fig. 1 (d)). Overall, samples at pH 7.5– 10.5 approached the AmCO₃OH(s) solubility in 70 d. As the colloidal americium particles and precipitates matured over time, their physicochemical properties, including particle size and structure, presumably changed, which could accompany the pH changes in the samples and thus cause equilibrium shifts. Indeed, we have often observed time-dependent pH changes in solutions containing americium colloidal particles in an Am(III)-OH system (data not shown). A more systematic and long-term study of americium solubility with close monitoring of solution conditions, including pH and temperature, is needed.

4. Conclusions

The solubility of americium was estimated under over-saturation conditions in Syn-DB3 groundwater. The absorption and luminescence spectroscopic data indicated that colloidal Am(III) particles initially formed and precipitated rapidly over a couple of days. The measured solubility of americium in the Syn-DB3 groundwater condition was comparable to that controlled by amorphous AmCO₃OH(am) instead of crystalline phases, when compared with the geochemical modeling results. Our results suggested that kinetic information on the transformation into or between different solid phases was an important factor for predicting the solubility of radionuclides. In addition, our study demonstrated that geochemical modeling results should be interpreted in conjunction with experimental results for a more reliable evaluation of radionuclide mobility. A long-term study of americium solubility with close tracking of the pH changes, tight control of temperatures, and better phase separation, such as ultrafiltration, was needed for a more reliable description of Am behavior in domestic groundwater.