A reliable prediction of the chemical behavior of actinides in aqueous solutions is necessary in performance assessment analysis of deep geological nuclear waste repositories. Although geological or geo-technical barrier systems may prevent or hinder formation water from contacting the waste, intrusion of aqueous solutions into a repository has to be taken into account for several scenarios on the long-term evolution of a repository. At a depth of several hundreds of meters, relevant argillaceous and granitic type formation waters are Na+/Cl− type solutions, characterized by anaerobic conditions and a pH in the range of 7 to almost 9 . Sedimentary and crystalline rock type formation waters are normally characterized by low ionic strength, although formation waters with elevated ionic strength are also found in sedimentary bedrocks in the Canadian Shield  and Cretaceous argillites in Northern Germany , among other examples. Fluid inclusions and brine pockets in rock salt formations are characterized by high ionic strength (Im > 5 m) dominated by high concentrations of mainly Na+, Mg2+, K+ and Cl−. Corrosion of cementitious waste forms in MgCl2 dominated brines may lead to high CaCl2 concentrations (≥2 M) and highly alkaline pHm (≈12) conditions .
High concentrations of nitrate (≥1 M) are expected in wastes from nuclear fuel reprocessing, although other sources/processes can be also responsible for elevated nitrate concentration/inventory in repositories for waste disposal [5, 6]. In the Waste Isolation Pilot Plant (WIPP), a deep underground salt mine for transuranic waste disposal (New Mexico, USA), the initial quantity of nitrate in waste is calculated to 2.74 × 107 mol . Very high nitrate concentrations in combination with hyperalkaline pH conditions are also expected in several waste tanks of the Hanford Site (Washington, USA) . Slow nitrate reduction kinetics expected at 25 °C and moderate H2 partial pressure  may affect the aqueous speciation of radionuclides and thus impact their mobilization into the biosphere.
Reducing conditions develop after the closure of a deep underground repository for nuclear waste disposal due to anoxic iron/steel container corrosion, producing dissolved Fe(II) species, Fe(II)/Fe(III) phases and hydrogen. In these reducing geochemical conditions, actinides are expected to prevail in the tri- and tetravalent redox states. Although the aqueous chemistry of AnIII/IV under repository-relevant conditions is dominated by hydrolysis reactions under absence of other strongly complexing ligands, the interaction of actinides with salt brines leads to unique geochemical boundary conditions affecting significantly actinide solubility and speciation . A number of recent studies have shown the formation of hitherto unknown ternary Ca–An–OH complexes in alkaline CaCl2 brines accompanied with an increase in AnIII/IV solubility [11–14]. The formation of complexes with other inorganic or organic ligands can also impact the behavior of actinides and promote its mobilization into the biosphere.
The last update book released within the OECD Nuclear Energy Agency – thermochemical database project (NEA-TDB)  selected only thermodynamic data for the complex AmNO32+, whereas no selection was provided for the system Pu(III)–NO3. Recent Cm(III)–TRLFS studies with nitrate concentrations up to 4.61 m proposed the formation of CmNO32+ and Cm(NO3)2+ complexes and provided both lg β° and ΔrH° data [16, 17]. So far all these studies focused on acidic conditions, leaving aside the assessment of nitrate effects under neutral to alkaline, repository relevant pH conditions.
The interaction of nitrate with metal cations is considered to be weak, but slightly stronger than in the corresponding chloride systems [10, 18, 19]. A challenge in the modelling of these systems arises from the distinction between formation of (weak) complexes and variations in the activity coefficients [19, 20]. Usually, large changes of the ionic medium are necessary to study weak complex formation, which additionally cause changes of the activity coefficients of the species involved. Note that the NEA-TDB guidelines for the extrapolation to zero ionic strength discourages the replacement of more than 10 % of the background electrolyte by the ligand when assessing the formation of weak complexes . Besides classical solution chemistry studies, the use of advanced spectroscopic techniques can help in the identification of new aqueous species, and thus provide key inputs in the definition of correct chemical models for complex systems.
This study aims at providing a comprehensive thermodynamic description of AnIII/LnIII–NO3 interaction under repository-relevant pH conditions in the absence of carbonate and T = 22 ± 2 °C. For this purpose, extensive batch solubility experiments were conducted with Nd(III) in dilute to concentrated NaCl–NaNO3, MgCl2–Mg(NO3)2 and CaCl2–Ca(NO3)2 mixed solutions. Complementary spectroscopic measurements [Cm(III)–TRLFS, Nd–LIII EXAFS] and a detailed solid phase characterization (XRD, SEM-EDS) were performed to gain detailed insights into the aqueous speciation and solid phases prevailing in the studied systems. The combination of these approaches allows the development of chemical, thermodynamic and Pitzer activity models for the system Nd3+/Cm3+–H+–Mg2+–OH−–Cl−–NO3−–H2O. Considering the widely accepted chemical analogy between NdIII/CmIII/AmIII/PuIII, this study provides the basis for an accurate assessment of the behavior of trivalent actinides in nitrate-rich wastes under repository-relevant conditions.
All solutions were prepared with ultrapure water, purified with a Milli-Q academic apparatus (Millipore, 18.2 MΩ·cm, 22 ± 2 °C, pore size 0.22 μm) and purged several hours with Ar before use. NaCl (p.a.), NaNO3 (p.a.), MgCl2·6H2O (p.a.), Mg(NO3)2·6H2O (p.a.), CaCl2·2H2O (p.a.), Ca(NO3)2·4H2O (p.a.), portlandite [Ca(OH)2(cr), p.a.], C8H17NO3S (CHES, p.a.), C8H18N2O4S (HEPES, p.a.), HCl (Titrisol®) and NaOH (Titrisol®) were obtained from Merck, C4H11NO3/C4H11NO3·HCl (TRIS/TRIS·HCl, p.a.) from Sigma-Aldrich, whereas brucite (Mg(OH)2(cr), BioUltra > 99 %) was purchased from Fluka. Nd(OH)3(s) was prepared by hydration of crystalline Nd2O3(cr) (p.a., Merck) in pure water under argon atmosphere. The complete solid phase transformation was confirmed by XRD (PDF 70-0215) [13, 21]. Mg-oxychloride [Mg2(OH)3Cl·4H2O(cr)] was synthesized by transformation of brucite in concentrated MgCl2 solutions (≥2.5 M) as described in .
A 2 × 10−5 M Cm(III) stock solution in 0.1 M HClO4 was used for TRLFS experiments. The isotopic composition of curium was 89.7 % Cm–248, 9.4 % Cm–246, 0.4 % Cm–243, 0.3 % Cm–244, 0.1 % Cm–245 and 0.1 % Cm–247.
The molal H+ concentration (pHm = −lg mH+) was determined with combination pH electrodes (type Orion Ross, Thermo Scientific). Calibration against pH (activity scale) standard buffers (pH 6–12, Merck) yields operational “measured” pHexp values in salt solutions of ionic strength Im > 0.1 m, with pHm = pHexp + Am. The empirical parameter Am includes the individual activity coefficient γH+ and a contribution ΔEJ entailing the difference in liquid junction potential Ej between dilute pH buffer solutions for calibration and samples with high concentration of background electrolyte. Empirical Am values for NaCl, MgCl2 and CaCl2 systems were previously reported elsewhere [11, 22]. Am values of pure nitrate systems and nitrate-chloride mixtures were experimentally determined in the present work with standard solutions containing 2 × 10−2–6.25 × 10−4 M HCl in 0.1–6.02 m NaCl–NaNO3, 0.25–5.24 m MgCl2–Mg(NO3)2 and 2.91/4.02 m CaCl2–Ca(NO3)2 mixtures.
Am values determined for NaCl–NaNO3, MgCl2–Mg(NO3)2 and CaCl2–Ca(NO3)2 solutions are provided in Figure 1 and Table 1, together with Am values available for NaCl, MgCl2 and CaCl2 systems as reported in literature . Am values for the different Cl−–NO3− mixtures considered in this work were based on linear correlation with total NO3− concentration.
Nd(III) solubility and solid phase characterization
All samples were stored and prepared in argon glove boxes under exclusion of oxygen (O2 < 5 ppm, T = 22 ± 2 °C) and CO2. All solutions were prepared volumetrically and the resulting molar concentrations were converted into the molal scale with the density values calculated using Pitzer equations with data reported elsewhere [23, 24]. Batch solubility experiments of Nd(III) were performed from undersaturation conditions with Nd(OH)3(s) as solid phase in Polyvials® (HDPE, Zinsser Analytic). Five series of solubility experiments were prepared with different total salt concentrations of NaCl–NaNO3 (0.1–6.02 m) and MgCl2–Mg(NO3)2 (0.25–5.24 m) mixtures, whereas only two series were prepared in the case of CaCl2–Ca(NO3)2 (2.85 and 4.02 m) mixtures (see Table 2). Na+, Mg2+ and Ca2+ concentrations were fixed within series of constant ionic strength and the concentrations of Cl− and NO3− were varied according to a fixed pattern (see Table 2). Nitrate concentrations ranged between 0 and 8.0 m per batch solubility experiment. The solubility of nitrate in the series 5.24 m MgCl2–Mg(NO3)2 was limited to mNO3− ≤ 2.32 m. For each ionic strength and Cl−/NO3− mixture, a set of (at least) three independent samples was prepared within 7.5 ≤ pHm ≤ 13.2, with 15 mL background electrolyte and 6–12 mg Nd(OH)3(s). The pH in the samples was adjusted by the addition of organic buffers (150 μL 1.0 M HEPES (pH 7.5–7.9), TRIS (pH 8.3–8.4) or CHES (pH 8.6–10.0) to reach a final buffer concentration of 10 mM in the samples), brucite/Mg-oxychloride, portlandite, HCl and NaOH. The impact of these organic buffers on AnIII/LnIII speciation was evaluated in a previous study . The solubility of brucite/Mg-oxychloride constrain the pHm within a range of 8.6 and 9.1 depending upon MgCl2 and nitrate concentration . In the calcium system, portlandite buffers the pHm to ∼12. The pHm controlled by these solid phases is denoted as “pHmax” for MgCl2–Mg(NO3)2 and CaCl2–Ca(NO3)2 systems in the following.
The pH and mNd(III) were systematically monitored up to 398 days. Nd(III) concentration in the aqueous solution was determined by ICP-MS (X-Series II, Thermo Scientific) after 10 kD (∼1.5 nm) ultrafiltration (Pall Life Sciences). Detection limit for Nd(III) is typically in the range of 10−9–10−10 M depending on background electrolyte concentration in the sample and required dilution steps. The analytical uncertainties of the ICP-MS analyses are 5–10 %. The outcome of these solubility studies was compared with previous solubility experiments in nitrate-free NaCl, MgCl2 and CaCl2 solutions .
After ensuring equilibrium conditions (constant pH and mNd(III)), about 2 mg of Nd(III) solid phase recovered from selected batch experiments by centrifugation were characterized by X-ray diffraction (XRD). Although equilibria were typically obtained within a few days, XRD were taken after >79 days. Powder samples were washed four times with ethanol to remove the background electrolyte which can interfere in the XRD analysis, resuspended in approximately 20 μL ethanol and placed on a single crystal silicon waver.
XRD measurements were performed using a D8 Advance diffractometer (Bruker AXS) equipped with a Cu radiation tube, Ni filter and a Sol-X detector, working at an X-ray source current of 25 mA and a voltage of 40 kV. Diffractograms were recorded in the range 5° ≤ 2Θ ≤ 60° with a step size of 0.04° 2Θ, 6 s counting time per step and variable slit widths. Measured reflexes were compared to XRD patterns of relevant phases of the JCPDS database .
A small fraction of the washed powder was further characterized by scanning electron microscope – energy-dispersive X-ray spectrometry (SEM-EDS). SEM-EDS analyses were carried out using a Quanta 650 FEG instrument (FEI) equipped with a Thermo Scientific UltraDry silicon drift X-ray detector. SEM images were collected at an electron acceleration voltage of 20 kV, whereas SEM-EDS measurements were done at 30 kV. Data analysis was performed using the NORAN System 7 X-ray microanalysis system for the Quanta 650 FEG instrument.
Nd–LIII EXAFS studies in MgCl2–Mg(NO3)2 solutions
EXAFS measurements were conducted on the supernatant of a solubility sample taken from the 4.05 m MgCl2–Mg(NO3)2. The sample was characterized by mNO3− = 5.81 m, pHm = 8.15 and mNd(III) = 1.49 × 10−3 m and was measured after 10 kD (∼1.5 nm) ultrafiltration to avoid the contribution of colloidal species in the EXAFS spectra. The sample was prepared in a 400 μL polyethylene vial, double-sealed in a plastic envelop inside an Ar-glovebox and transported to the beamline.
Nd–LIII edge X-ray absorption fine structure (XAFS) spectra were recorded at the INE Beamline for Actinide and Radionuclide Science at ANKA. Spectra were energy calibrated to the first inflection point in the XANES spectra of a Mn metal foil (6.539 keV), which was measured simultaneously. The XAFS signal was recorded at room temperature in fluorescence mode using a vortex Si-drift detector. Si<111> crystals were used in the double crystals monochromator, operating in fixed-exit mode. The parallel alignment of the crystal faces was detuned to ∼70 % of the maximum beam intensity at the beginning of each scan. The incident intensity was then held constant by means of a piezo-driven feedback system to the second crystal.
EXAFS fits were performed with D-Artemis, a program of the Demeter IFEFFIT package , using phase and amplitude data calculated for a 59 atom cluster (∼5.8 Å diameter sized centered on the individual metal cations). Feff6L delivered as standard with the package was replaced by Feff8.4 for these calculations. For magnesium atoms, single path scattering files for phase and amplitude were used. The k-range used in the modelling was [2.45–9.6 Å−1]. Fits were performed in the R-space [1.25–4.5 Å] using simultaneously the k1-, k2- and k3-weighted data. The structures used as model to fit to the data are Nd(OH)2NO3 [ICSD 63550]  and Nd(NO3)3(H2O)4(H2O)2 [ICSD 37181] .
Time resolved laser fluorescence spectroscopy (TRLFS) measurements were performed with 1 × 10−7 M Cm(III) per sample with 5.61/6.02 m NaCl–NaNO3, 0.25/4.05 m MgCl2–Mg(NO3)2 and 4.02 m CaCl2–Ca(NO3)2 as background electrolyte, containing 0–8.0 m NO3−. The pHm in both MgCl2–Mg(NO3)2 series was buffered to pHmax with brucite and Mg-oxychloride, respectively. In addition samples in 4.05 m MgCl2–Mg(NO3)2 were further titrated to pHm = 2.94 with 0.1 M HCl (of same ionic strength and composition). Samples in 5.61/6.02 m NaCl–NaNO3 and 4.02 m CaCl2–Ca(NO3)2 were buffered with TRIS or CHES to 7.93 ≤ pHm ≤ 8.95.
TRLFS measurements were performed using a Nd:YAG laser (Surelite II Laser, Continuum) pumping a dye laser (Narrowscan Dye Laser, Radiant Dyes). The repetition rate of the Nd:YAG laser was 10 Hz, with a maximum laser energy of 2 mJ. To filter out Rayleigh- and Raman scattering, as well as short-lived background fluorescence, the emission spectra were recorded in a range of 570–630 nm with a delay of 1 μs and in a time window of 1 ms. Emission spectra of each sample were integrated over 500 accumulations. Excitation of Cm(III) was performed at λex = 396.6 nm, which corresponds to the maximum spectral absorbance band of Cm3+(aq). The fluorescence emission was detected by a spectrograph (Shamrock A-SR-303i-B, Andor Technology) with a 300, 600 and 1200 lines/mm grating and an ICCD camera (iStar ICCD, Andor Technology).
The obtained spectra can be interpreted qualitatively by peak position and shape. For a quantitative interpretation, a peak deconvolution and knowledge of the respective pure component spectra is necessary. Single components were extracted by subtracting known pure component spectra from experimentally determined spectra in MgCl2–Mg(NO3)2 systems.
Results and discussion
Solubility of Nd(III) in dilute to concentrated NaCl–NaNO3, MgCl2–Mg(NO3)2 and CaCl2–Ca(NO3)2 solutions
Figure 2 shows examples of the experimental solubility data of Nd(OH)3(s) in 0.10, 0.51, 1.02, 2.64 and 5.61 m NaCl–NaNO3 mixtures obtained in this study (for all data see appendix, Table A.1). Figure 3 shows exemplarily the experimental solubility data of Nd(OH)3(s) obtained in 2.80 m/4.05 m MgCl2–Mg(NO3)2 mixtures with 1.1 m ≤ mNO3− ≤ 5.8 m and 4.02 m CaCl2–Ca(NO3)2 mixtures with mNO3− = 5.75 m. Solubility data in 0.25 and 1.04 m MgCl2–Mg(NO3)2, as well as data in 2.80 and 4.05 m MgCl2–Mg(NO3)2 with mNO3− < 1.1 m are provided in Fig. A.1 in the appendix. Figures 2 and 3 also show the experimental solubility data reported by Neck et al.  in nitrate-free NaCl, MgCl2 and CaCl2 solutions under analogous pH and ionic strength conditions (except for the system 1.02 m NaCl–NaNO3, which is compared to 0.51 m NaCl reference data). Solubility curves of Nd(OH)3(s) calculated with the thermodynamic and activity models reported by the same authors are also appended in the figures for comparison purposes.
No significant effect of nitrate on the solubility of Nd(OH)3(s) is observed for any of the studied NaCl–NaNO3 systems. On the contrary, these solubility data show a very good agreement with Nd(III) solubility previously reported in nitrate-free NaCl solutions of analogous pH and ionic strength , which clearly indicate that no Nd(III)–NO3 complexation affects solubility in Na–systems with 7.5 ≤ pHm ≤ 13.2.
Nitrate has a negligible effect on the solubility of Nd(III) for the systems 0.25 and 1.04 m MgCl2–Mg(NO3)2 with nitrate concentrations up to 2.14 m (see appendix, Table A.2). These data are in good agreement with the previously reported Nd(OH)3(s) solubility in nitrate-free MgCl2 solutions of analogous pH and ionic strength . However, a clear increase in Nd(III) solubility occurs for 2.80 m and 4.05 m MgCl2–Mg(NO3)2 mixtures and pHm 8–9 (Fig. 3a–f) compared to the nitrate-free reference data by Neck and co-workers . The increase in solubility becomes more pronounced with increasing nitrate concentration and/or ionic strength. The maximum increase in Nd(OH)3(s) solubility (about one order of magnitude) is observed in the 4.05 m MgCl2–Mg(NO3)2 mixture with mNO3− = 5.81 m (Fig. 3f). The influence of nitrate complexation on the Nd(OH)3(s) solubility can be shown by comparing experimentally determined Nd(III) concentrations at pHm = 8.80 in 4.05 m MgCl2–Mg(NO3)2, see Fig. 4. There is a clear and systematic trend to higher Nd(III) solubility with increasing nitrate concentration, e.g., mNd(III) = −4.49 m at mNO3− = 0.24 m, mNd(III) = −4.14 m at mNO3− = 0.58 m, mNd(III) = −3.93 m at mNO3− = 1.16 m, mNd(III) = −3.72 m at mNO3− = 2.33 m and mNd(III) = −3.50 m at mNO3− = 5.81 m (Fig. 4). These findings hint to the likely formation of Nd(III)–NO3 complexes under near-neutral to weakly alkaline pH conditions.
The slope analysis of the experimental solubility data in MgCl2–Mg(NO3)2 mixtures shows two well-defined pH regions with slope −2 (pHm ≤ 8.3) and −1 (pHm > 8.3). This indicates the uptake of 2 and 1 H+, respectively, in the chemical reaction controlling the solubility of Nd(III) in these pH regions. Based upon a system with solubility controlled by Nd(OH)3(s) (see next section), the formation of the aqueous species with stoichiometries Nd:OH 1:1 and 1:2 is proposed.
In contrast to the MgCl2–Mg(NO3)2 system, no significant enhancement in Nd(OH)3(s) solubility occurs in 2.85 m (see appendix, Table A.3) and 4.02 m CaCl2–Ca(NO3)2 mixtures with nitrate concentrations up to 5.75 m compared to nitrate-free CaCl2 systems under virtually the same experimental conditions . This observation suggests the likely participation of Mg2+ in the formation of Nd(III)–NO3 complexes in concentrated nitrate bearing MgCl2–Mg(NO3)2 solutions and pHm 8–9.
Solid phase characterization in Na–Mg–Cl–NO3 systems
In all evaluated Na–Mg–Cl–NO3 systems with mCl− ≤ 5.82 m, the diffractograms of solid phases recovered from selected solubility experiments match the XRD pattern of the initial Nd(OH)3(s) powder (Fig. 5) and are in excellent agreement with the Nd(OH)3(s) reference diffractogram in  (PDF 70-0215). Besides Nd(OH)3(s), reflexes belonging to brucite (PDF 74-2220; ) and Mg-oxychloride (PDF 07-0412; ) initially added to the samples are clearly identified in those solid phases recovered from experiments with pHm = pHmax (data not shown).
A number of solid phases recovered from the solubility experiments in MgCl2–Mg(NO3)2 solutions were also analyzed by scanning electron microscopy. SEM-EDS analysis show a considerable enrichment of chlorine in Nd(III) solid phases equilibrated with solutions at mMg2+ ≥ 4.05 m and mCl− ≥ 5.82 m. These results hint towards a solid phase transformation in concentrated chloride brines. A comprehensive study evaluating the formation and stability of Nd(III)–OH–Cl(s) phases is ongoing at KIT–INE.
Cm(III)-TRLFS studies in NaCl–NaNO3, MgCl2–Mg(NO3)2 and CaCl2–Ca(NO3)2 solutions
Figure 6a–e shows a selection of experimentally determined Cm(III)–TRLFS spectra in 5.61 m NaCl, 6.02 m NaNO3, 4.05 m MgCl2, 4.05 m MgCl2–Mg(NO3)2 and 4.02 m CaCl2–Ca(NO3)2 systems. The Cm3+(aq) reference spectrum measured in a dilute acidic solution is shown in all the figures. All other Cm(III) spectra collected in this work for MgCl2–Mg(NO3)2 systems are provided in Fig. B.1 in the appendix.
In the absence of other complexing ligands, a number of binary Cm(III) hydroxide, chloride and nitrate species are expected to be present in solution under acidic to weakly alkaline pH conditions. Their relative contribution depends on the pH, ionic strength and electrolyte composition. Wavelengths reported for binary Cm(III)–OH, Cm(III)–Cl and Cm(III)–NO3 species have been included in the figures for comparison purposes [12, 16, 32–34].
The second Cm(III) hydrolysis species dominate in 5.61 m NaCl (Fig. 6a) over the investigated pH range (between 7.95 and 8.95), whereas binary Cm(III)–NO3 species prevail in 6.02 m NaNO3 and pHm = 7.93 (Fig. 6b). At pHm ≥ 8.93, Cm(OH)2+ becomes also predominant in 6.02 m NaNO3 (Fig. 6b). The definition of additional species is not necessary to explain Cm(III) spectra collected in NaCl–NaNO3 solutions, except some minor chloro-species in 5.61 m NaCl.
Cm(III) spectra collected in 4.05 m MgCl2 (Fig. 6c) at 2.95 ≤ pHm ≤ 8.45 are dominated by the Cm3+ aquo ion and binary Cm(III)–Cl species, whereas the second hydrolysis species of Cm(III) increases with pHm and dominates at pHm > 8.45. In the case of the 4.05 m MgCl2–Mg(NO3)2 system (Fig. 6d), binary Cm(III)–NO3 species dominate the spectra in the pH range 2.94 ≤ pHm ≤ 8.14. Besides the binary Cm(III)–NO3 species, additional Cm(III) complexes with a very clear pH dependency form in this system. Peak deconvolution confirms the presence of two new components at λ = 602.7 nm and λ = 606.0 nm. These species are absent in pure MgCl2 solutions of the same ionic strength, which clearly indicates the participation of nitrate in the complex formation. The component at λ = 602.7 nm has been assigned to the moiety “CmNO3OH+,” based on λ(Cm(NO3)2+) = 602.2 nm  and λ(Cm(OH)2+) = 603.5 nm [12, 32, 33], and considering that stronger ligands (e.g., OH− > NO3−) induce greater red-shift in the Cm(III) spectra than weaker ones. The strongly red-shifted Cm(III) shoulder at λ = 606.0 nm has been assigned to the moiety “CmNO3(OH)2(aq),” taking also as a reference the wavelength reported for the complex Cm(OH)3(aq) (λ = 607.5 nm) .
The spectra collected in 4.02 m CaCl2–Ca(NO3)2 (Fig. 6e) differ significantly from the 4.05 m MgCl2–Mg(NO3)2 system and can be explained considering mainly the formation of binary Cm(III)–NO3 species with minor contributions of chloro and hydroxo complexes. No mixed Cm(III)–NO3–OH complexes are observed in relevant concentrations at pHm ≤ 8.63. The absence of these species in CaCl2–Ca(NO3)2 solutions under the same experimental conditions (pHm, Im, mNO3−) as in the MgCl2–Mg(NO3)2 systems hints towards the participation of Mg2+ in the complexation reaction. Consequently, the formation of the quaternary aqueous species Mg[CmNO3OH]3+ and Mg[CmNO3(OH)2]2+ is proposed and has been considered in the thermodynamic modelling summarized below. Note that under mildly alkaline conditions (pHm > 10) and absence of nitrate, Neck and co-workers  proposed the formation of the ternary species Ca[Cm(OH)3]2+ holding the same ligand stoichiometry and charge. These Cm(III)–TRLFS observations are consistent with Nd(III) solubility data under analogous experimental conditions, and thus confirm that the observed increase in solubility for Nd(OH)3(s) is not caused by ion interaction processes but is related to a genuine Cm(III) complexation reaction with nitrate.
All pure component spectra derived by peak deconvolution using the complete set of TRLFS spectra collected in 4.0 m MgCl2–Mg(NO3)2 mixtures with 2.93 ≤ pHm ≤ 8.95 and 0.00 ≤ mNO3− ≤ 8.04 are shown in Fig. 6f.
Nd–LIII EXAFS studies in MgCl2–Mg(NO3)2 solutions
Figure 7 shows the k2-weighted Nd–LIII Fourier transform EXAFS spectrum collected for the supernatant (after 10 kD ultrafiltration) of the Nd(III) solubility sample at pHm = 8.15, 4.05 m MgCl2–Mg(NO3)2 with 5.81 m NO3− (see Fig. 3f). The structural parameters resulting from EXAFS evaluation are shown in Table 3. The best fit in the FT range [1.25–4.5 Å] is obtained using 6 shells. In addition to O and N shells, a shell with Mg2+ as backscatterer is needed to fit the peak at ∼4 Å. Furthermore, no Nd–Nd interaction can be evidenced within the R range taken into account. The results point out the presence of one to two nitrate groups, more likely a mixture of species having one and two nitrate groups where at least one of these species has Mg2+ in the direct vicinity of OH- and/or NO3-groups. Further these results are in good agreement with Cm(III)–TRLFS observations in MgCl2–Mg(NO3)2 mixtures (see Fig. 6d) where under the EXAFS sample characteristics a mixture of binary and quaternary nitrate-bearing aqueous species is expected. In conclusion, Nd–LIII EXAFS confirms the formation of quaternary Mg–Nd(III)–NO3–OH inner-sphere complexes with participation of Mg2+ and the absence of polymeric/colloidal Nd(III) species in solution.
Chemical, thermodynamic and activity models
The thermodynamic model derived in  for Ln(III) and An(III) in NaCl, MgCl2 and CaCl2 solutions is extended in the present work to nitrate-bearing systems by combining the results from solubility data, spectroscopic techniques and solid phase characterization. Nd(OH)3(s) is confirmed by XRD and SEM-EDS as solid phase controlling the solubility in all the evaluated systems with mCl− ≤ 5.82 m. The formation of the aqueous species Mg[AnIII/LnIIINO3OH]3+ and Mg[AnIII/LnIIINO3(OH)2]2+ is proposed based on slope analysis, Cm(III)–TRLFS and Nd–LIII EXAFS. Hence, equilibrium reactions (1) and (2) can be defined, in combination with equations (3)–(6) for the calculation of the corresponding lg*K′s,(1,1,n,1) and lg*K°s,(1,1,n,1) with n = 1 or 2.
Provided the very high ionic strength in most of the evaluated systems where the effect of nitrate is observed (Im ≥ 8.52 m), the Pitzer formulism has been preferred for the thermodynamic modelling . Parameters reported in [23, 24] are used for the calculation of aw and the activity coefficients of Mg2+, Cl−, NO3− and H+. Pitzer ion interaction coefficients for Nd(III) species derived in chloride media are reported in . Conditional equilibrium constants reported for binary Cm(III)–NO3 species at T = 25 °C and 0.10 m ≤ mNO3− ≤ 4.61 m (I ≤ 4.61 m)  have been used in the present work to determine the corresponding lg β° and ion interaction parameters using the Pitzer approach, see Table 4. For the new species Mg[AnIII/LnIIINO3OH]3+ and Mg[AnIII/LnIIINO3(OH)2]2+, only the binary parameters β(0) and the corresponding stability constant at I = 0 were calculated. Typical values for analogous cation-anion pairs of the same valence type are used for β(1), whereas CΦ was set to 0 [37, 38]. Ternary Pitzer parameters are generally set to 0. The difference between experimental and modelled solubility was minimized by optimizing β(0) and lg*K°s,(1,1,n,1) for the species Mg[AnIII/LnIIINO3OH]3+ and Mg[AnIII/LnIIINO3(OH)2]2+. A relevant constrain in the modelling exercise is the assumption that ion interaction parameters of cations with Cl− and NO3− are the same. This assumption is based on the very similar solubility of Nd(OH)3(s) in NaCl–NaNO3 and CaCl2–Ca(NO3)2 mixtures compared to the nitrate-free reference systems  and the very similar SIT ion interaction parameters reported for +1 to +4 monomeric cations  (see Fig. C.1 in the appendix). This permits to importantly decrease the number of unknown parameters whilst avoiding relevant variations in the activity coefficients for mixed solutions of same ionic strength but different Cl−/NO3− ratio.
Only solubility data collected for 0.25, 1.04, 2.80 and 4.05 m MgCl2–Mg(NO3)2 were considered for the optimization exercise, consistently with the validity of the thermodynamic model derived in . Note that solubility samples in pure 4.05 m Mg(NO3)2 were disregarded in the calculations due to alterations in XRD pattern and SEM-EDS characterization, with respect to the original Nd(OH)3(s) material. Stability constants and Pitzer ion interaction coefficients resulting from the optimization exercise are shown in Tables 4 and 5, respectively, together with the constants and parameters of all other species and solid phases of relevance in the system Ln3+/An3+–H+–Na+–Mg2+–Ca2+–OH−–Cl−–NO3−–H2O. Solubility curves and underlying aqueous speciation calculated with the thermodynamic and activity models summarized in Tables 4 and 5 are shown in Figs. 2, 3 and Fig A.1.
Summary and conclusions
The effect of nitrate on the solubility of AnIII/LnIII was studied in dilute to concentrated NaCl–NaNO3, MgCl2–Mg(NO3)2 and CaCl2–Ca(NO3)2 solutions using batch solubility experiments, Cm(III)–TRLFS, Nd–LIII EXAFS and extensive solid phase characterization (XRD and SEM-EDS), in order to derive a robust thermodynamic description of repository relevant aqueous systems.
Nitrate has a significant influence on the solubility of Nd(OH)3(s) in concentrated weakly alkaline MgCl2–Mg(NO3)2 solutions with total salt concentration ≥ 2.83 m and mNO3− ≥ 1.13 m. However, no effect of nitrate is observed in NaCl–NaNO3 and CaCl2–Ca(NO3)2 mixtures under analogous experimental conditions, thus indicating the relevant role of Mg2+ in the interaction between Nd(III) and nitrate. Cm(III)–TRLFS and Nd–LIII EXAFS confirm the participation of Mg2+ in the formation of quaternary inner-sphere complexes of the type Mg–AnIII/LnIII–NO3–OH.
The combination of slope analyses, TRLFS, EXAFS and solid phase characterization permits to confirm the relevance of the equilibrium reactions AnIII/LnIII(OH)3(s) + H+ + NO3− + Mg2+ ⇔ Mg[AnIII/LnIIINO3(OH)2]2+ + H2O and AnIII/LnIII(OH)3(s) + 2H+ + NO3− + Mg2+ ⇔ Mg[AnIII/LnIIINO3OH]3+ + 2H2O in the control of the solubility of AnIII/LnIII in concentrated MgCl2–Mg(NO3)2 brines. Based on the newly generated data, the chemical, thermodynamic and activity models described in  for Ln(III) and An(III) are further extended to Ln3+/An3+–H+–Na+–Mg2+–Ca2+–OH−–Cl−–NO3−–H2O systems.
To our knowledge, this is the first comprehensive experimental study providing a quantitative thermodynamic description of the effect of nitrate on the aquatic chemistry of Ln(III) and An(III) under repository-relevant pH conditions. This study also highlights the key role of spectroscopic techniques in defining correct chemical models of complex systems, especially when involving the formation of “weak complexes” in contraposition to matrix effects or pure ion interaction processes.
The authors highly appreciate the technical contribution by M. Böttle, N. Finck, F. Geyer and E. Soballa of KIT-INE. The valuable comments by V. Metz (KIT-INE) and three anonymous reviewers significantly contributed to the improvement of this manuscript are gratefully acknowledged.
V. Metz, H. Geckeis, E. Gonzalez-Robles, A. Loida, C. Bube, B. Kienzler. Radiochim. Acta100, 699 (2012).Google Scholar
S. K. Frape, P. Fritz. Geochim. Cosmochim. Ac.48, 1617 (1984).Google Scholar
W. Brewitz. GSF-T 114, Ges. f. Strahlen- u. Umweltforschung, Braunschweig (1980).Google Scholar
C. Bube, V. Metz, E. Bohnert, K. Garbev, D. Schild, B. Kienzler. Phys. Chem. Earth64, 87 (2013).Google Scholar
A. P. Paiva, P. Malik. J. Radioanal. Nucl. Chem.261, 485 (2004).Google Scholar
B. A. Crawford, S. A. Lott, L. D. Sparks, G. Van Soest, B. McInroy. In Waste Management 2006 Conference, WM’06 Conference, Tucson, Arizona (2006).Google Scholar
D. C. Kicker, C. G. Herrick. WIPP:1.2.5:PA:QA-L:559199, Carlsbad (2013).Google Scholar
R. C. Philip Hill. WRPS-49044-FP, Richland (2011).Google Scholar
M. Altmaier, X. Gaona, T. Fanghänel. Chem. Rev.113, 901 (2013).Google Scholar
M. Altmaier, V. Neck, T. Fanghänel. Radiochim. Acta96, 541 (2008).Google Scholar
T. Rabung, M. Altmaier, V. Neck, T. Fanghänel. Radiochim. Acta96, 551 (2008).Google Scholar
V. Neck, M. Altmaier, T. Rabung, J. Lützenkirchen, T. Fanghänel. Pure Appl. Chem.81, 1555 (2009).Google Scholar
D. Fellhauer, V. Neck, M. Altmaier, J. Lützenkirchen, T. Fanghänel. Radiochim. Acta98, 541 (2010).Google Scholar
R. Guillaumont, T. Fanghänel, V. Neck, J. Fuger, D. A. Palmer, I. Grenthe, M. H. Rand. Chemical Thermodynamics Vol. 5. Update on the Chemical Thermodynamics of Uranium, Neptunium, Plutonium, Americium and Technetium, Elsevier, Amsterdam (2003).Google Scholar
A. Skerencak, P. J. Panak, W. Hauser, V. Neck, R. Klenze, P. Lindqvist-Reis, T. Fanghänel. Radiochim. Acta97, 385 (2009).Google Scholar
L. Rao, G. Tian. Dalton Trans.40, 914 (2011).Google Scholar
W. Hummel. Report AN-44-01-28, Villigen (2001).Google Scholar
W. Hummel, G. Anderegg, L. Rao, I. Puigdomenech, O. Tochiyama. Chemical Thermodynamics Vol. 9. Chemical Thermodynamics of Compounds and Complexes of U, Np, Pu, Am, Tc, Se, Ni and Zr with Selected Organic Ligands, Elsevier, Amsterdam (2005).Google Scholar
I. Grenthe, F. Mompean, K. Spahiu, H. Wanner. Guidelines for the Extrapolation to Zero Ionic Strength, OECD-NEA, Issy-les-Moulineaux (2013).Google Scholar
M. Altmaier, V. Metz, V. Neck, R. Müller, T. Fanghänel. Geochim. Cosmochim. Acta67, 3595 (2003).Google Scholar
M. Weiser, T. Rabung, H. Geckeis. In 13thInternational Conference on the Chemistry and Migration Behaviour of Actinides and Fission Products in the Geosphere, MIGRATION’11, pp. 62–63, Beijing, China (2011).Google Scholar
JCPDS. Powder diffraction files. Joint Commitee on Powder Diffraction Standards, Swarthmore, USA (2001).Google Scholar
B. Ravel, M. Newville. J. Synchrotron Radiat.12, 537 (2005).Google Scholar
D. J. Rogers, N. J. Taylor, G. E. Toogood. Acta Cryst.39, 939 (1983).Google Scholar
G. Isetti. Period. Mineral.34, 327 (1965).Google Scholar
P. M. de Wolff. Technisch Physische Dienst, Delft (1957).Google Scholar
T. Fanghänel, J. I. Kim, P. Paviet, R. Klenze, W. Hauser. Radiochim. Acta66/67, 81 (1994).Google Scholar
T. Fanghänel, J. I. Kim. J. Alloy Compd.271–273, 728 (1998).Google Scholar
T. Fanghänel, J. I. Kim, R. Klenze, Y. Kato. J. Alloy Compd.225, 308 (1995).Google Scholar
I. I. Diakonov, B. R. Tagirov, K. V. Ragnarsdottir. Radiochim. Acta81, 107 (1998).Google Scholar
T. Könnecke, T. Fanghänel, J. I. Kim. Radiochim. Acta76, 131 (1997).Google Scholar
I. Grenthe, A. Plyasunov. Pure Appl. Chem.69, 951 (1997).Google Scholar
M. H. Rand, J. Fuger, I. Grenthe, V. Neck, D. Rai. Chemical Thermodynamics Vol. 11. Chemical Thermodynamics of Thorium, OECD Publications, Paris (2008).Google Scholar
V. Neck, T. Fanghänel, J. I. Kim. Report FZKA 6110, Karlsruhe (1998).Google Scholar
The online version of this article (DOI: 10.1515/pac-2014-1205) offers supplementary material, available to authorized users.