All experiments were conducted at T = (22 ± 2) °C in Ar gloveboxes with O₂ < 2 ppm. Solutions were prepared with purified water (Millipore Milli-Q Advantage A10 (18.2 MΩ·cm at 22°C, 4 ppb TOC) with Millipore Millipak^® 40 0.22 µm), purged with Ar for >1 h to remove traces of O₂ and CO₂.

Glutaric acid (C₅H₈O₄, 99%, solid), α-hydroxyisobutyric acid (C₄H₈O₃, 99%, solid), 3-hydroxybutyric acid (C₄H₈O₃, 99%, liquid), portlandite (Ca(OH)₂ (cr)) and tin (II) chloride (SnCl₂, p.a) were purchased from Sigma-Aldrich. NaOH, HCl (both Titrisol), hydroquinone (C₆H₆O₂, HQ, p.a.), concentrated HNO₃ (Supra- and Ultrapure) were obtained from Merck.

⁶³Ni and ¹⁵²Eu radiotracers were purchased from Eckert and Ziegler, with total activities of 100 μCi/3.7 MBq in 1 mL of 0.5 and 0.1 M HCl, respectively. Although both radiotracers were commercialized as “carrier-free”, the concentrations of stable Ni in the ⁶³Ni stock solution and stable Eu in the ¹⁵²Eu stock solutions were quantified as (3.0 ± 0.3)∙10⁻⁵ M and (1.1 ± 0.1)∙10⁻⁵ M. The isotopic composition of the Pu(VI) stock used in this work was 99.4 wt% ²⁴²Pu, 0.58 wt% ²³⁹Pu, 0.005 wt% ²³⁸Pu and 0.005 wt% ²⁴¹Pu, as characterized in a previous study (Altmaier et al., 2008). The original Pu(VI) stock solution ([Pu]_tot = 5.63·10⁻² M) in 3 M HClO₄ was diluted to [Pu]_tot = 2.64·10⁻⁶ M in 0.2 M HCl. The latter solution was used for the preparation of the sorption samples containing plutonium. Pu(IV) and eventually Pu(III) are expected to prevail in the reducing conditions of the experiments (see Tasi et al., 2018a), but the choice of the +VI oxidation state for the spike of Pu aimed at avoiding high initial oversaturation conditions that may result in the formation of Pu(IV) colloids (see Tasi et al., 2021a).

The hardened cement paste used in this work was provided by SKB as a monolith, which was prepared from a mixture of Swedish structural Portland cement for civil engineering (CEM I 42.5N BV/SR/LA) and deionized water at a water-to-cement weight ratio of 0.5. In a previous study (Tasi et al., 2021a), the HCP monolith was milled, sieved to a particle size of <100 μm, extensively characterized and stored under Ar atmosphere until use in the sorption experiments. Tasi and co-workers reported a (Brunauer–Emmett–Teller, BET) surface area of 79.2 m²⋅g⁻¹ for the HCP powder material used in the present study.

The cement porewater was prepared using the protocol described in Tasi et al. (2021a). The HCP was weighted in a 2 L Kautex flask and contacted with Milli-Q water to reach a solid-to-liquid ratio (S:L) of 25 g·dm⁻³. After a contact time of 1 week, the initial leachate was separated by decantation from the HCP, and the solid material was equilibrated again with fresh Milli-Q water during one additional week. The same HCP material and cement pore water were used in the solubility and sorption studies with proxy ligands for UP2W degradation products (Szabo et al., 2022a; Szabo et al., 2022b). The composition of the cement porewater is buffered by portlandite at pH ≈ 12.5 and [Ca] ≈ 0.02 M, and is characterized by a low content of Na and K (≤3⋅10⁻⁴ M). A detailed description of the impact of the proxy ligands investigated in this work on the HCP is provided in our previous study dedicated to this binary system (Szabo et al., 2022a).

The pH was measured using a ROSS combination pH electrode (Thermo Scientific, OrionTM) calibrated against standard buffer solutions (pH = 2 to 12, Merck).

Redox potentials of plutonium systems containing either HQ or Sn(II) were measured using Pt combination electrodes (Metrohm) with Ag/AgCl as a reference system. Measurements were performed following the protocol described in the literature (Altmaier et al., 2010). The measured potentials were converted to E _h values (referred to the standard hydrogen electrode, SHE) by correcting for the potential of the Ag/AgCl reference in 3.0 M KCl at T = 22°C, i.e. +207 mV. E _h values were converted to pe (with pe = −log a _e−) as E _h = −RT ln (10) F⁻¹ log a _e−, where R is the ideal gas constant (8.3144 J⋅mol⁻¹⋅K⁻¹) and F is the Faraday constant (96,485.3 C mol⁻¹). Uncertainties of pH and E _h measurements (calculated as 2σ of averaged measurements) ranged between ±0.05/±0.10 and ±15/± 40 mV, respectively.

All batch sorption samples were prepared using the HCP and cement porewater described in Section 2.2. Stock solutions of the proxy ligands were titrated with 1.0 M NaOH to pH ≈ 12.5 in order to avoid any impact on the pH of the cement porewater.

Suspensions of HCP and cement porewater (10 mL) at the targeted S:L ratio were equilibrated for 2 days before the addition of the corresponding radionuclide. In the experiments with plutonium, the HCP was mixed with cement porewater containing 2 mM of either HQ or Sn(II), and the suspension was equilibrated for 2 days, after which pH and E _h were determined as described in Section 2.3. Small volumes (14–351 μL) of the diluted radionuclide stock solutions were pipetted into these systems, aiming at log[⁶³Ni(II)]₀ = −8/−9, log[¹⁵²Eu(III)]₀ = −8/−8.67 and log[²⁴²Pu(IV)]₀ ≈ −8. After the addition of the corresponding radionuclides, the values of pH, E _h (only in the case of Pu) and total radionuclide concentration [RN]_aq, in the aqueous phase were monitored for ≤4 months [RN]_aq was measured after phase separation, i.e., ultrafiltration with 10 kD filters at 3,400–3,900 g (Nanosep^®, Pall Life Sciences; pore size ≈ 2-3 nm) (Ni, Eu, Pu) or ultracentrifugation (694,000 g, 1 h; Beckman XL-90, rotor type 90Ti) (Eu). After each sampling, the volume of the original sample was restored with cement pore water. The aliquots obtained after phase separation were diluted with 2% HNO₃ (Suprapur). In the samples containing nickel, the acidified aliquot was mixed with 10 mL of liquid scintillation counting (LSC) cocktail (Ultima Gold XR, Perkin Elmer) and measured for 30 min in a LSC Tri-Carb 3110 TR (Perkin Elmer) equipment considering an energy window between 0 and 100 keV. Detection limits achieved with this counting time and sample characteristics were ≈6⋅10⁻¹² M, calculated as three times the standard deviation of blank solutions. The concentration of Eu was determined by gamma counting (Wizard2 10—Detector Gamma Counter, Perkin Elmer) for an energy window between 15 and 2000 keV and a counting time of 30 min. Detection limits calculated for these conditions were ≈10⁻¹³ M. Pu was quantified by sector field inductively coupled plasma mass spectrometry (SF-ICP-MS; type Element XR, Thermo Fisher), with a detection limit of ≈10⁻¹³ M in the conditions of the present study. The solutions in the HCP-Ni system were additionally quantified for total nickel concentration by ICP-MS.

The same sequence was used in the preparation of the sorption samples in the ternary systems HCP-RN-proxy ligands. Two days after the addition of the corresponding radionuclide, the suspensions were spiked with appropriate volumes of GTA, HIBA and HBA stock solutions (0.2, 5·10⁻² and 5·10⁻³ M) targeting the range of initial ligand concentrations −4 ≤ log([L]₀/M) ≤ −1. Note that the sorption of the proxy ligands on HCP at the S:L ratio used in the ternary system HCP-RN-proxy ligands is very minor (Szabo et al., 2022a), and thus [L]₀ is considered equal to the concentration of the ligand in equilibrium conditions [L]_aq. After the addition of the proxy ligands, the values of pH, E _h (only in the case of Pu) and [RN]_aq were monitored for ≤3 months. The sampling procedure was the same as described above for the binary system HCP-RN. After each sampling, the volume of the original sample was restored with cement pore water containing the corresponding proxy ligand and ligand concentration. The sorption of Ni(II) and Eu(III) on the walls of the vessels was evaluated after terminating the sorption experiments. Hence, the suspension remaining in the vessel was discarded, the vessel was washed three times with MQ-water, filled in with 2% HNO₃ and left overnight. The concentration of Ni and Eu in the resulting solution was quantified by LSC and gamma counting, respectively. From these results, the fraction of ⁶³Ni and ¹⁵²Eu sorbed on the vessel’s walls was determined as 30% and 4%, respectively. Distribution ratios determined in this work are corrected accordingly. A negligible sorption of Pu on the vessel’s walls under analogous experimental conditions was previously reported by Tasi et al. (2021a). The main experimental conditions considered in this work for the binary HCP-RN and ternary HCP-RN-proxy ligands systems are summarized in Table 1.

Sorption was evaluated in terms of distribution ratios, R_d (in m³∙kg⁻¹), calculated as the ratio of radionuclide concentration in the solid ([RN]_solid, in mol∙kg⁻¹) and aqueous ([RN]_aq, in M) phases: R d = M s o l i d M a q ∙ 1 1000 = M 0 − M a q M a q ∙ V m (1) where [M]₀ is the initial radionuclide concentration in the aqueous phase (M), V is the volume of sample (m³) and m is the mass of HCP (kg).

Figure 2 shows the uptake of ⁶³Ni(II) by HCP in the absence of proxy ligands, as determined in this work for the degradation stage II of cement (pH ≈ 12.5). The figure shows also sorption data reported by Wieland and co-workers for the sorption of ⁶³Ni(II) by HCP in the degradation stage I (pH = 13.3), both in the absence and presence of added stable Ni(II) (Wieland et al., 2000; Wieland and Van Loon, 2003; Wieland et al., 2006).

R_d values determined in this work remain mostly constant within the investigated range of S:L ratios (1–20 g⋅dm^–3). Slight differences in the quantified R_d values are identified between sorption experiments with [⁶³Ni]₀ = 10⁻⁸ and 10⁻⁹ M, i.e. (3.6 ± 1.8) m³⋅kg^–1 and (1.8 ± 0.7) m³⋅kg^–1, although both values agree within the respective uncertainties. In contrast to this, Wieland and co-workers observed a systematic decrease of the measured R_d values with increasing S:L ratio (see Figure 2). The authors proposed isotopic exchange with stable nickel present in pristine HCP as main mechanism for the uptake of ⁶³Ni, and used the partition coefficient (α) to describe the isotopic distribution of ⁶³Ni between the HCP and the cement pore water: α = n s o l i d , N i 63 / n s o l i d , N i n a q , N i 63 / n a q , N i = R d , N i 63 R d , N i (1a) where n s o l i d , N i 63 and n s o l i d , N i are the number of moles of ⁶³Ni and stable Ni in HCP, n a q , N i 63 and n a q , N i denote the number of moles of ⁶³Ni and stable Ni in the cement pore water, and R d , N i 63 and R d , N i are the distribution coefficients of ⁶³Ni and stable Ni, respectively. The values of α approach 1 if the total inventory of ⁶³Ni in HCP is accessible to isotopic exchange.

Wieland et al. reported α = (3.3 ± 1.7)⋅10⁻² (absence of added stable Ni), α = (2.9 ± 0.6)⋅10⁻² (with stable [Ni]_added = 10⁻⁶ M) and α = (4.5 ± 1.0)⋅10⁻² (with stable [Ni]_added = 10⁻⁴ M) for the uptake of ⁶³Ni by HCP (in the degradation stage I) (Wieland et al., 2006). On the basis of these results, Wieland and co-workers argued that only a small fraction of stable Ni present in pristine HCP is available for isotopic exchange with ⁶³Ni.

Measurements of stable Ni in the HCP and corresponding pore water used in this work resulted in (3.7 ± 1.0)⋅10⁻⁴ mol⋅kg^–1 and (1.6 ± 0.8)⋅10⁻⁷ M, ¹ respectively, which allow the calculation of R d , N i = (2.3 ± 0.9) m³⋅kg^–1. The combination of this value with R d , N i 63 determined in this work results in α = (1.0 ± 0.5) and (1.9 ± 1.0) for experiments conducted at [⁶³Ni]₀ = 10⁻⁹ and 10⁻⁸ M, respectively. The partition coefficients (α) determined in this work are clearly higher than those reported in Wieland et al. (2006). Such discrepancies can be possibly explained by the different boundary conditions considered in both studies, i.e. cement degradation stage II (this work; pH ≈ 12.5) and I (Wieland et al., 2006; pH = 13.3). Hence, measured zeta potentials of the HCP surface sharply decrease from ≈20 mV at pH ≈ 12.5 to ≈ −20 mV at pH = 13.3 (Pointeau et al., 2006; Pointeau et al., 2008; Tasi et al., 2021a; Szabo et al., 2022a), whereas Ni(II) speciation is expected to change from Ni(OH)₂(aq) to Ni(OH)₃ ^– for the same variation of pH (González-Siso et al., 2018; Missana et al., 2022). In a recent study, Missana and co-workers reported R d , N i 63 ≈ 1 m³⋅kg^–1 (i.e. log R d , N i 63 ≈ 0) for the uptake of ⁶³Ni by C-S-H with Ca:Si = 1.4. Higher R d , N i 63 were obtained with decreasing pH and Ca:Si ratios, i.e. log R d , N i 63 ≈ 1 at Ca:Si = 0.8.

The concentration of stable Ni in the cement porewater is by a factor of 10 or 100, respectively, higher than the added ⁶³Ni. At the relatively high stable Ni concentration which is most likely determined by solubility of a Ni-bearing solid phase (LDH phases or Ni(OH)₂ (am)) we can expect that Ni species are not only abundant in a solid phase and as dissolved Ni(OH)₂(aq), but also as surface sorbed onto or incorporated into cement phases such as e.g., C-S-H. Spiked ⁶³Ni will undergo the same reactions as stable Ni so that isotope exchange is assumed to be the dominant retention mechanism under our experimental conditions. Note further that the partition coefficients measured in this work are consistent (within the uncertainties) with α = 1, which suggests the full availability of stable Ni in pristine HCP (in the degradation stage II) for isotopic exchange with ⁶³Ni.

Figure 3 shows the effect of proxy ligands (with 10⁻⁴ M ≤ [L]_tot ≤ 0.1 M) on the uptake of ⁶³Ni by HCP in degradation stage II. The figure includes also data reported by Bruno and co-workers for the uptake of ⁶³Ni under the same conditions but in the presence of ISA (Bruno et al., 2018). HIBA, HBA and GTA have a minor impact on the retention of ⁶³Ni(II) by HCP, and a slight decrease in the R d , N i 63 is observed only at [L]_tot = 0.1 M. These results are consistent with solubility data previously reported for β-Ni(OH)₂(cr) in the presence of the same proxy ligands (Szabo et al., 2022b), where an increase in the solubility was only observed at [L]_tot = 0.1 M. In contrast to the proxy ligands investigated in this work, Figure 3 shows that ISA decreases the retention of ⁶³Ni by HCP in degradation stage II at [ISA]_tot > 10⁻³ M (Bruno et al., 2018). This is again in line with solubility data reported for β-Ni(OH)₂(cr) in the presence of ISA, where the complexes Ni(OH)ISA(aq), Ni(OH)₂(ISA)^– and Ni(OH)₃(ISA)^2– were reported to control the aqueous speciation of Ni(II) in alkaline conditions (González-Siso et al., 2018). These observations support that the formation of stable complexes Ni(II)-OH-L is mostly responsible for the decrease in the retention of ⁶³Ni(II) by HCP. Differences observed between the proxy ligands and ISA are attributed to the presence of several alcohol groups in the latter ligand, which are known to form stable chelates with hard Lewis acids like Ni(II).

Figure 4 shows the uptake of ¹⁵²Eu(III) by HCP in degradation stage II, both in the absence and presence of increasing concentrations of proxy ligands. For comparative purposes, the figure shows also the sorption of Eu(III) by HCP in degradation stage I in the absence and presence of ISA and GLU, as reported by Wieland and co-workers (Wieland et al., 1998; Wieland and Van Loon, 2003; Wieland, 2014).

A very strong sorption is observed in the absence of proxy ligands, with R_d values ranging from ≈3⋅10³ m³⋅kg⁻¹ and ≈1⋅10⁴ m³⋅kg⁻¹ at S:L = 1–4 g⋅dm⁻³. These values are slightly higher but in line with sorption studies available in the literature. Wieland and co-workers reported R_d values ≥ 10³ m³⋅kg⁻¹ at S:L = 0.01 g⋅dm⁻³ and > 10² m³⋅kg⁻¹ at S:L = 0.1 g⋅dm⁻³ (Wieland et al., 1998; Wieland and Van Loon, 2003). Our results are also consistent with the uptake observed for Eu(III) in the presence of C-S-H phases (Tits et al., 2000; Pointeau et al., 2001), thus strongly supporting that the latter phases are the main sink of Eu(III) (as well as An (III)) in cementitious systems. The uptake mechanism investigated by time-resolved laser fluorescence spectroscopy and X-ray absorption spectroscopy for Eu(III), Nd(III) and Cm(III) supports a fast adsorption with the formation of surface complexes, followed by the slower incorporation of the trivalent lanthanide or actinide in the C-S-H structure (Pointeau et al., 2001; Tits et al., 2003; Stumpf et al., 2004; Mandaliev et al., 2010a; Mandaliev et al., 2010b). Note that stable Eu(III) is also present in HCP (quantified as 2.7⋅10⁻⁶ mol⋅kg⁻¹ in the CEM I material used in this work), although the inventory is significantly lower than in the case of stable Ni(II) (3.7⋅10⁻⁴ mol⋅kg⁻¹) and has a negligible effect in the retention of ¹⁵²Eu(III) in the current laboratory study. ²

The uptake of Eu(III) by HCP is not affected by the presence of the proxy ligands up to ligand concentrations of [L]_tot = 0.1 M, which is consistent with solubility data previously reported for Nd(OH)₃(s) in the presence of the same proxy ligands (Szabo et al., 2022b). Although these ligands are known to form stable complexes with Ln(III)/An(III) in acidic to near-neutral pH conditions (Deelstra and Verbeek, 1964; Wang et al., 2000; Giroux et al., 2002), they cannot compete against hydrolysis in hyperalkaline systems or against the strong sorption occurring in HCP systems. This observation is further supported by thermodynamic calculations conducted by Keith-Roach and Shahkarami (Keith-Roach and Shahkarami, 2021). In contrast to the proxy ligands investigated in this work, the uptake of Eu(III) is strongly decreased in the presence of GLU and, to a lesser extent, ISA (Wieland, 2014) (see Figure 4). Even though HIBA and HBA contain a carboxylic group with an alcohol group in α- or β-position, this configuration is insufficient to form stable complexes with Eu(III) in hyperalkaline systems. As discussed in Szabo et al. (2022a), a greater number of alcohol groups (as occurs in ISA or GLU) is required to form stable chelates in hyperalkaline conditions and thus affect aqueous complexation and sorption phenomena.

The measurements of pH and E _h in the Pu systems containing HQ and Sn(II) resulted in (pe + pH) = (10.0 ± 0.5) and (2 ± 1), respectively. These values are consistent with previous studies available in the literature using the same reducing chemicals (Tasi et al., 2018a; Tasi et al., 2018c). Considering the redox distribution of plutonium in alkaline reducing conditions, Pu(IV) is expected to prevail in the aqueous phase of HQ-buffered systems, whereas both Pu(III) and Pu(IV) may coexist in the strongly reducing conditions imposed by Sn(II) (Fellhauer, 2013; Tasi et al., 2018c; Grenthe et al., 2020). Figures 5A, B (HQ-system) and (Sn(II)-system) show the uptake of Pu (in terms of log R_d values) by HCP in the degradation stage II, both in the absence and presence of increasing concentrations of proxy ligands. The figure includes also sorption data reported for Pu in the presence of ISA under analogous experimental conditions (Tasi et al., 2021a).

Sorption data in the investigated Pu samples are characterized by a large dispersion. This is a common feature of An(IV) systems in pH-conditions where the neutral species An(OH)₄(aq) prevails, and it is caused by difficulties in the phase separation and the possible presence of intrinsic colloids. A very strong sorption of Pu is observed in HQ-systems and absence of proxy ligands, with log R_d values ranging from 2.2 to 3.5 (with R_d in m³∙kg⁻¹). These results are consistent with sorption data reported by Tasi and co-workers under analogous experimental conditions (Tasi et al., 2021a). The reference book by Ochs et al. recommends a log R_d of 1.5 for the uptake of Pu(IV) by HCP in the degradation stage II, with log R_d = 3 defined as upper limit (Ochs et al., 2016), also in line with our experimental observations.

Strong sorption is observed also in the presence of proxy ligands, with analogous log R_d values determined in HQ- and Sn(II)-systems. Considering the strong sorption described also in Section 3.2 for Eu(III), the experimental observations for Pu in Sn(II)-systems do not allow the unequivocal assignment of the oxidation state prevailing under these very reducing conditions. A slight decrease in the log R_d values can be speculated at [L]_tot > 10⁻² M, especially in the case of HQ-systems. However, this observation is uncertain due to the large dispersion of the data, and by the increase of the detection limit with increasing ligand concentration caused by the steady increase of the dilution factors required for ICP-MS. In spite of this, R_d values remain high in all the investigated systems containing proxy ligands, i.e. log R_d > 1. The weak impact of the proxy ligands on the retention of Pu by HCP in degradation stage II is in line with the results of solubility experiments in the presence of analogous concentrations of proxy ligands, in which the solubility of PuO₂(ncr, hyd) remains unaffected up to [L]_tot = 0.1 M, both in HQ- and Sn(II)-buffered systems (Szabo et al., 2022b). Although HIBA, HBA and GTA are known to form stable binary complexes with An (III/IV) under acidic conditions, these complexes can neither outcompete hydrolysis nor the strong sorption on HCP under the hyperalkaline pH conditions characteristic of the degradation stage II of cement.

In contrast to the proxy ligands, Figure 5 shows the remarkable impact that ISA on the uptake of Pu, which results in a decrease of up to 3 orders of magnitude in the R_d values (Tasi et al., 2021a). This is attributed to the strong ternary and quaternary complexes (Ca-)Pu-OH-ISA forming in hyperalkaline conditions (Tasi et al., 2018a; Tasi et al., 2018b). Moreover, distinct trends are observed at high ISA concentrations, involving the formation of stable surface complexes than increase sorption. Tasi and co-workers explained this observation with the strong uptake of ISA by cement, which modifies the surface properties at [ISA]_tot ≥ 10⁻³ M. This again highlights the significantly different behaviour of proxy ligands and ISA, not only with respect to the interaction with radionuclides but also concerning the interaction with cement surfaces. Similar observations have been reported for the uptake of Pu(IV) and Th(IV) by HCP and C-S-H phases in the presence gluconate, a polyhydroxocarboxylic acid analogous to ISA (Moreton, 1993; Moreton et al., 2000; Wieland, 2014; Guidone et al., 2022).