Compilation of Technical Notes on Less Studied Elements.Pdf

SCK•CEN/12201513 SCK•CEN ER-0323 2016/Ssa/Pub-10 Compilation of Technical Notes on less studied elements

1st Full version

Sonia Salah1, Norbert Maes1, Christophe Bruggeman1, Lian Wang1, Richard Metcalfe2, James Wilson2

1SCK•CEN, 2QUINTESSA

Publication date: January 2017

Contract naam: Contrat de R&D “Gestion à long terme des déchets radioactifs” (2015-2020) Contract nummer: CO-90-14-3690-00 / CCHO 2015-0304/00/00

Contract information: RS 15-SCK-MIG-29: Finalisation of technical notes on less known elements

SCK•CEN Boeretang 200 BE-2400 Mol Belgium

Phone +32 14 33 21 11 Fax +32 14 31 50 21

http://www.sckcen.be

Contact: Knowledge Centre [email protected]

COPYRIGHT RULES All property rights and copyright are reserved to SCK•CEN. In case of a contractual arrangement with SCK•CEN, the use of this information by a Third Party, or for any purpose other than for which it is intended on the basis of the contract, is not authorized. With respect to any unauthorized use, SCK•CEN makes no representation or warranty, expressed or implied, and assumes no liability as to the completeness, accuracy or usefulness of the information contained in this document, or that its use may not infringe privately owned rights.

SCK•CEN, Studiecentrum voor Kernenergie/Centre d'Etude de l'Energie Nucléaire Stichting van Openbaar Nut – Fondation d'Utilité Publique ‐ Foundation of Public Utility Registered Office: Avenue Herrmann Debroux 40 – BE‐1160 BRUSSEL Operational Office: Boeretang 200 – BE‐2400 MOL

SCK•CEN/12201513 Page 2 of 208 Compilation of Technical Notes on less studied elements

Table of Content Configuration Control ...... 3 Table of Content ...... 4 Abstract ...... 6 Keywords...... 7 List of Figures ...... 8 List of Tables ...... 13 1 Introduction ...... 15 1.1 Group I: Reference element for non-retarded transport: Tritiated water (HTO) ...... 18 1.2 Group II: Elements characterized by anion exclusion ...... 18 1.3 Group III: Elements characterized by cation exchange sorption ...... 19 1.4 Group IV: Elements characterized by transport behavior linked to DOM ...... 19 1.5 Subgroup IVa: Transition metals (+ Be and Sn) ...... 20 1.6 Subgroup IVb: Trivalent lanthanides and actinides (La/Ac) ...... 21 1.7 Subgroup IVc: Tetravalent lanthanides and actinides (La/Ac) ...... 22 1.8 References ...... 22 2 GROUP II: Elements characterized by anion exclusion ...... 24 2.1 Subgroup IIa: Monovalent anions ...... 24 2.1.1 Technical note for Carbon (C) ...... 24 2.1.2 Technical note for Chlorine (Cl) ...... 35 2.1.3 Technical note for Selenium (Se) ...... 40 2.2 Subgroup IIa: Divalent anions ...... 53 2.2.1 Technical note for Molybdenum (Mo) ...... 53 3 GROUP III: Elements characterized by cation exchange sorption ...... 60 3.1 Subgroup IIIa: Monovalent cations (alkali metals) ...... 60 3.1.1 Technical note for Rubidium (Rb) ...... 60 3.2 Subgroup IIIb: Divalent cations (alkali earth metals) ...... 63 3.2.1 Technical note for Calcium (Ca) ...... 63 3.2.2 Technical note for Radium (Ra)...... 68 4 GROUP IV: Elements characterized by DOM linked transport ...... 74 4.1 Subgroup IVa: Transition metals (+ Be and Sn) ...... 74 4.1.1 Technical note for Silver (Ag)...... 74 4.1.2 Technical note for Beryllium (Be) ...... 79 4.1.3 Technical note for Niobium (Nb) ...... 90 4.1.4 Technical note for Nickel (Ni) ...... 100 4.1.5 Technical note for Palladium (Pd) ...... 109 4.1.6 Technical note for Tin (Sn) ...... 112 4.1.7 Technical note for Zirconium (Zr) ...... 117 4.2 Subgroup IVb: Trivalent lanthanides and actinides ...... 122 4.2.1 Technical note for Actinium (Ac) ...... 122 4.2.2 Technical note for Curium (Cm) ...... 130 4.2.3 Technical note for Plutonium (Pu) ...... 144 4.2.4 Technical note for Samarium (Sm) ...... 162

SCK•CEN/12201513 Page 4 of 208 Compilation of Technical Notes on less studied elements

4.3 Group IVc: Tetravalent lanthanides and actinides (+ pentavalent Pa) ...... 166 4.3.1 Technical note for Neptunium (Np) ...... 166 4.3.2 Technical note for Thorium (Th) ...... 176 4.3.3 Technical note for Protactinium (Pa) ...... 195 5 ANNEX I ...... 206

SCK•CEN/12201513 Page 5 of 208 Compilation of Technical Notes on less studied elements

Abstract

The current report represents one of the documents intended to substantiate the assessment basis to be used within the scope of SFC 1 and represents a compilation of literature, speciation, solubility, sorption, and migration data for 21 less studied waste and safety relevant radionuclides discriminated by ONDRAF/NIRAS. The preparation of the so-called Technical Notes for 21 elements presented herein has been a common effort of SCK•CEN and QUINTESSA, with the former summarizing the knowledge on elements for which in-house data (experimental and modeling data) were available, while QUINTESSA mainly focused on gathering literature information on the elements for which no or only limited data existed. As compared to the “reference elements”, i.e. HTO, Cs, Sr, Am, Tc, and U for which so-called Topical Reports (TR’s) are available, the above mentioned radionuclides are considered to represent “less studied elements”, due to which O/N requested a specific report synthesizing the available and safety relevant information for these radionuclides.

In this context, it should be referred to the research strategy developed by the Waste & Disposal Research Unit at SCK•CEN together with ONDRAF/NIRAS during the last years in preparation of the Safety and Feasibility Case 1. As studying all waste relevant radionuclides in detail was impossible, due to safety reasons budget and time limitations, the strategic approach consisted of subdividing all radionuclides into different groups based on analogous chemical characteristics (e.g. oxidation state, place in the periodic table) and behaviour, such as their hydrolysis/complexation, their retention and migration (Bruggeman et al., 2008). Each group is represented by one or two “reference elements", which have been studied in detail and for which so-called "phenomenological models" were elaborated. These phenomenological or geochemical models represent the cornerstone of the developed methodology, describing in a qualitative and quantitative way the retention and migration processes under undisturbed Boom Clay conditions. The described methodology is based on a huge experimental programme, which is supported by geochemical calculations as well as transport modeling. In total, four main groups and several subgroups of RN's have been differentiated, characterized by an increasing complexity in interaction mechanisms, which coincides with their charge and valence/oxidation state (ref).

Besides compiling all relevant and available information on the “less studied elements”, the current report aims at justifying the developed research strategy of radionuclide grouping/ranking. At the end of each Technical Note, the reader will thus find a justification summarizing the reasoning for associating the element under consideration to a specific group. It should be mentioned here, that due to the lack of literature and/or experimental data, this task has not been straightforward for certain elements and revisions will be undertaken, in case more data become available and existing knowledge gaps can be filled.

SCK•CEN/12201513 Page 6 of 208 Compilation of Technical Notes on less studied elements

References

Bruggeman, S. Salah, N. Maes, L. Wang, A. Dierckx , and M. Ochs (2008) Outline of the experimental approach adopted by SCK•CEN for developing radionuclide sorption parameters, External Report SCK•CEN-ER-73.

Bruggeman, N., Maes, N., (2016) Radionuclide migration and retention in Boom Clay, External Report SCK•CEN-ER-0345.

ONDRAF/NIRAS (2011) Plan Déchets pour la gestion à long terme des déchets radioactifs conditionnées de haute activité et/ou de longue durée de vie et aperçue de questions connexes, Rapport NIROND 2011-02 F.

ONDRAF/NIRAS (2013) ONDRAF/NIRAS Research, Development and Demonstration (RD&D) Plan for the geological disposal of high-level and/or long-lived radioactive waste including irradiated fuel if considered as waste. State-of-the-art report as of December 2012. Belgian Agency for radioactive Waste and Enriched Fissile Materials, NIROND-TR 2013-12E.

Keywords radionuclides, sorption, migration, speciation, solubility, cation exchange, surface complexation, retardation, transport, Boom Clay, geological disposal

SCK•CEN/12201513 Page 7 of 208 Compilation of Technical Notes on less studied elements

List of Figures Figure 1: Elements belonging to group II: highlighted in blue, group III: highlighted in pink, group IVa highlighted in green, group IVb and IVc: highlighted in violet; (group I = HTO) not illustrated ...... 17 Figure 2: a) Generic Eh-pH diagram for water (taken from Brookins, 1988) and b) Eh-pH diagram of carbon (C-S- - -2 O-H) for the BC reference porewater system. Assumed activity of [HCO3 ] = 1.4 × 10 . Database: MOLDATA_R2. Code: The Geochemist's Workbench- 10.0...... 25 i Figure 3: Apparent dispersion coefficient, D app, versus the apparent velocity, Vapp. Figure a (top) concerns all clay cores from Mol-1 (see Table 1). The cores with bad fits are highlighted by different markers. Figure b (bottom) contains six clay cores from Mol-1 as well as the cores from the pulse injection experiments i from Table 1. The line is the best fit according to the expression (D app = Dapp + αVapp). In Figure a, the linear extrapolation concerns six clay cores in the "homogeneous range" of the Mol-1 series with good fitting results, while in Figure b, all shown points are included in the linear fit. (Aertsens et al., 2010) ..... 27 Figure 4: The product ηR of the (diffusion) accessible porosity and the retardation factor as a function of depth. The clay cores for which the quality of the fit is not excellent (191 m, 211 m, 239 m and 291 m, see Table 1) are indicated by a square. The two horizontal lines are the maximal and minimal ηR obtained considering the results of the pulse injection experiments mentioned in Table 2 (Aertsens et al., 2010) .. 29 Figure 5: Set-up of the Tribicarb-3D experiment: three piezometers, each with a number of filters, are placed in the Boom Clay next to the URF. Both piezometers R34-1 and R32-3 are approximately parallel to one another and to the bedding plane of the clay. Piezometer R32-2 is inclined. Tracer (HTO and H14CO3-) is initially injected in filter 6 of piezometer R32-3 (Aertsens et al., 2013)...... 31 14 - Figure 6: Experimental data and blind prediction of the H CO3 evolution in the filters of the a) injection piezometer R32-3, b) the piezometer R34-1 parallel to the injection piezometer, and c) the inclined ...... 32 Figure 7: Temperature dependence of Cl speciation in the Mol reference water reported in De Craen et al. (2004). The speciation calculations were undertaken with PHREEQC version 3.1.1 and the thermodynamic database “llnl.dat”, which is distributed with the PHREEQC package...... 36 Figure 8: Aqueous Cl concentrations limited by the solubility of three alternative Cl-bearing minerals, compared with the concentration of Cl in the Mol Reference Water reported in De Craen et al. (2004). Solubility calculations were undertaken with PHREEQC version 3.1.1 and the thermodynamic database “data0.ypf.R2” (USDOE, 2007). This database supports the Pitzer approach for calculating activity coefficients in highly saline solutions ...... 36 Figure 9: Eh-pH diagram of selenium (Se-C-S-O-H) for the BC reference porewater system. Assumed activities of dissolved [Se] = 10-8. Database: MOLDATA_R2. Code: The Geochemist's Workbench - 10.0...... 40 Figure 10: X-ray absorption near-edge structure spectra for different Boom Clay fractions (A > 1 µm, 250 nm < B < 1 µm, 40 nm < C 250 nm) equilibrated with an initial Se concentration of 5×10-5 M ("5"), 5×10-4 M ("4") and 5×10-3 M ("3") (Breynaert et al., 2010) ...... 41 75 2- 75 2- Figure 11: Evolution of SeO3 (a) and SeO4 (b) concentrations in FeS2 suspensions (2.5 g/L and 10 g/L) measured by a combination of ion chromatography and gamma counting (Bruggeman et al., 2002) ...... 43 2- Figure 12: Time evolution of SeO3 in supernatant solutions of batch experiments with Boom Clay (0.05 kg/L and 0.21 kg/L) at two initial Se concentrations (1×10-6 and 5×10-6 M) (Bruggeman et al., 2005) ...... 44 Figure 13: The electromigration experimental setup (Beauwens et al., 2005) ...... 45 Figure 14: Experimental results from electromigration experiments with selenate source. Incomplete oxidation of the source (originally in selenite form) causes the simultaneous occurrence of an immobile and a mobile species (Beauwens et al., 2005) ...... 45 Figure 15: Se concentration in outlet (75Se, as Bq/L, recalculated towards start of the percolation) as function of time ...... 46 Figure 16: Se concentration in outlet (75Se, as Bq/L, recalculated towards start of the percolation) as function of time ...... 47 Figure 17: Se concentration in outlet (mol/L) as function of percolated volume (mL) ...... 47 Figure 18: Se concentration profile (in Bq/g) in the two clay cores, obtained after ~ 800-1000 days percolation. The profiles are given relative to the position of the administered 75Se source (in mm). The percolation direction is from left to right...... 49 Figure 19: Se concentration profile (in Bq/g, logarithmic scale) in the two clay cores, obtained after ~ 800-1000 days percolation. The profiles are given relative to the position of the administered 75Se source (in mm). The percolation direction is from left to right...... 49 Figure 20: Eh-pH diagrams of molybdenum (in the system Mo-Ca-C-S-Cl-F-O-H for a) to c) and in the same system plus Se for d)) for the BC reference porewater from De Craen et al. (2004), calculated using

SCK•CEN/12201513 Page 8 of 208 Compilation of Technical Notes on less studied elements

Geochemist’s Workbench version 7.0 and assuming an activity of dissolved Mo, [Mo] = 10-5 . a) Calculated using the Visual MINTEQ thermodynamic database release 2.40 with no minerals or species suppressed. b) Calculated using the Visual MINTEQ release 2.40 with all minerals suppressed. c) As for a), but calculated using the LLNL V8 R6 "combined" dataset, thermo.com.V8.R6. d) As for b), but calculated using the LLNL V8 R6 "combined" dataset, thermo.com.V8.R6.full and with activity of Se, [Se] = 10-8...... 55 Figure 21: Solubility diagrams for molybdenum (in the system Mo-Ca-C-S-Cl-F-O-H for a) and in the same system plus Se for b)) for the BC reference porewater from De Craen et al. (2004), calculated using Geochemist’s Workbench version 7.0. a) Calculated using the Visual MINTEQ thermodynamic database release 2.40 with no minerals or species suppressed. b) As for a), but calculated using the LLNL V8 R6 "combined" dataset, thermo.com.V8.R6.full and with activity of Se, [Se] = 10-8. Note that the only Mo- solids in the database used in b) are Mo and MoSe2; if the latter is suppressed than no minerals are calculated to be stable...... 56 Figure 22: Eh-pH diagram of rubidium (Rb-C-S-O-H) for the BC reference porewater system. Assumed activity of dissolved [Rb] = 10-8. Diagram a) MOLDATA/NEA TDB. Code: The Geochemist's Workbench - 8.08...... 60 Figure 23: Eh-pH diagram of calcium (Ca-C-S-O-H) for the BC reference porewater system. Activity of dissolved [Ca] = 2.44 × 10-5. Left diagram: aqueous speciation, right diagram: solid phases included in calculation. Database: MOLDATA_R2. Code: The Geochemist's Workbench- 10.0 ...... 63 Figure 24: Left: Schematic representation of the electromigration set-up. Right: Typical electromigration dispersion profiles in the clay obtained for radionuclides with different charges...... 64

Figure 25: Results from the Ca electromigration experiments on Boom Clay conductedi at different electrical DDapp= app +α D V app fields resulting in different Vapp. The line represents the linear relationship ...... 65 Figure 26: Eh-pH diagram of radium (Ra-C-S-O-H) for the BC reference porewater system...... 68 Figure 27: Left: Schematic representation of the electromigration set-up. Right: Typical electromigration dispersion profiles in the clay obtained for radionuclides with different charges...... 71 Figure 28: The linear relationship between the apparent convection velocity and the apparent dispersion coefficient gives the apparent diffusion coefficient (intercept). The squares and circles are ...... 72 Figure 29: Solubility diagrams for silver, assuming Ca2+ activity buffered by calcite, sulphate buffered by pyrite, 2+ - Fe buffered by siderite, log a Cl = -3.155, log f CO2(g) = -2.44, log f O2(g) = -71.4 (Eh = -0.28 mV at pH 8.36, 16 °C, 1 bar). Diagrams generated using ‘Act2’ module of Geochemist’s Workbench® (Bethke, 2008), speciated for carbonate. Upper diagram includes all solids. Native silver is suppressed in the lower diagram...... 75 Figure 30: Eh-pH diagram of beryllium (Be-C-S-O-H) for the BC reference porewater system...... 80 Figure 31: Distribution of Be-hydrolysis products (x,y) at (a) I = 1 m Be(II) and (b) 10-5 m Be(II) (copied from Baes and Mesmer, 1976) ...... 81 Figure 32: Dependence of measured Kd values on the pH value of the solution. Three different systems...... 83 7 Figure 33: Variations in partition coefficients (in log Kd) of Be with different particle types in the presence or absence of model macromolecular organic compounds, i.e. humic acid (HA), acid polysaccharide (APS), and protein (BSA) (Yang et al., 2013) ...... 84 Figure 34: Calculated hydraulic conductivity, K (in m/s), as function of time in 2 percolation experiments ...... 85 Figure 35: Be concentration in outlet (7Be, as Bq/L, recalculated towards start of the percolation) as ...... 86 Figure 36: Be concentration in outlet (mol/L) as function of percolated volume ...... 86 Figure 37: Be concentration profile (in Bq/g) in the two clay cores, obtained after ~ 280 days percolation. The profiles are given relative to the position of the administered 7Be source (in mm)...... 87 Figure 38: Aqueous speciation of niobium (10-8 M) in saline Olkiluoto reference water as pH is varied between 6 and 11, after Ervanne et al. (2013). The curves were calculated by PHREEQC and ThermoChimie database v 7b (note that Nb data in this version of the thermodynamic database are the same as those in v 9 used to calculate the results in Table 15 below). Similar speciation results were obtained for all the Olkiluoto reference waters...... 91 Figure 39: Solubility of Nb(V) under anaerobic conditions, as measured by Yajima (1992) and Yajima et al (1994), as plotted in Kitamura et al. (2010a). Plots with mark “�” were taken from the first run (analyzed by ICP-OES) and others were taken from a second run (analyzed by ICP-MS). The solid line is a least- - square regression assuming that dissolved Nb is in the form Nb(OH)5(aq) and Nb(OH)6 , with the solubility limiting phase being Nb2O5(s); dashed lines show the contribution from each aqueous species...... 93 Figure 40: Eh-pH diagram of niobium (Nb-C-S-O-H) for the BC reference porewater system. Diagram a) activity of dissolved [Nb] = 10-8, b) minerals included in calculation, activity of dissolved [Nb] = 10-5. Code: The Geochemist's Workbench -8.08...... 94

SCK•CEN/12201513 Page 9 of 208 Compilation of Technical Notes on less studied elements

Figure 41: Rd (m3/kg) values calculated for Nb(V) sorption based on literature Rd data and presented as an empirical cumulative distribution function for Olkiluoto tonalite (TVO-GW) and Rapakivi granite samples (IVO-GW) (after Crawford, 2010)...... 96 Figure 42: Solubility diagrams for nickel, assuming Ca2+ activity buffered by calcite, sulphate buffered by pyrite, 2+ - Fe buffered by siderite, log f CO2(g) = -2.44, log a Cl = -3.155, log f O2(g) = -71.4 (Eh = -0.28 mV at pH 8.36, 16 °C, 1 bar). Solute activity buffers are the same as those assumed for models of the Boom Clay porewater discussed by Salah and Wang (2014). Diagrams generated using ‘Act2’ module of Geochemist’s Workbench® (Bethke, 2008) and speciated for carbonate. Diagram (a) includes all solids; (b) excludes NiS(gamma); (c) excludes NiS(gamma) and NiS(beta)...... 102 Figure 43: Sorption isotherms and log Kd-values as function of the Ni-equilibrium concentrations for Ni on Silver Hill Illite at various conditions. In the left graphs the isotherms measured at pCO2atm = 0.04% and in CO2- free atm with different amounts of organic carbon (0-30 mg/L) are shown. The right graphs show the sorption results measured at in-situ pCO2 of 0.4 % and different amounts of organic carbon (0-40 mg/L)...... 103 Figure 44: Sorption edge for Ni on Illite du Puy (S/L:1 g/L) at IS 0.1 M NaClO4. Error ...... 104 Figure 45: Log Kd-values of Ni on Boom Clay determined in SBCW and RBCW...... 104 Figure 46: Eh-pH diagram of palladium (Pd-C-S-O-H) for the BC reference porewater system. Assumed activity of dissolved [Pd] = 10-8. Database: MOLDATA_R2. Code: The Geochemist's Workbench - 10.0...... 109 Figure 47: Eh-pH diagram of tin (Sn-C-S-O-H) for the BC reference porewater system. Assumed activity of dissolved [Sn] = 10-8. Diagram a) LLNL TDB, b) ANDRA (ThermoChimie) and MOLDATA TDB, c) NAGRA/PSI TDB, d) NEA TDB (not meaningful, as Sn2+ is the only aqueous species), e) MOLDATA TDB solid phases included, f) same diagram as e, but [Sn] = 10-6, g) same diagram as f, but cassiterite suppressed in calculation. Code: The Geochemist's Workbench (from Salah and Wang, 2014)...... 113 Figure 48: Eh-pH diagrams of zirconium (Zr-C-S-O-H) for the BC reference porewater system. Assumed activity of dissolved Zr = 10-8. Diagram a) LLNL TDB, b) ANDRA TDB c) NAGRA/PSI TDB, d) NEA TDB, e) MOLDATA Code: The Geochemist's Workbench (reproduced from Salah and Wang, 2014)...... 118 Figure 49: Solubility diagrams for zirconium, assuming Ca2+ activity buffered by calcite, sulphate buffered by 2+ pyrite, Fe buffered by siderite, log f CO2(g) = -2.44, log f O2(g) = -71.4 (Eh = -0.28 mV at pH 8.36, 16 °C, 1 bar), log a Cl- = -3.155. Solute activity buffers are the same as those used for model Boom Clay porewater described by Salah and Wang (2014). Diagrams generated using ‘Act2’ module of Geochemist’s Workbench® (Bethke, 2008). Upper diagram includes all solids...... 119 Figure 50: Eh-pH diagrams of actinium (in the system Ac-C-S-Cl-F-O-H) to for the BC reference porewater from De Craen et al. (2004), calculated using Geochemist’s Workbench version 7.0 and the JAEA-TDB version 140331, and assuming an activity of dissolved Ac, [Ac] = 10-5 and a temperature of 16 ̊C. In a) all aqueous species and solid phases in the database are allowed to form; in b) solid phases are suppressed...... 124 Figure 51: Solubility diagram for actinium (in the system Ac-C-S-Cl-F-O-H) for the BC reference porewater from De Craen et al. (2004), calculated using Geochemist’s Workbench version 7.0 and the JAEA-TDB version 140331...... 125 Figure 52: Left: Calculated Cm(III) speciation in carbonate containing systems...... 131 Figure 53: Cm(III) fluorescence spectra in the EG/BS < 100kDa size fraction in presence (left) and absence (right) of 15 mM bicarbonate at two excitation wavelengths at pH 8.7...... 131 Figure 54: Compilation of the Cm(III) fluorescence spectra recorded by the “indirect excitation” method. The spectra are scaled to the same peak height...... 132 Figure 55: Species distribution of Cm(III) as a function of pH (Stumpf et al., 2001) ...... 133 Figure 56: Effect of metal concentration on the interaction constant of trivalent actinides with Aldrich humic acids (at a ionic strength of 0.1 M). TRLIF: Time-Resolved Laser-Induced Fluoresence; SP: spectrophotometry ...... 135 Figure 57: Separation of curium humate TRLFS spectrum after four weeks of contact time (Cm-humate) into the spectrum of the slow dissociation mode and a rest representing curium humate that has dissociated during the five hours contact with the cation exchanger Chelex 100. The latter spectrum is a mixture between fast dissociation mode and a shoulder for the slow mode that also shows partial dissociation within the five hours contact time with Chelex (Monsallier et al., 2003) ...... 135 Figure 58: Trivalent metal ion humic acid complexation in dependence of the metal ion loading for 1×10-8 and -7 1×10 M Cm(III), at a constant pH (6.0) and humic acid concentration of 10 mg/L in 0.01 M NaClO4. The 3+ spectrum of the free Cm aq in 0.1 M HClO4 in the absence of humic acid is added to (a) ...... 136

SCK•CEN/12201513 Page 10 of 208 Compilation of Technical Notes on less studied elements

Figure 59: Cm(III) humic acid complexation in dependence of pH for 1×10-7 M Cm(III) and 10 mg/L humic acid in 3+ 0.01 M NaClO4. The spectrum of the free Cm aq in 0.1 M HClO4 in the absence of humic acid is added to (a) ...... 137 Figure 60: Calculated hydraulic conductivity, K (in m/s), as function of time in 2 percolation experiments (Cm244m1c7 and Cm244m1c8) and sequential migration experiment (Cm244m1c7 SM) ...... 138 Figure 61: Cm concentration in outlet (244Cm, as Bq/L) as function of time ...... 139 Figure 62: Cm concentration in outlet (mol/L) as function of percolated volume ...... 139 Figure 63: Conceptual model used for the interpretation of organic matter linked radionuclide migration in Boom clay (Maes et al., 2011) ...... 140 Figure 64: Results of the fitting of the Cm elution curve in a sequential migration experiment using the proposed conceptual model (Maes et al., 2011) ...... 141 Figure 65: Eh-pH diagram of plutonium (Pu-C-S-O-H) for the BC reference porewater system. Assumed dissolved activity of [Pu] = 10-8. Database MOLDATA_R2. Code: The Geochemist's Workbench - 10.0 145 Figure 66: Evolution in Pu(IV) concentrations (after filtration at 30000 MWCO) in H2O, SBCW (with increasing concentrations of TROM humic substances) and in RBCW. Measurements after 15 and 84 days (Maes et al., 2004) ...... 146 Figure 67: Concentration of colloidal/polymeric Pu(IV) (squares with crosses inside) and Pu concentration in the supernatant after ultrafiltration (crosses) determined in a PuO2 solubility study under Ar atmosphere . 148 Figure 68: Solid-liquid and redox equilibria of plutonium in the presence of oxygen (Neck et al., 2007) ...... 148 Figure 69: Measured ζ-potential as a function of pH of intrinsic Pu colloids and natural smectite colloids ...... 149 Figure 70: Plutonium sorption (Kd) as a function of the reciprocal of the absorbance at 280 nm ...... 150 Figure 71: 30 day Pu(V) (circles) and 30 day Pu(IV) (squares) sorption isotherms for SWy-1 Na-montmorillonite (1 g/L) in 0.7 mM NaHCO3, 5 mM NaCl buffer solution at pH 8 (Begg et al., 2013) ...... 150 Figure 72: Relationship of Kd of reduced Pu to dissolved organic carbon concentration for 15 lakes and four 237 rivers. The results of four laboratory measurements of Kd using added Pu tracer are shown for comparison ...... 151 Figure 73: Calculated hydraulic conductivity, K (in m/s), as function of time in 3 percolation experiments (Pu238m3c2, Pu238m2c4, Pu238m2c6) and 1 sequential migration experiment (Pu238m2c6 SM) ...... 153 Figure 74: Pu concentration in outlet (238Pu, in Bq/L) as function of time...... 154 Figure 75: Pu concentration in outlet (mol/L) as function of percolated volume (mL) ...... 154 Figure 76: Conceptual model used for the interpretation of organic matter linked radionuclide migration in Boom clay (Maes et al., 2011) ...... 155 Figure 77: Results of the fitting of the Pu elution curve in a sequential migration experiment using the proposed conceptual model (Maes et al., 2011) ...... 156 Figure 78: Normalized column breakthrough curves of Pu, smectite colloids and conservative tritium tracer. The dotted line curve is obtained by dividing the measured effluent Pu activity at any time by the maximum measured effluent activity (Abdel-Fattah et al., 2013) ...... 158 Figure 79: Eh-pH diagram of samarium (Sm-C-S-O-H) for the BC reference porewater system...... 163 Figure 80: Eh-pH diagram of neptunium (Np-C-S-O-H) for the BC reference porewater system...... 167 Figure 81: Neptunium activity in filtrates and ultrafiltrates. Initial Np ...... 168 Figure 82: Calculated hydraulic conductivity, K (in m/s), as function of time in 2 percolation experiments (Np237m1c5 and Np237m1c6) and 1 sequential migration experiment (Np237m1c5 SM) ...... 170 Figure 83: Np concentration in outlet (237Np, in Bq/L) as function of time...... 170 Figure 84: Np concentration in outlet (mol/L) as function of percolated volume (mL) ...... 171 Figure 85: Conceptual model used for the interpretation of organic matter linked radionuclide migration ..... 172 Figure 86: Results of the fitting of the Np elution curve in a sequential migration xperiment using the proposed conceptual model (Maes et al., 2011) ...... 173 Figure 87: Thorium speciation in 0.01 M NaCl as function of pH in absence of carbonates. a) [Th] = 1×10-8 M, b) [Th] = 1×10-3 M. Database: Moldata_R2.Code: Geochemist's Workbench 10.0...... 177 Figure 88: Eh-pH diagram of thorium (Th-C-S-O-H) for the BC reference porewater system...... 177 Figure 89: Eh-pH diagram of thorium (Th-C-S-O-H) for the BC reference porewater system...... 177 Figure 90: Th-concentrations in SBCW after 0.45 µm and 2 nm (ultra)filtration as function of the initial TOC concentrations Dashed line gives detection limit (i.e. 4.3×10-9 M; 1 ppb). Copied from Delécaut (2004; p. 132)...... 179 Figure 91: ThO2(cr) solubilities as function of different TOC concentrations (Liu et al., unpublished data)...... 180 Figure 92: Solubility of ThO2 as function of pH measured by different authors. The solid line represents predictions made with the equilibria described above in equations 10-13 (Hummel et al., 2002)...... 183

SCK•CEN/12201513 Page 11 of 208 Compilation of Technical Notes on less studied elements

Figure 93: Th-solubility as function of pH calculated with values recommended by Rard et al. (2008) ...... 187 Figure 94: Sorption isotherms and solid/liquid distribution coefficients (Kd-values) for Th-experiments performed in SBCW and RBCW after centrifugation (NF) and ultrafiltration (UF) ...... 188 Figure 95: Eh-pH diagram of protactinium (Pa-C-S-O-H) for the BC reference porewater system. Assumed activity of [Pa] = 10-8. Database: MOLDATA (2010_MOLDATA_nov_b O2.dat). Code: The Geochemist's Workbench - 8.12...... 195 Figure 96: Calculated hydraulic conductivity, K (in m/s), as function of time in 2 percolation experiments (Pa231m3c3-top and Pa231m3c4a-middle) and sequential migration experiment (Pa231m3c4b-bottom) ...... 199 Figure 97: Pa concentration in outlet (231Pa, as Bq/L) as function of time in 2 percolation experiments (Pa231m3c3-top and Pa231m3c4a-middle) and sequential migration experiment (Pa231m3c4b-bottom) ...... 201 Figure 98: Conceptual model used for the interpretation of organic matter linked radionuclide migration in Boom clay (Maes et al., 2011) ...... 202 Figure 99: Results of the fitting of the Pa elution curve in a sequential migration experiment using the proposed conceptual model (Maes et al., 2011) ...... 203

SCK•CEN/12201513 Page 12 of 208 Compilation of Technical Notes on less studied elements

List of Tables 14 - Table 1: Results for the samples of the Mol-1 coring selection for the H CO3 injection campaign...... 27 14 - Table 2: Overview of results of H CO3 experiments (through- diffusion, clay core codes F-T and pulse injection - D2, clay core codes NRM) performed at normal ionic strength. The effective diffusion coefficient is the i product ηRDapp, the effective dispersion coefficient the product ηRD app (Aertsens et al., 2010) ...... 29 14 - Table 3: Results for the samples of the H CO3 injection campaign at high ionic strengths (0.1 mol/L, 0.5 mol/lL and 1.0 mol/L – horizontal cores H2 and H4, vertical cores V5 and V6). The effective dispersion coefficient i is the product ηRD app (Aertsens et al., 2010)...... 30 Table 4: Parameters used by Le Gal La Salle et al. (2013) to model profiles of Cl and 37Cl/35Cl across Jurassic mudrocks at Tournemire, Southern France, as determined by the radial diffusion method ...... 37 Table 5: Solubility of Se in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV...... 42 Table 6: Molybdenum (Mo): Kd values recommended values and upper and lower limits ...... 57 Table 7: Measured diffusion coefficients for HTO in mudrocks at three different localities (m2/s)...... 58 Table 8: Site types and distributions for "reference illite" (CEC = 0.2 eq/kg) ...... 61 Table 9: Selectivity coefficients for the "reference illite" (CEC = 0.2 eq/kg) ...... 61 Table 10: The results of electromigration experiments with 45Ca2+ at different electrical fields ...... 65 Table 11: Solubility of Ra in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV. Database: MOLDATA_R2. Code: The Geochemist's Workbench- 10.0 ...... 69 Table 12: Species distribution of Ra in equilibrium with RaSO4(s)...... 69 Table 13: The results of electromigration experiments with 226Ra2+ at different electrical fields ...... 72 Table 14: Solubility of Be in the BC reference porewater system at 25°C, pH 8.355 and Eh -281 mV. Database MOLDATA_R2 (adapted for Be-species from MINTEQ v2.4. Code: The Geochemist’s Workbench ...... 81 Table 15: Speciation of Nb(V) when concentrations are constrained by equilibrium with NaNbO3(s), Nb(cr) or Nb2O5(s) in the presence of BC reference porewater (composition from De Craen et al., 2004). The speciation and saturation indices were calculated using PHREEQC Interactive v2.18.3514 and the Thermochimie version 9.0 thermodynamic database (Duro et al., 2006a; Grivé et al., 2014) for a temperature of 16 ̊C. Note that Nb data in this version of the thermodynamic database are the same as those in v.7c used to calculate the results in Figure 38 above)...... 92 Table 16: Solubility of Nb in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV...... 94 Table 17: Species distribution of Nb in equilibrium with Nb2O5(cr)...... 94 Table 18: Solubility of Ni in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV. Database: MOLDATA. Code: The Geochemist's Workbench (from Salah and Wang, 2014)...... 100 3 Table 19: Calculated Kd values (m /kg) for nickel sorption to kaolinite KGa-1b under Finnish reference water conditions (from Hakanen et al., 2014)...... 105 3 Table 20: Calculated Kd values (m /kg) for nickel sorption to illite-IMt1 under Finnish reference water conditions (from Hakanen et al., 2014)...... 106 Table 21: Solubility of Pd in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV...... 110 Table 22: Solubilities of Ac recommended by Berner (2002) and Wersin and Schwyn (2004) for use in NAGRA’s Project Opalinus Clay safety assessment. The reference conditions are for an Na-Cl dominated porewater with an ionic strength of 0.323, pH of 7.25 and Eh of -193.6 mV...... 125 Table 23: Measured diffusion coefficients for HTO in mudrocks at three different localities (m2/s)...... 127 Table 24: Summary of the fitted parameters kdecomp, RRN-DOM and RRN (Maes et al., 2011)...... 141 Table 25: Solubility of Pu in the BC reference porewater system at 25°C, pH 8.355 and Eh -281 mV. Database: MOLDATA. Code: The Geochemist's Workbench - 10.0...... 145 Table 26: Species distribution of Pu in equilibrium with PuO2(am,hyd)...... 146 Table 27: Summary of the fitted parameters kdecomp, RRN-DOM and RRN (Maes et al., 2011)...... 156 Table 28: Solubility of Sm in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV...... 163 Table 29: Solubility of Np in the BC reference porewater system at 25°C, pH 8.355 and Eh -281 mV...... 167 Table 30: Summary of the fitted parameters kdecomp, RRN-DOM and RRN (Maes et al., 2011)...... 173 Table 31: Solubility of Th in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV...... 178 Table 32: Species distribution of Th in equilibrium with ThO2(am,hyd,aged)...... 178 Table 33: Th-sorption data on Boom Clay in absence (SBCW) and presence of DOM (RBCW) ...... 189 Table 34: Th(IV) sorption data on Na-montmorillonite (Bradbury and Baeyens, 2005) ...... 190 Table 35: Th(IV) sorption data on Na-illite (Bradbury and Baeyens, 2009) ...... 190 Table 36: Hydrolysis constants of different databases and publications ...... 191 Table 37: Solubility of Pa in the BC reference porewater system at 25°C, pH 8.355 and Eh -281 mV...... 196

SCK•CEN/12201513 Page 13 of 208 Compilation of Technical Notes on less studied elements

Table 38: Summary of the fitted parameters kdecomp, RRN-DOM and RRN (Maes et al., 2011)...... 203 Table 39: Thermodynamic data and geochemical conditions used by QUINTESSA for speciation and solubility calculations ...... 206

SCK•CEN/12201513 Page 14 of 208 Compilation of Technical Notes on less studied elements

1 Introduction Within the framework of the Belgian Safety and feasibility Case 1, SCK•CEN was asked by ONDRAF/NIRAS to provide scientific information on the retention and migration behavior of about 25 waste relevant radionuclides under chemical conditions representative for Boom Clay at the Mol site. Dissolved gases (i.e. H2, CH4, He, Ar, Xe) are not discussed in the current report.

As a detailed study of each waste relevant radionuclide is not achievable (due to time restrictions), a pragmatic approach was developed by SCK•CEN to ″subdivide″ all radionuclides into groups based on (assumed) similar chemical characteristics (e.g. oxidation/valence state) and behaviour, such as their speciation (hydrolysis, complexation), their sorption (affinity and mechanism) and migration.

Based on this approach, each group is represented by one or two ″key″ or ″reference elements″, which have been studied in detail and for which so-called ″phenomenological models″ have been elaborated. These phenomenological models represent the cornerstone of the developed methodology, describing in a qualitative and quantitative way the RN retention/migration behaviour under undisturbed Boom Clay (i.e. far-field) conditions at Mol under present-day conditions. Assigning consistent parameter ranges for the identified groups represents the last step of the described approach and the basis for a well founded long-term safety assessment.

Besides the reference elements, there are however waste relevant elements for which the retention/migration behavior is less or not known at all, as only limited experimental and/or literature data are available. In order to overcome this drawback, SCK•CEN together with QUINTESSA compiled so-called Technical Notes (TN) on the “less studied elements”, comprising literature information, speciation calculations, and the available experimental data with respect to sorption and migration. Based on the latter and calling upon analogy –where possible- it was tried to associate the “less studied elements” to the different groups. At the end of each TN the main lines of reasoning for the grouping are summarized in form of a justification.

An ANNEX was added, as the reader may notice that the Pourbaix diagrams and solubility values included in this report were not all calculated with the same geochemical code and thermodynamic database (TDB). As mentioned above, the compilation of the Technical Notes (TN’s) represented a common effort of QUINTESSA and SCK•CEN. QUINTESSA compiled the TN’s for which no experimental data were available at SCK•CEN and for which also only limited thermodynamic data exist. In order to provide however information on the general geochemistry, the speciation and transport behaviour for the elements belonging to this group (Ac, Ag, Mo, Nb, Ni, Pa, Sm, Zr), QUINTESSA performed a detailed literature review and summarized the most relevant outcome in the respective paragraphs. This explains why the style of the TN’s compiled by QUINTESSA and SCK•CEN are not uniform. Besides this, different calculation approaches and several TDB’s were used by QUINTESSA, due to which the results might be (slightly) different from the ones obtained by SCK•CEN (Salah and Wang, 2014). Despite this fact, no major differences were recognized. For further details concerning the thermodynamic databases that were used for the calculations and the geochemical conditions, it is referred to ANNEX I.

SCK•CEN/12201513 Page 15 of 208 Compilation of Technical Notes on less studied elements

As mentioned above, the developed methodology, i.e. the grouping approach relies on an extensive experimental program and datasets gathered during the last decades, and has been substantiated by geochemical calculations as well as different transport modeling approaches. As described in detail in Bruggeman and Maes (2016), besides similarities in the aqueous speciation and the sorption behavior, also similarities in the migration behaviour were observed, which lead to the subdivision of the radionuclides into four main groups characterized by an increasing complexity of their chemical behavior and interaction mechanisms. Although the elements belonging to one group may exhibit differences in their chemical behavior, the same phenomenological model remains valid. In order to narrow down these differences (e.g. in valence state or speciation) group II, III and IV were further subdivided in several subgroups. The 1st group comprises only HTO, which represents the reference element (Group I) for non-retarded transport, the 2nd group (Group II) comprises elements characterized by anion exclusion, and based on their different valence state, two subgroups were differentiated, i.e. IIa) the monovalent anions and IIb) the divalent anions. Reference elements for this group are I and Se, respectively. The 3rd group (Group III) is formed by elements characterized by cation exchange sorption. Also here 2 subgroups were discriminated, with IIIa) subgroup of monovalent alkaline cations and IIIb) subgroup of divalent alkali earth elements. Reference elements for the latter group are Cs and Sr. The largest group (Group IV) consists of elements showing a high affinity for dissolved organic matter, mainly present in form of colloids in the Boom Clay pore water, and mediating their transport through the clay. Within group IV, three subgroups have been differentiated, i.e. IVa) the subgroup of transition metals, IVb) the subgroup of trivalent lanthanides (La) and actinides (Ac) and IVc) the subgroup of tetravalent La and Ac. Pentavalent protactinium was included also in the latter subgroup. Reference elements for the 3 subgroups are Tc, Am and U.

In the last years, so-called Topical Reports (TR’s) were compiled for the above mentioned ″key″ or ″reference elements″ of each group/phenomenological model. They reflect the state-of-the-art knowledge on the chemistry and transport behavior of the respective element/group, relevant literature information, in-house obtained experimental data, and derived retention and migration parameters.

It is considered that within the TR’s all important processes/phenomena that may affect the retention and migration behaviour of the safety-relevant radionuclides under the current geochemical BC conditions are covered and discussed. Therefore, it is recommended to consult them in order to get a detailed overview on the current understanding of the ″group behavior″.

SCK•CEN/12201513 Page 16 of 208 Compilation of Technical Notes on less studied elements

IA VIIIA

IIA IIIA IVA VA VIA VIIA Be C

IIIB IVB VB VIB VIIB VIIIB IB IIB Cl Ca Ni Se

Rb Sr Zr Nb Mo Tc Pd Ag Sn I Cs

Ra Ac

Th Pa U Np Pu Am Cm

Figure 1: Elements belonging to group II: highlighted in blue, group III: highlighted in pink, group IVa highlighted in green, group IVb and IVc: highlighted in violet; (group I = HTO) not illustrated

Furthermore, it is referred to some other documents for further details on the experimental approach adopted by SCK•CEN for developing radionuclide sorption parameters, i.e. Bruggeman et al. (2008), on the Natural Organic Matter (NOM) in Boom Clay, i.e. Bruggeman and de Craen (2012), and on the developed phenomenological model for organic matter linked radionuclide transport, i.e. Maes et al. (2011b).

In the following, the different groups, reference elements and associated radionuclides are summarized.

Group I): Reference for non-retarded transport  HTO (tritiated water) Topical report (HTO): Bruggeman et al. (2013)

Group II) Elements characterized by anion exclusion Subgroup II a) Monovalent anions - -  HCO3 (bicarbonate), Cl- (chloride), HSe (biselenide) Reference elements: I, Se Topical report (I): Bruggeman et al. (2010a) Supporting documents: State-of-the-art-report (Se): De Cannière et al. (2011) Publication: Aertsens et al. (2008)

Subgroup II b) Divalent anions 2-  MoO4 (molybdate) No reference element

Group III) Elements characterized by cation exchange sorption Subgroup III a) Monovalent alkali cations  Rb Reference elements: Cs Topical report (Cs): Maes et al. (2011a)

SCK•CEN/12201513 Page 17 of 208 Compilation of Technical Notes on less studied elements

Subgroup III b) Divalent alkali earth cations  Ca, Ra Reference element: Sr Topical Report (Sr): Maes et al. (2012)

Group IV) Transport behaviour linked to DOM Subgroup IV a) Transition metals:  Ag, Be, Nb, Ni, Pd, Sn and Zr Reference element: Tc Topical Report (Tc): Bruggeman et al. (2010b)

Subgroup IV b) Trivalent lanthanides and actinides  Ac, Cm, Pu, Sm, Reference elements: Am Topical report (Am): Bruggeman et al. (2011)

Subgroup IV c) Tetravalent lanthanides and actinides (+ pentavalent protactinium):  Th, Np and Pa Reference element: U Topical report (U): Salah et al. (2015)

The reasoning behind the assignment of the radionuclides to the different groups is shortly outlined hereunder. For further details, it is referred to the previously mentioned reports and publications.

1.1 Group I: Reference element for non-retarded transport: Tritiated water (HTO) The first group comprises the reference conservative tracer HTO. HTO is very frequently used in both lab-scale migration experiments and in meter-scale in situ experiments, in order to obtain reference transport parameters for the Boom Clay. No other elements were assigned to this group, due to which this ″group″ is not further discussed in this report and it is referred to the Topical Report for further details.

1.2 Group II: Elements characterized by anion exclusion Reference elements: Iodine (I) and selenium (Se)

Based on the speciation calculations, C, Cl, Se and Mo are present as monovalent or divalent - - - 2- anionic species, i.e. HCO3 , Cl , HSe and MoO4 in the undisturbed BC porewater. As such, - they may represent ligands for metals and form complexes with them (especially HCO3 and Cl-). Sorption of these negatively charged species onto the negatively charged clay surfaces is – as expected – negligible or only weak under the slightly alkaline BC conditions. Consequently, also retardation of these species is considered to be low (0

effect. With respect to solubility, the carbon and chlorine concentrations are considered not to be solubility limited, while for molybdenum and selenium different solubility limiting solids have been proposed.

1.3 Group III: Elements characterized by cation exchange sorption Reference elements: Cs and Sr

Elements comprised in this group are monovalent alkaline (Rb+) and divalent alkaline earth elements/cations (Ca2+ and Ra2+). The speciation of the monovalent cations in the porewater is predicted to be simple, i.e. they are generally occurring as free cations (i.e. Cs+, Rb+), while the divalent cations tend to form inorganic hydroxyl- or carbonate complexes (i.e. SrCO3, RaCO3,aq). Organic complexation has been revealed to be not important for this group of elements. While the aqueous concentrations for the monovalent cations seem not to be solubility limited, Sr, Ca and Ra concentrations may be controlled by carbonates (i.e. calcite, strontianite) and/or sulphates (i.e. celestite). Solid solution formation may also play an important role for the before mentioned cations. Uptake by the host rock for the the elements of this group is mainly controlled by cation exchange processes related to the clay minerals. In the cases of the alkali metals Cs and Rb, micaceous clay minerals such as illite present the largest sorption sink because of the presence of the so-called frayed edge sites (FES). These sites have a high affinity (but low capacity) for cations with low hydration energy like Cs and Rb. The uptake of these elements depends mainly on the illite content. Sorption models like the 3-site ion exchange model (or 3-IEX, Bradbury and Baeyens, 2000) are suitable to describe the sorption onto Boom Clay under varying conditions. In the cases of alkali earth metals (Ca, Sr, Ra), sorption is dominated rather by ion exchange in the interlayer of swelling clays (smectites) and to a lesser extent by micaceous and other clay minerals (such as chlorite). Besides cation exchange, also surface complexation may play a role for the divalent cations. Sorption models like the 2-site protolysis non-electrostatic surface complexation/cation exchange (or 2 SPNE SC/CE, Bradbury and Baeyens, 1997) are suitable to describe the sorption onto the Boom Clay under varying conditions. The most significant sorption competitors with the radionuclides belonging to this group are the cations present in the porewater (Na, K, Mg, Ca). With respect to transport, double layer enhanced diffusion (i.e. surface diffusion) has been put forward as an important process for this group of elements.

1.4 Group IV: Elements characterized by transport behavior linked to DOM This group encompasses a wide range of different elements and may be further divided into three different subgroups: (1) the transition metals; (2) the trivalent lanthanides and actinides; and (3) the tetravalent (and pentavalent) actinides. Notwithstanding the extent of this group, there is a dominant feature common to all elements, namely DOM-related transport which is assumed to overrule the differences between the elements. Therefore, the phenomenological model for the three different subgroups is quite generic.

SCK•CEN/12201513 Page 19 of 208 Compilation of Technical Notes on less studied elements

With regard to speciation and under the geochemical conditions imposed by the Boom Clay all elements (the only exception being Ag) within this large group, feature either as a hydrolysed species or as a carbonate complex (or as a mixed hydroxy-carbonate complex). The hydrolysis or carbonate complexation of these elements implies also that their solubility is limited by the corresponding (oxy)hydroxide, carbonate or hydroxyl-carbonate precipitate. The tendency to form complexes in aqueous solution also translates itself into the formation of strong (inner-sphere) surface complexes with OH-containing surface functional groups. The strongest inorganic sorbents for these elements in the Boom Clay are therefore the clay minerals, mostly illite and smectite.

However, the strong hydrolyzing tendency of these elements also makes them susceptible to form complexes with organic functional groups present on DOM colloids. The nature of these "complexes" might be quite diverse due to the vast array of organic functional groups, and also due to different possible interaction mechanisms like uni-/multidentate/chelate complex formation, charge neutralization, and/or colloid-colloid association. However, the importance of the interaction of these elements with humic particles is undisputed.

With regard to transport, the complex speciation of these elements in the Boom Clay necessitates a different approach compared to the previously discussed groups. Especially the interaction with organic colloids can no longer be described by the classical diffusion- advection equation. Indeed, it has been observed for representative elements from all three subgroups that a pure equilibrium approach (assumed in diffusion-advection theory) is no longer valid. Apparently, dissociation of these elements from DOM colloids is dominated by a slow kinetic mechanism and this dissociation kinetics seems to depend mostly on the organic colloid itself rather than on the specific chemistry of the binding element. The transport behaviour of these elements must therefore be described by two conservation equations for diffusive-reactive transport. One for the transport of pure dissolved species of the element itself and a second one describing the transport of the species linked to the OM. A term to describe the kinetic exchange between the 2 species is furthermore needed.

A more detailed description of the 3 subgroups follows below.

1.5 Subgroup IVa: Transition metals (+ Be and Sn)  Reference element is Technetium (Tc/TcO(OH)2(aq)):

This group comprises the transition metals ″sensu strictu″, i.e. d-block elements of the periodic table, such as Ag, Nb Ni, Pd, and Zr, but additionally Be and Sn were associated to this group. The behavior of these elements is quite complex with respect to their occurrence in various oxidation states and their speciation. As revealed by the speciation calculations, - - some of them are strongly hydrolyzing, such as Nb(OH)6 , Pd(OH)2(aq), Sn(OH)5 , and 2- Zr(OH)4(aq), but they may also form carbonate complexes, such as Ni(CO3)2 , hydrogen 2- sulfide complexes, such as AgHS(aq), and/or occur as oxy-anions, like BeO2 . It is well known that strongly hydrolyzing elements have also a strong tendency to form polynuclear species and (eigen)colloids, respectively. With respect to sorption, information is quite scattered, which can be explained by the different speciation behavior. As can be seen above, the different elements occur predominantly as neutral species or anions. Consequently, their sorption is expected to be rather limited under the reference Boom Clay conditions, with the

SCK•CEN/12201513 Page 20 of 208 Compilation of Technical Notes on less studied elements

clay surfaces being negatively charged. It should be mentioned however, that thermodynamic data especially for this group of elements are limited, as e.g. for Be and Nb and often associated with high uncertainties, as for e.g. Pd and Sn. Therefore, the predicted speciation has to be considered with caution and may be different using another thermodynamic database. Besides this, studying sorption of redox-sensitive elements is always a challenge. Uptake of the different elements of this group is however reported to occur predominantly via cation exchange (when present as free cation), but also via surface complexation with the latter being pH dependent. The literature review revealed that sorption of these elements might be influenced by the presence of dissolved organic matter, which may decrease (e.g. Be see the respective paragraph), but also increase the sorption (e.g. Ag, see the respective paragraph). In the phenomenological model, the transport of the elements belonging to group IV is considered to be colloid facilitated, through formation of Rn-DOM pseudocolloids. Except for the reference element of this group, i.e. Tc, experimental evidence, i.e. migration experiments are lacking for the elements attributed to this group.

1.6 Subgroup IVb: Trivalent lanthanides and actinides (La/Ac) 2-  Reference element is Americium (Am/Am(CO3)3 ):

Elements associated to this group are Ac, Cm, Pu and Sm. While for Ac and Sm thermodynamic data, as well as sorption and migration data are limited or even lacking, the geochemistry of Pu and Cm has been extensively studied in the last decades and is quite well known. However, it should be mentioned, that despite the huge work on Pu that was performed in the past, the geochemical behavior of this element remains challenging and interpretation of experimental data is generally not straightforward, due to its complex redox-chemistry.

Sm belongs to the group of rare earth elements (REE) and also the chemical behavior of Ac appears to be similar to REE (especially La), which as a group has been extensively studied too. The lack of thermodynamic data for Ac is generally bridged by using Am-data assuming chemical analogy between Ac(III) and Am(III). With respect to inorganic speciation, the elements attributed to this group show a very consistent picture, i.e. forming all negatively 2− charged dicarbonato-complexes (La/Ac/REE(CO3) ). Experimental observations revealed however also a strong affinity of the trivalent lanthanides and actinides for the dissolved organic matter present in BC porewater, due to which it is assumed that a large fraction of these elements occurs also as humic-associated species (either as true complexes or as colloidal associations). Sorption of the trivalent La/Ac(III) is assumed to be high with surface complexation being the main mechanism. It follows from this expectation that the sorption of La/Ac(III) would thus rather be influenced by pH changes than by changes in the solution’s ionic strength. In contrast to elements of group IIIa, a large set of migration/diffusion data for the trivalent La/Ac has been compiled over the last years, which in fact enabled the development of the colloid facilitated transport and phenomenological model.

SCK•CEN/12201513 Page 21 of 208 Compilation of Technical Notes on less studied elements

1.7 Subgroup IVc: Tetravalent lanthanides and actinides (La/Ac) 4-  Reference element is Uranium (U/U(OH)4(aq)/UO2(CO3)3 ):

Elements comprised in this are Np, Th and Pa. In contrast to Np and Pa, which are redox- sensitive and may both occur in the tetra- and pentavalent oxidation state, Th is not redox- sensitive and occurs only in the tetravalent state. Based on speciation calculations performed with our in-house developed database MOLDATA, Np(OH)4(aq) is predicted as the major species under undisturbed BC conditions. But as for other tetravalent actinides, colloid formation of Np(IV) is an important process and should be taken into account when evaluating the speciation of Np. Besides this, tetravalent neptunium has been also observed to form complexes with humic and fulvic acids (see Np paragraph) and as reported by different authors, actinides are bound to the carboxylic and possibly phenolic groups of the humics as inner-sphere complexes. With respect to Th, a mixed Th-OH-CO3 complex (131- complex) is calculated to represent the dominant aqueous species in the BC porewater. Np(IV) generally sorbs strongly, as it is also strongly hydrolysing and surface complexation representing the main mechanism in neutral and slightly alkaline environments. Also the uptake of Th was shown to be high, but it was also revealed that the extent of sorption was influenced by the formation of intrinsic and pseudocolloids. The strong competition between dissolved humic substances and clays for complexation and sorption, respectively, was also evidenced for Np. With respect to solubility, amorphous oxides are generally assumed to control the aqueous actinide concentrations in the far-field, but as for sorption intrinsic colloid and pseudo-colloid formation may lead to “apparent solubilities exceeding the thermodynamic solubilities by several orders of magnitude. The strong influence of humic substances is also reflected in the transport behaviour, so that also for this group Rn-DOM facilitated transport was put forward in the phenomenological model.

With respect to Pa, it can be noted that only little is known about its complexation behaviour under groundwater conditions. In undisturbed BC porewater, Pa(OH)5(aq) is predicted to represent the prevalent species, it should however be mentioned that thermodynamic data for Pa are scarce. Concerning sorption, Bradbury and Baeyens (2003a, 2003b, see respective paragraph) state that despite the difficulties associated to work with protactinium, the available sorption data suggest strong sorption on almost all rock types under neutral to slightly-alkaline conditions. As for the other elements of the third group, surface complexation is expected to be the dominant sorption mechanism. Besides this, a strong affinity of Pa to sorb onto colloids is reported. Although experimental evidence is lacking, Pa- NOM association is assumed to govern the transport behaviour of pentavalent actinides.

1.8 References Aertsens, M., Van Gompel, M., De Canniere, P., Maes, N., and Dierckx, A. (2008). Vertical distribution of - (HCO3 )-C-14 transport parameters in Boom Clay in the Mol-1 borehole (Mol, Belgium). Physics and Chemistry of the Earth 33, S61-S66.

Bruggeman, C., Salah, S., Maes, N., Wang, L., Dierckx, A. and Ochs, M. (2008) Outline of the experimental approach adopted by SCK•CEN for developing radionuclide sorption parameters. Period 2007-2013. External Report, SCK•CEN-ER-73, SCK•CEN, Mol, Belgium.

SCK•CEN/12201513 Page 22 of 208 Compilation of Technical Notes on less studied elements

Bruggeman, C., Aertsens, M., Maes, N., and Salah, S., (2010a) Iodine retention and migration behaviour in Boom Clay. Topical Report, First Full Draft, SCK•CEN External Report, SCK•CEN-ER-119, Mol, Belgium.

Bruggeman, C., Maes, N., Aertsens, A., Govaerts, J., Martens, E., Jacops, E., Van Gompel, M., Van Ravestyn, L. (2010b). Technetium retention and migration behaviour in Boom Clay. Topical Report, First Full Draft, External Report, SCK•CEN-ER-101, SCK•CEN, Mol, Belgium.

Bruggeman, C., Salah, S., Maes, N. (2011) Americium retention and migration behaviour in Boom Clay. Topical Report, First full draft, SCK•CEN External Report SCK•CEN-ER-201, Mol, Belgium.

Bruggeman, C., De Craen, M. (2012). Boom Clay natural organic matter. Status report 2011. SCK•CEN- ER-206, SCK•CEN, Mol, Belgium.

Bruggeman, C. and Maes, N. (2016) Radionuclide migration and retention in Boom Clay. External Report, SCK•CEN-ER-0345, SCK•CEN, Mol, Belgium.

De Canniere, P., Maes, A., Williams, S. J., Bruggeman, C., Beauwens, T., Maes, N., and Cowper, M. M., (2010) Behaviour of Selenium in Boom Clay. State-of-the-art report. SCK•CEN-ER-120, SCK•CEN, Mol, Belgium.

Bruggeman C., Maes N., Aertsens M., De Cannière P. (2013) Tritiated water retention and migration behaviour in Boom Clay. External Report, SCK•CEN-ER-248, SCK•CEN, Mol, Belgium,

Maes, N., Salah, S., Bruggeman, C., Aertsens, M., and Martens, E., (2011a) Cesium retention and migration behaviour in Boom Clay. Topical report, First Full Draft, External Report, SCK•CEN-ER-153, SCK•CEN, Mol, Belgium.

Maes, N., Bruggeman, C., Govaerts, J., Martens, E., Salah, S., and Van Gompel, M. (2011b) A consistent phenomenological model for natural organic matter linked migration of Tc(IV), Cm(III), Np(IV), Pu(III/IV) and Pa(V) in the Boom Clay. Physics and Chemistry of the Earth 36, 1590-1599.

Maes N., Salah S., Bruggeman C., Aertsens M., Martens E., Van Laer, L. (2012) Strontium retention and migration behaviour in Boom Clay. Topical Report, First Full Draft, SCK•CEN-ER-197, SCK•CEN, Mol, Belgium.

Salah, S., Bruggeman, C., and Maes, N. (2015) Uranium retention and migration behavior in Boom Clay. Topical Report, Status 2014, External Report, SCK•CEN-ER-305, SCK•CEN, Mol, Belgium.

SCK•CEN/12201513 Page 23 of 208 Compilation of Technical Notes on less studied elements

2 GROUP II: Elements characterized by anion exclusion

2.1 Subgroup IIa: Monovalent anions

2.1.1 Technical note for Carbon (C)

2.1.1.1 General Carbon belongs to group 14 of the periodic table and its atomic number is 6. Carbon is non- metallic and the most common oxidation state of carbon in inorganic compounds is +4. In the tetravalent oxidation state, C is found as carbon dioxide in the atmosphere of the Earth and dissolved in all natural waters. In carbon monoxide and other transition metal carbonyl complexes, the oxidation state of C is +2. Inorganic carbon is a component of calcium and magnesium carbonates, such as limestones and dolomites, respectively.

Apart from the inorganic carbon species, it has to be stressed that the chemistry of carbon is widely diverse. Carbon forms more compounds than any other element, with almost ten million pure organic compounds described to date. Carbon is therefore widely distributed in nature and carbon compounds form the basis of all known life on Earth. In nature, also elemental carbon can be found and C is present in three allotropic forms: amorphous, graphite, and diamond. Graphite is one of the softest known materials while diamond is one of the hardest. Graphite exists in two forms: alpha and beta. These have identical physical properties, except for their crystal structure. All carbon allotropes are solids under normal conditions with graphite being the most thermodynamically stable form. In this technical note, we only consider carbon as inorganic carbonate/carbon dioxide type species.

Carbon has seven isotopes, of which three are naturally occurring, i.e. 12C (98.9%) and 13C (1.1%), representing stable isotopes and 14C being radioactive with a half-life of about 5,700 years. The latter is mainly produced by cosmic rays and in nuclear reactors through neutron activation of 14N, which is comprised in oxide fuels, coolant water and structural materials. A detailed overview about 14C generation in nuclear power plants is given by Wang et al. (2009).

2.1.1.2 Speciation and solubility Despite the fact, that BC porewater comprises also dissolved organic carbon, only the inorganic speciation is discussed in the following. In Figure 2a the Eh-pH domains for different natural waters and the boundaries (sulphide-sulfate boundary; organic carbon- carbonate boundary) for important geochemical reactions are illustrated. As mentioned above, carbon occurs in different oxidations states. The boundary for organic carbon, generally represented as CH2O oxidizing to carbonate (oxidation state of C = +IV) represents the transition from oxidation state 0 to +IV. Under even more reducing conditions, carbon may also occur in C(-IV), as in methane (Figure 2b). The carbonate concentration in the -2.44 reference porewater is imposed by a partial pressure of CO2(g) = 10 atm at pH 8.36 and - 25 °C, which corresponds to a HCO3 concentration of 0.014 M (log activity = -1.9). Figure 2b reveals that between pH ~6 to ~10 bicarbonate represents the stable species. Under more

SCK•CEN/12201513 Page 24 of 208 Compilation of Technical Notes on less studied elements

acidic and oxidizing conditions, carbonic acid (H2CO3 or here CO2(aq)) is the important species. In more alkaline environments (pH >10), the carbonate ion becomes dominant.

The carbon concentration is not solubility limited. However, bicarbonate is in equilibrium with calcite following the equations:

2+ 2- CaCO3(s) ⇔ Ca + CO3 2- - + H2O + CO3 ⇔ HCO3 + H - H2O + HCO3 ⇔ H2CO3(aq) H2CO3(aq) ⇔ H2O + CO2(g)

a) b)

- 2- HSO4 SO4

.5 CO2(aq)

- HCO3

Eh (volts) 2- 0 H2S(aq) CO3

Methane µ

–.5 HS- 25°C 0 2 4 6 8 10 12 14 pH

Figure 2: a) Generic Eh-pH diagram for water (taken from Brookins, 1988) and b) Eh-pH diagram of carbon (C-S-O- - -2 H) for the BC reference porewater system. Assumed activity of [HCO3 ] = 1.4 × 10 . Database: MOLDATA_R2. Code: The Geochemist's Workbench- 10.0.

2.1.1.3 Sorption and retardation Due to the presence of significant amounts of carbonate-type species (see equations above) in the Boom Clay, isotopic exchange with waste-derived carbon isotopes is foreseen. In particular, ingrowth of 14C in carbonate solid phases is a particular retention mechanism which may play a role during the slow diffusion through the formation.

14 - Besides precipitation/ingrowth in carbonates, also adsorption of H CO3 may be considered, both on oxide-type surfaces forming a surface complex through ligand exchange (Su and Suarez, 1997; Van Geet et al., 1994), or as part of a ternary surface complex with metal cations (Marques Fernandes et al., 2010).

14 - Up to now, there are no batch adsorption data for H CO3 on Boom Clay or on specific minerals under Boom Clay conditions. However, from percolation data a slight retention of 14 - H CO3 is observed, in line with expectations from literature (see further).

SCK•CEN/12201513 Page 25 of 208 Compilation of Technical Notes on less studied elements

2.1.1.4 Migration and diffusion 14 - Several experiments on H CO3 transport in Boom Clay have been performed, both in lab- scale setups and in situ on a larger scale in the HADES underground research facility (URF, Mol, Belgium). The results from these studies can be found in several reports and publications (Put and De Cannière, 1994; Aertsens et al., 2008; Aertsens et al., 2009; Aertsens et al., 2010; Aertsens et al., 2013; Aertsens, 2013).

14 - 131 Put and De Cannière (1994) describe through-diffusion tests with H CO3 , HTO and I. The tests are performed on clay pastes reconsolidated to a pressure of 2.0 or 4.4 MPa. During the tests, an exponential decreasing concentration of 14C in the inlet compartment was observed, which was interpreted as a loss of CO2 due to evaporation. The decrease in the inlet compartment was taken into account during modelling by assuming a decreasing exponential input concentration as model input. By doing this, fitting of the 14C quantity diffusing out of the clay core revealed similar ηR values (the product of the diffusion accessible porosity, η, and the retardation factor, R, between 0.07 and 0.18) as for I-. Hence, 14C was considered as a non-sorbing tracer and no (observable) isotopic exchange was said to have taken place. The range of values for the apparent diffusion coefficient, Dapp, was between 7.5×10-11 m²/s (for a consolidation pressure of 4.4 MPa) and 3.0×10-10 m²/s (for a consolidation pressure of 2.0 MPa).

14 - Aertsens et al. (2008; 2010) describe the transport parameters of H CO3 determined by pulse injection experiments on confined clay cores (32 mm length, 38 mm diameter) taken from different depths in the Mol-1 borehole (Mol, Belgium). The cores were percolated with Real Boom Clay Water (RBCW) sampled from the EG/BS piezometer in the HADES URF. The measured and fitted values for the experiments are presented in Table 1. Based on the measured hydraulic conductivity, K, a range (from 191 m to 281 m depth below drilling table, marked in brown in Table 1) was delineated in which the Boom Clay was considered 14 - homogeneous (with respect to transport). Within this range, the average Dapp of H CO3 in -11 the direction perpendicular to the bedding plane is (6±3)×10 m²/s. This average Dapp is i based on the linear extrapolation of the apparent dispersion coefficients, D app, versus the apparent velocity, Vapp, to zero velocity (Figure 3). The product ηR is 0.26±0.03 (Figure 4), lying between the values obtained for iodide (ηR = 0.16) and HTO (ηR = 0.37) on samples from the same borehole and in similar experimental setups.

SCK•CEN/12201513 Page 26 of 208 Compilation of Technical Notes on less studied elements

14 - Table 1: Results for the samples of the Mol-1 coring selection for the H CO3 injection campaign. The region where the Boom Clay can be considered as homogeneous is indicated in brown. The yellow line correponds to the double band. In the column 'Flow variation', the fits which are not excellent are coloured in red. These data are not taken up in the calculated averages. (Aertsens et al., 2010)

Strati- Clay Core Depth Module Hydraulic Pressure V Darcy Apparent App Disp η R Eff Disp Flow graphy Code [m BDT] Cell Conductivity Gradient Velocity Coeff Coeff Variation -1 -1 -1 -1 2 -1 2 -1 [—] [##×-MIG] [m] [—] [m s ] [bar m ] [m s ] [m s ] [m s ] [—] [m s ] [%]

Voort 31b-MIG 180.68-180.78 Module 7-3 1.5E-11 106.3 1.7 E-08 6.1E-08 2.5E-10 0.27 6.7E-11 6 41b-MIG 190.61-190.71 Module 6-4b 3.6E-10 3.8 1.4 E-08 2.9E-08 3.2E-11 0.50 1.6E-11 166 51b-MIG 200.55-200.65 Module 7-5 9.8E-12 106.3 1.1 E-08 4.2E-08 1.6E-10 0.26 4.1E-11 10 Transition Zone 62b-MIG 211.33-211.43 Module 7-13 1.8E-12 106.3 2.0 E-09 1.8E-08 1.1E-10 0.11 1.1E-11 546 70b-MIG 219.24-219.34 Module 5-12D 3.0E-12 375 1.2 E-08 4.3E-08 1.4E-10 0.27 3.8E-11 16 83b-MIG 232.13-232.23 Module 5-8D 4.9E-12 375 1.9 E-08 6.3E-08 1.6E-10 0.31 4.9E-11 136 Putte 90b-MIG 238.99-239.09 Module 5-1b 1.2E-12 390.6 4.8 E-09 1.5E-08 7.9E-10 0.31 2.5E-10 117 101b-MIG 249.78-249.88 Module 5-7D 2.3E-12 375 8.8 E-09 3.3E-08 1.1E-10 0.27 2.9E-11 54 111b-MIG 259.79-259.89 Module 7-2 4.7E-12 106.3 5.2 E-09 2.3E-08 1.3E-10 0.23 2.9E-11 22 Terhagen 121b-MIG 269.67-269.77 Module 5-13b 1.2E-12 384.4 4.9 E-09 2.0E-08 6.8E-11 0.25 1.7E-11 21 Belsele- 132b-MIG 280.63-280.73 Module 5-11 5.0E-12 106.3 5.5 E-09 2.5E-08 2.7E-10 0.22 5.9E-11 19 Waas 142b-MIG 290.63-290.73 Module 6-12c 1.7E-11 31.3 5.6 E-09 2.5E-08 3.2E-10 0.23 7.1E-11 593 Average 4.4E-12 3.5E-08 1.5E-10 0.26 3.7E-11 Standard deviation 2.8E-12 1.5E-08 6.2E-11 0.03 1.4E-11 BDT: Below Drilling Table.

a) 181 m Voort Formation 8 E-10 191 m Voort Formation Homogeneous Boom Clay

6 E-10 211 m /s) 2 239 m (m 291 m 4 E-10 Linear regression

2 E-10 Apparent dispersion coefficient

0 E+00 0 E+0 1 E-8 2 E-8 3 E-8 4 E-8 5 E-8 6 E-8 7 E-8 Apparent velocity (m/s)

2.0 E-10 b)

1.5 E-10 /s) 2 (m 1.0 E-10 Homogeneous Boom Clay NRM 96 97 5.0 E-11 alkaline plume/ Ecoclay Apparent dispersion coefficient Linear regression

0.0 E+00 0 E+00 1 E-08 2 E-08 3 E-08 4 E-08 5 E-08 6 E-08 7 E-08 Apparent velocity (m/s)

i Figure 3: Apparent dispersion coefficient, D app, versus the apparent velocity, Vapp. Figure a (top) concerns all clay cores from Mol-1 (see Table 1). The cores with bad fits are highlighted by different markers. Figure b (bottom) contains six clay cores from Mol-1 as well as the cores from the pulse injection experiments from Table 1. The line i is the best fit according to the expression (D app = Dapp + αVapp). In Figure a, the linear extrapolation concerns six clay cores in the "homogeneous range" of the Mol-1 series with good fitting results, while in Figure b, all shown points are included in the linear fit. (Aertsens et al., 2010)

SCK•CEN/12201513 Page 27 of 208 Compilation of Technical Notes on less studied elements

In Aertsens et al. (2010) the results from the pulse injection experiments are also compared with older published results from different experimental setups. These experiments, can be subdivided in two groups: (i) experiments conducted at "normal" Boom Clay pore water conditions (ionic strength ~ 0.016 mol/L), and (ii) experiments, aimed at studying the influence of ionic strength on the transport parameters (from 0.1 to 1.0 M). The data from the first group are summarized in Table 2 and consists of two types of experiments: through- diffusion and pulse injection experiments. The second group concerns only pulse injection experiments and the data are summarized in Table 3.

The through-diffusion tests in Table 2 are the ones described in Put and De Cannière (1994). It is remarked that these tests give directly a diffusion coefficient while the pulse injection experiments only provide dispersion coefficients. The pulse injection experiments reported in Table 2 have been obtained on cores taken from the HADES URF. All cores are perpendicular to the bedding plane. The first experiments were conducted in the early 90s (codes NRM002A/B and NRM019A/B; named NRM95-97 in Figure 3a. A second set of experiments was performed in the framework of research on the effect of an alkaline plume on the migration behaviour (Alkaline plume/Ecoclay). All points from the through-diffusion and pulse injection experiments are shown in Figure 3b. Fitting the data according to a linear -11 regression results in an apparent diffusion coefficient Dapp = (5.5±1.0)×10 m²/s.

The second set of experiments discussed in Aertsens et al. (2010) concerns pulse injection experiments performed at different ionic strength. The data shown in Table 3 are part of an exhaustive set of experiments, aimed at studying the influence of ionic strength on the 14 - transport parameters of HTO, iodide and H CO3 (Aertsens et al., 2009). Two horizontal (H2 and H4, percolation orientation parallel to the bedding plane) and two vertical (V5 and V6, percolation perpendicular to the bedding plane) Boom Clay cores were used in the experiments. The ionic strength was stepwise increased from RBCW to 0.1, 0.5 and 1.0 mol/L by addition of NaNO3 salt. An overview of the injection scheme is given in Aertsens et al. (2009).

The results clearly show the effect of increasing ionic strength on the transport parameters. The values of ηR are clearly higher than for RBCW and increase with ionic strength (average ηR = 0.39, 0.49 and 0.56 at 0.1 M, 0.5 M and 1.0 M, respectively). These results are in line with established theories of anion exclusion, which is also observed for iodide (and other anions) in Boom Clay (Moors, 2005; Aertsens et al., 2009; Bruggeman et al., 2010), as well as in other compacted clays (Garcia-Gutierrez et al., 2004; Van Loon et al., 2003; Van Loon et al., 2007).

14 - At higher ionic strength, ηR values for H CO3 become higher than the value for tritiated water, which lies around 0.37. The explanation given at this point is a slight retardation (R > 14 - 1) for H CO3 which might be attributable to isotopic exchange with stable carbonate and/or sorption. Under the assumption that iodide has the same diffusion accessible porosity as 14 - 14 - H CO3 and diffuses conservatively, the retardation factor R for H CO3 can be estimated and would amount to 1.6-2.0 under all ionic strength conditions (Aertsens et al., 2010).

SCK•CEN/12201513 Page 28 of 208 Compilation of Technical Notes on less studied elements

0.50

0.40 ( )

R 0.30 η

0.20

0.10 180 200 220 240 260 280 300 Depth (m)

Figure 4: The product ηR of the (diffusion) accessible porosity and the retardation factor as a function of depth. The clay cores for which the quality of the fit is not excellent (191 m, 211 m, 239 m and 291 m, see Table 1) are indicated by a square. The two horizontal lines are the maximal and minimal ηR obtained considering the results of the pulse injection experiments mentioned in Table 2 (Aertsens et al., 2010)

14 - Table 2: Overview of results of H CO3 experiments (through- diffusion, clay core codes F-T and pulse injection - D2, clay core codes NRM) performed at normal ionic strength. The effective diffusion coefficient is the product i ηRDapp, the effective dispersion coefficient the product ηRD app (Aertsens et al., 2010)

Through-diffusion Strati- Clay Core Fit Program App Diff η R Eff Diff graphy Code Sample Length C 0 a a Coeff Coeff [—] [##×-MIG] [cm] [Bq cm-3] [hour-1] [sec-1] [m2 s-1] [—] [m2 s-1]

F-T1 fit36(exp) 1.05 1402 3.47E-04 9.64E-08 7.5E-11 0.18 1.4E-11 Reconsolidated F-T2 fit36(exp) 2.1 6403 7.71E-04 2.14E-07 1.5E-10 0.12 1.8E-11 Boom Clay F-T3 fit36(exp) 2.15 5451 6.12E-04 1.70E-07 1.8E-10 0.10 1.8E-11 pastes F-T4 fit36(exp) 1.05 2497 3.37E-04 9.35E-08 2.2E-10 0.11 2.4E-11 F-T5 fit36(exp) 1.03 1069 1.07E-03 2.98E-07 3.0E-10 0.03 9.0E-12 Through Average 1.9E-10 0.11 2.0E-11 Diffusion Standard deviation 8.3E-11 0.05 4.5E-12

Pulse injection

Strati- Clay Core Fit program Module Flow V Darcy Apparent App Disp η R Eff Disp graphy Code Cell Variation velocity Coeff Coeff

-1 -1 2 -1 2 -1 [—] [%] [m s ] [m s ] [m s ] [—] [m s ] Confined undisturbed NRM002A D2_fit C1422 4 2.4E-09 8.7E-09 1.0E-10 0.27 2.8E-11 Boom Clay NRM002B D2_fit C1425 10 2.1E-09 7.1E-09 8.0E-11 0.30 2.4E-11 perpendicular to NRM019A D2_fit_V2 C1422 58 2.3E-09 8.3E-09 8.9E-11 0.27 2.4E-11 stratification NRM019B D2_fit_V2 C1425 56 2.0E-09 7.3E-09 6.7E-11 0.28 1.9E-11 NRM 95-97 Average 7.9E-09 8.5E-11 0.28 2.4E-11 Standard deviation 8.0E-10 1.5E-11 0.01 1.7E-13 Confined undisturbed NRM031A D2_fit_V2 mod8cel2 24 4.3E-09 1.5E-08 5.8E-11 0.28 1.6E-11 Boom Clay NRM031B D2_fit_V2 mod8cel3 12 6.0E-09 2.0E-08 6.8E-11 0.30 2.0E-11 perpendicular to NRM026A D2_fit_V2 mod8cel5 6 4.7E-09 1.6E-08 5.8E-11 0.29 1.7E-11 stratification NRM026B D2_fit_V2 mod8cel6 5 5.9E-09 1.9E-08 7.1E-11 0.31 2.2E-11 Ecoclay Average 1.8E-08 6.4E-11 0.29 1.9E-11 Standard deviation 2.3E-09 6.9E-12 0.01 7.9E-14

With regard to the apparent dispersion coefficient, it is noted that the orientation of the clay i cores has an influence on the D app value. Therefore the apparent dispersion coefficient is

SCK•CEN/12201513 Page 29 of 208 Compilation of Technical Notes on less studied elements

considered to be anisotropic and the anisotropy factor was calculated to be ~ 2.3 (Aertsens et al., 2003). It is remarked that this value is calculated based on dispersion coefficients, and not on diffusion coefficients. The ionic strength does not seem to have a significant influence on the apparent dispersion coefficient.

14 - -11 The effective diffusion coefficient for H CO3 (~ 1.4×10 m²/s in RBCW) is lower than the values for iodide (~ 2.2×10-11 m²/s) and HTO (~ 8.5×10-11 m²/s) (Aertsens et al., 2010). This order corresponds to the difference for the diffusion coefficients in water: 2.4×10-9 m²/s -9 - -9 14 - (HTO), 2.0×10 m²/s (I ) and 1.2×10 m²/s (H CO3 ) (Li and Gregory, 1974).

14 - The diffusion of H CO3 was also tested in a large scale in situ experiment in the HADES URF, coded TRIBICARB-3D (Aertsens et al., 2013). The set-up is shown in Figure 5. The experiment involves three piezometers equipped with a number of filters (between 4 and 8), placed in the Boom Clay next to the URF. Two piezometers are horizontal while the third one is inclined. 14 - Both HTO and H CO3 tracers are injected, through filter 6 of the R32-3 piezometer.

14 - Table 3: Results for the samples of the H CO3 injection campaign at high ionic strengths (0.1 mol/L, 0.5 mol/lL and 1.0 mol/L – horizontal cores H2 and H4, vertical cores V5 and V6). The effective dispersion coefficient is the i product ηRD app (Aertsens et al., 2010).

Ionic Clay Series V Darcy Apparent App Disp η R Eff Disp Flow strength Core Code Velocity Coeff Coeff Variation Code -1 -1 2 -1 2 -1 [—] [m] [m s ] [m s ] [m s ] [—] [m s ] [%]

0.1 mol/l H2 c 1.06E-08 2.58E-08 1.1E-10 0.41 4.3E-11 10 H4 c 9.34E-09 2.44E-08 9.9E-11 0.38 3.8E-11 9 V5 c 2.85E-09 7.85E-09 4.1E-11 0.36 1.5E-11 39 V6 c 2.96E-09 7.38E-09 4.8E-11 0.40 1.9E-11 21

0.5 mol/l H2 f 8.46E-09 1.70E-08 1.2E-10 0.50 6.1E-11 12 H4 f 7.54E-09 1.57E-08 1.2E-10 0.48 5.7E-11 9

1.0 mol/l H2 i 1.10E-08 1.94E-08 1.2E-10 0.57 6.9E-11 33 H4 i 9.80E-09 1.77E-08 1.2E-10 0.55 6.5E-11 33

SCK•CEN/12201513 Page 30 of 208 Compilation of Technical Notes on less studied elements

Figure 5: Set-up of the Tribicarb-3D experiment: three piezometers, each with a number of filters, are placed in the Boom Clay next to the URF. Both piezometers R34-1 and R32-3 are approximately parallel to one another and to the bedding plane of the clay. Piezometer R32-2 is inclined. Tracer (HTO and H14CO3-) is initially injected in filter 6 of piezometer R32-3 (Aertsens et al., 2013).

14 - The results for H CO3 are shown in Figure 6 (a-c). The data lower than ~ 1000 Bq/L show a 14 - large scatter. This is mainly due to the instability of the H CO3 tracer which tends to 14 evaporate from water under the gaseous form of CO2 (although measures have been taken to prevent this), and to the overlap of the 14C in the liquid scintillation spectrum with the HTO activity. Therefore only values > 1000 Bq/L are considered accurate and reliable.

SCK•CEN/12201513 Page 31 of 208 Compilation of Technical Notes on less studied elements

14 - Figure 6: Experimental data and blind prediction of the H CO3 evolution in the filters of the a) injection piezometer R32-3, b) the piezometer R34-1 parallel to the injection piezometer, and c) the inclined piezometer R32-2 of the Tribicarb-3D experiment.

14 - The transport of H CO3 in this experiment was predicted blindly and using a relatively simple analytical model (Aertsens et al., 2013). For this blind prediction Dapp values parallel the bedding plane of 1.2×10-11 m²/s and perpendicular to the bedding plane of 5.8×10-11 m²/s were used, together with an ηR value of 0.33. These values differ somewhat with best available laboratory results. The results from the blind prediction are shown in Figure 6 and show already a quite good agreement with experimental data (in the regions where these data are reliable). A fitting exercise was then performed, starting from two fixed values for ηR: 1) ηR = 0.33, the value of the blind prediction, and 2) ηR = 0.26, based on best available data

SCK•CEN/12201513 Page 32 of 208 Compilation of Technical Notes on less studied elements

for laboratory experiments. Both ηR values provided good fits to the data. With ηR = 0.26, -11 the fitted Dapp value parallel the bedding plane was 1.0×10 m²/s and perpendicular to the -10 bedding plane 1.1×10 m²/s. With ηR = 0.33, the fitted Dapp value parallel the bedding plane was 1.0×10-11 m²/s and perpendicular to the bedding plane 7.0×10-11 m²/s, which is closer to the values determined in the laboratory. Both fits were judged to be equally well reproducing the experimental data (Aertsens et al., 2013).

2.1.1.5 Justification The rationale behind the grouping for carbon is twofold, on the one hand it is revealed by - the speciation calculations that the bicarbonate anion (HCO3 ) represents the dominant aqueous species under BC reference conditions, and on the other hand by the different types 14 14 - of migration experiments showing no or only slight retardation of C/H CO3 . An additional argument to associate carbon to the anion group is supported by the observed influence of ionic strength on the transport behaviour, i.e. revealing anion exclusion.

2.1.1.6 References Aertsens, M., Van Gompel, M., De Cannière, P., Maes, N., Dierckx, A. (2008) Vertical distribution of 14 - H CO3 transport parameters in Boom Clay in the Mol-1 borehole (Mol, Belgium), Physics and Chemistry of the Earth, 33, S61-S66

Aertsens, M., De Cannière, P., Moors, H., Van Gompel, M. (2009) Effect of ionic strength on the 14 - transport parameters of tritiated water, iodide and H CO3 in Boom Clay, Scientific Basis for Nuclear Waste Management XXXIII, In: Materials Research Society Symposium Proceedings, Burakov, B.E., Aloy, A.S., Eds., 1193, 497-504

14 - Aertsens, M., Dierckx, A., Moors, H., De Cannière, P., Maes, N. (2010) Vertical distribution of H CO3 transport parameters in Boom Clay in the Mol-1 borehole (Mol, Belgium) and comparison with data from independent measurements, SCK•CEN-ER-66, SCK•CEN, Mol, Belgium, 31 pp.

Aertsens, M., Maes, N., Van Ravestyn, L., Brassinnes (2013) Overview of radionuclide migration experiments in the HADES Underground Research Facility at Mol (Belgium), Clay Minerals, 48, 153-166

Aertsens, M. (2013) Overview of radionuclide migration experiments in the HADES Underground Research Facility at Mol, SCK•CEN-ER-164, SCK•CEN, Mol, Belgium

Brookins, D.G. (1988) Eh-pH Diagrams for Geochemistry. Springer Verlag Berlin, Heidelberg, New York, London, Paris, Tokyo. 176 pages.

Bruggeman, C., Aertsens, M., Maes, N., Salah, S. (2010) Iodine retention and migration behaviour in Boom Clay, Topical Report, First full draft, SCK•CEN-ER-119, SCK•CEN, Mol, Belgium

Garcia-Gutiérrez, M., Cormenzana, J., Missana, T., Mingarro, M. (2004) Diffusion coefficients and accessible porosity for HTO and 36Cl in compacted FEBEX bentonite, Applied Clay Science, 26, 65-73

Li, Y.-H., Gregory, S. (1974) Diffusion of ions in sea water and deep sea sediments, Geochimica et Cosmochimica Acta, 38, 703-714

Marques Fernandes, M., Stumpf, T., Baeyens, B., Walther, C., Bradbury, M.H. (2010) Spectroscopic identification of ternary Cm-carbonate surface complexes, Environmental Science & Technology, 44, 921-927

SCK•CEN/12201513 Page 33 of 208 Compilation of Technical Notes on less studied elements

Moors, H. (2005) Topical report on the effect of the ionic strength on the diffusion accessible porosity of Boom Clay. SCK•CEN-ER-02, SCK•CEN, Mol, Belgium

Put, M.J., De Cannière, P. (1994) Migration behaviour of 14C labelled bicarbonate, HTO and 131I in Boom Clay, Radiochimica Acta, 66/67, 385-388

Su, C., Suarez, D.L. (1997) In situ infrared speciation of adsorbed carbonate on aluminum and iron oxides, Clays and Clay Minerals, 45, 814-825

Van Geen, A., Robertson, A.P., Leckie, J.O. (1994) Complexation of carbonate species at the goethite surface: Implications for adsorption of metal ions in natural waters, Geochimica et Cosmochimica Acta, 58, 2073-2086

Van Loon, L., Soler, J., Bradbury, M. (2003) Diffusion of HTO, 36Cl- and 125I- in Opalinus Clay samples from Mont Terri. Effect of confining pressure. Journal of Contaminant Hydrology, 61, 73-83

Van Loon, L., Glaus, M., Müller, W. (2007) Anion exclusion in compacted bentonites: Towards a better understanding of anion diffusion , Applied Geochemistry, 22, 2536-2552

Wang, L., Martens, E., Jacques, D., De Cannière, P., Berry, J., Mallants, D. (2009) Review of sorption values for the cementitious near-field of a near surface radioactive waste disposal facility. Project near surface disposal of category A waste at Dessel. NIRAS-MP5-03 DATA-LT(NF). NIROND-TR 2008-23 E.

SCK•CEN/12201513 Page 34 of 208 Compilation of Technical Notes on less studied elements

2.1.2 Technical note for Chlorine (Cl)

2.1.2.1 General Chlorine (Cl) is a halogen with an atomic weight of 35.4527 and an atomic number of 17 (CRC, 2011). Naturally it does not occur in significant quantities in a free state, although it does form in very small quantities transiently in the atmosphere, owing to the action of sunlight on ionic Cl, for example derived from seawater. Instead, almost all natural Cl occurs in a combination of aqueous solutions and a wide range of solid phases. In aqueous solution Cl occurs in anionic form and in a wide range of complexes. A number of natural solid phases have Cl as a major constituent, principally halite (NaCl), carnallite (KMgCl3·6H2O) and sylvite (KCl). Additionally, Cl occurs in minor or trace quantities in numerous hydroxyl-bearing solid phases, such as micas (including muscovite, biotite), amphiboles (including hornblende) and apatite.

Typically, chlorine exists in aqueous solution dominantly as the chloride ion, Cl-. Complexes with metal ions will form to some degree and become more important as pH becomes increasingly acidic, such that metals are hydrolysed to a decreasing extent. Cl may have several oxidation states: 7, 5, 3, 1, -1, but in nature its dominant oxidation state is -1.

Natural Cl is dominated by two stable isotopes, 35Cl (75.76%) and 37Cl (24.24%), but the radioactive isotope Cl-36, with a half-life of 3.01 x 105 years occurs in very small abundances and is widely used as an environmental tracer and for groundwater residence time studies.

2.1.2.2 Speciation and solubility When PHREEQC v3.1.1 is used together with the thermodynamic database llnl.dat to produce a geochemical model of the Mol reference porewater reported by De Craen et al. (2003) at 16°C, it is found that 99.8% of the chlorine in solution is in the Cl- form. The next most abundant Cl-bearing aqueous complex is calculated to be NaCl0, but this has an activity around two orders of magnitude smaller than Cl-. In progressively more saline waters, progressively more of the Cl becomes complexed with metal cations; calculations using PHREEQC v3.1.1 and the database llnl.dat show that in seawater at 16°C, 95.9% of the Cl is in the form of Cl-, while 3% is in the form of NaCl0, 1% is in the form of MgCl+ and 0.04% is in the form of CaCl+. Over relevant temperature ranges, the aqueous Cl speciation will vary very little.

The concentrations of Cl species will be influenced by hydration and dehydration reactions during diagenesis since these reactions consume and release water respectively (Worden, 1996). Since Cl will partition into the aqueous phase, consumption of water by hydration, solid-phase hydration will concentrate Cl in the remaining water. Water release by dehydration reactions will have the reverse effect.

SCK•CEN/12201513 Page 35 of 208 Compilation of Technical Notes on less studied elements

Figure 7: Temperature dependence of Cl speciation in the Mol reference water reported in De Craen et al. (2004). The speciation calculations were undertaken with PHREEQC version 3.1.1 and the thermodynamic database “llnl.dat”, which is distributed with the PHREEQC package.

Generally, except where Cl-bearing evaporite minerals are present in a rock (principally halite, although other Cl-bearing phases may also occur, such as sylvite and carnallite), aqueous concentrations of Cl will usually be well below any solubility limit (Figure 8). This is certainly expected to be the case in the Boom Clay at Mol, where the porewater has an ionic strength of only 0.015 molal.

Figure 8: Aqueous Cl concentrations limited by the solubility of three alternative Cl-bearing minerals, compared with the concentration of Cl in the Mol Reference Water reported in De Craen et al. (2004). Solubility calculations were undertaken with PHREEQC version 3.1.1 and the thermodynamic database “data0.ypf.R2” (USDOE, 2007). This database supports the Pitzer approach for calculating activity coefficients in highly saline solutions

2.1.2.3 Sorption and retardation Cl does not sorb significantly in natural sub-surface environments (e.g. Linklater et al., 2003). Solid phases in such environments have negligible negative anion exchange capacities, so

SCK•CEN/12201513 Page 36 of 208 Compilation of Technical Notes on less studied elements

that anion exchange reactions involving Cl- are also negligible. When modelled using the distribution coefficient approach, Kd is typically assigned a value of zero in radioactive waste management PA (Bradbury and Baeyens, 2003; Linklater et al., 2003; Quintessa and Geofirma Engineering Ltd, 2011). However, Linklater et al. (2003) report small distribution coefficient values (Rd) for the London Clay in the range 0.3 to 3.8 cm3 g-1 (3x 10-4 to 3.8 x 10–3 m3 kg-1). These values were obtained by batch and flow-through experiments.

There is some evidence that surface complexation reactions may influence the partitioning of Cl- between solid and aqueous phases. At low pH, there will be a positive charge at the edges of kaolinite sheets, onto which Cl- may sorb (Quirk, 1960; Sumner and Reeve, 1966). Generally, it is expected that any sorption will occur to a greater degree at lower pH when mineral surfaces will tend to have positive charges (Linklater et al., 2003).

There are also reports of Cl sorption being positively correlated with the presence of iron hydroxides in samples of kaolinite (Sumner and Reeve, 1966). This correlation has been interpreted to indicate sorption of Cl on the iron hydroxide by surface complexation. More acidic pH values would make this complexation stronger, owing to the surfaces of the iron hydroxide being more likely to retain a positive charge.

2.1.2.4 Migration and diffusion Le Gal La Salle et al. (2013) modelled Cl diffusion profiles across a Jurassic mudrock sequence at Tournemire in southern France. They found that a match between observations and model outputs could be attained using the parameter values in Table 4.

Table 4: Parameters used by Le Gal La Salle et al. (2013) to model profiles of Cl and 37Cl/35Cl across Jurassic mudrocks at Tournemire, Southern France, as determined by the radial diffusion method

Porewater Diffusion Cl-Accessible Formation Depth Interval (maSL) Coefficient, Dp, for Cl- Porosity θ (% (x10-11 m2 s-1) vol/vol) Upper Toarcian 1 558 513 1.7 8.0 Upper Toarcian 2 503 450 1.1 6.0 Upper Toarcian 3 450 430 1.2 6.4 Upper/Middle Toarcian 430 400 0.72 3.5 Middle Toarcian 400 380 0.8 7.2 Middle/Lower Toarcian 380 370 0.7 6.0 Lower Toarcian/Domerian 370 325 0.8 3.0 Domerian 325 298 1.4 4.0

In mudrocks the porosity that is accessible to diffusing halides is smaller than the water-filled porosity (Pearson, 1999). Barone et al. (1990) investigated the diffusion of chloride through shale having a relatively low water-content porosity (0.11). By comparing measurements of pore sizes in the shale with the size of the chloride ion, they concluded that Cl- diffusion could occur through at least 75%, but not all of the pore water.

SCK•CEN/12201513 Page 37 of 208 Compilation of Technical Notes on less studied elements

Wersin et al. (2013) report Cl profiles through a clay-rich Mesozoic rock sequence in northern Switzerland. They obtained different profiles using different methods (aqueous leaching and squeezing of rock core samples) and also determined Cl in one rock sample by an advective- displacement method and analysed Cl in one pumped groundwater sample. They found that consistent Cl concentrations could be modelled using the analytical data if a near-constant anion exclusion factor of about 0.5 was used. They noted that this value is consistent with that obtained by other studies.

Pearson (1999) discussed the concept of “geochemical porosity”, which is the proportion of the porosity in a rock within which chemical reactions of interest occur. Based on a review of published studies of several mudrocks he concluded that for halides the geochemical porosities would lie in the range 0.3 to 0.7 of other porosity values.

Hendry et al. (2000) studied the transport of radioactive and stable Cl isotopes (36Cl and 37Cl /35Cl respectively) through Cretaceous clay. They measured diffusion coefficients, effective porosities and total porosities on core samples and found that the diffusion coefficients lay in the range 1.5×10-10 m2s-1 to 1.6×10-10 m2s-1. Effective porosities (0.13-0.14) were found to be approximately 30% of the total porosity (0.36 – 0.38).

2.1.2.5 Justification As indicated by the speciation calculations, 99.8% of the chlorine in the BC pore water occurs in the Cl- form and only small distribution coefficients are reported in literature for clay rocks (e.g. London Clay: 3×10-4 to 3.8×10-3 m3 kg-1). No in-house data on chlorine/chloride sorption are available, but some evidence exists that surface complexation (SC) reactions may influence the partitioning of Cl- between the solid and aqueous phase with sorption/SC being higher/stronger at lower pH when mineral surfaces will tend to have (more) positive charges.

Unfortunately, migration of chlorine in BC has never been studied at SCK•CEN. But based on literature data and analogy to iodide, the transport behaviour of chlorine is considered to be (more or less) unretarded and influenced by anion exclusion.

2.1.2.6 References Barone F.S., Rowe R.K. and Quigley R.M. (1990) Laboratory determination of chloride diffusion coefficient in an intact shale. Canadian Geotechnical Journal, 27, 177-184.

Bradbury M.H. and Baeyens B. (2003) Near-field sorption data bases for compacted MX-80 bentonite for performance assessment of a high-level radioactive waste repository in Opalinus Clay host rock. Nagra Technical Report NTB02-18.

CRC (2011) Handbook of Chemistry and Physics, CRC Press, 92nd Edition.

De Craen M., Wang L., Van Geet M. and Moors H. (2004) The geochemistry of Boom Clay pore water at the Mol site, status 2004. SCK•CEN-BLG-990.

Hendry M.J., Wassenaar L.I. and Kotzer T. (2000) Chloride and chlorine isotope (36C1 and 37C1) as tracers of solute migration in a thick, clay-rich aquitard system. Water Resources Research, 36, 285 – 296.

SCK•CEN/12201513 Page 38 of 208 Compilation of Technical Notes on less studied elements

Le Gal La Salle C., Matray J.-M., Fethi B., Michelot J.-L., Dauzères A., Wittebroodt C., Frape S., Shouakar- Stashe O., Rebeix R. and Lancelot J. (2013) Modeling Cl- concentration and δ37Cl profiles in porewater across a 250m-thick indurated argillite at the Tournemire URL (France). Procedia Earth and Planetary Science, 7, 471 – 474.

Linklater C.M., Moreton A.D. and Tweed C.J. (2003) Analysis and Interpretation of Geosphere Sorption Data for a Nirex Performance Assessment. United Kingdom Nirex Limited, Nirex Report no. N/083.

Mazurek M., Alt-Epping P., Bath A., Gimmi T., Waber N., Buschaert S. (2011) Natural tracer profiles across argillaceous formations. Applied Geochemistry, 26, 1035–1064.

Pearson F.J. (1999) What is the porosity of a mudrock. In: Aplin AC, Fleet AJ and Macquaker JHS (eds) Muds and Mudstones. Physical and fluid flow properties. Geological Society, London, Special Publications, 157, 9-21.

Pearson F.J., Arcos D., Bath A., Boisson J.Y., Fernández A.M., Gäbler H.E., Gaucher E., Gautschi A., Griffault L., Hernán P. and Waber H.N; (2003) Geochemistry of water in the Opalinus Clay formation at the Mont Terri Rock Laboratory. Federal Office for Water and Geology, Bern, Switzerland, Series No. 5.

Quintessa and Geofirma Engineering Ltd (2011) OPG’s Deep Geologic Repository for Low- and Intermediate- Level Waste - Postclosure Safety Assessment: Data. NWMO Report NWMO DGR-TR- 2011-32.

Quirk (1960) Negative and positive adsorption of chloride by kaolinite. Nature, 188, 253 – 254.

Sumner ME and Reeve NG (1966) The effect of iron oxide impurities on the positive and negative adsorption of chloride by kaolinites. Journal of Soil Science, 17, 274–279.

United States Department of Energy (USDOE) (2007) In-Drift Precipitates/Salts Model. ANL-EBS-MD- 000045 REV 03.

Wersin P.,Waber N.H., Mazurek M., Mäder U.K., Gimmi T., Rufer D. and Traber D. (2013) Resolving Cl and SO4 profiles in a clay-rich rock sequence. Procedia Earth and Planetary Science, 7, 892 – 895.

Worden R.H. (1996) Controls on halogen concentrations in sedimentary formation waters. Mineralogical Magazine, 60, 259-274.

SCK•CEN/12201513 Page 39 of 208 Compilation of Technical Notes on less studied elements

2.1.3 Technical note for Selenium (Se)

2.1.3.1 General Selenium (atomic number: 34) is a redox-sensitive element and member of the sulfur family. As such, its chemistry is similar to that of sulfur. Natural Se comprises six stable isotopes, i.e. 74Se, 76Se, 77Se, 78Se, 80Se and 82Se. 79Se is a long-lived fission products (LLFP) and is considered as one of the main contributors to the dose-to-man for the disposal of spent fuel and high- level waste (HLW) in Boom Clay (De Cannière et al., 2010; Marivoet et al., 1999). Recently, a new value for the half-life of 79Se, i.e. 3.27 x 105 years was determined by Jörg et al. (2010). Selenium is a rare element in the Earth’s crust and also rarely occurs in its elemental state, but can be found in sulfide ores such as pyrite, where it partially replaces sulfur. Furthermore, Se occurs naturally in selenides (e.g. FeSe, HgSe, PbSe, ZnSe), selenates, which are analogous to sulfates and have similar chemistry, and selenites (e.g. Ag2SeO3, Na2SeO3).

A very detailed report on the selenium behaviour in Boom Clay was written by De Cannière et al. (2010). The current Technical Note does not pretend to be as detailed, but merely aims at delivering the main messages with respect to Se geochemistry in a reducing clay environment.

2.1.3.2 Speciation and solubility Selenium (Se) is redox-sensitive and occurs in several oxidation states, i.e. 0, –II, -I, +IV and +VI. Based on the speciation diagrams calculated with MOLDATA_R2 (Figure 9), hydrogen selenide anion (HSe-) is predicted to be the dominant aqueous Se species under the reducing conditions prevailing in Boom Clay.

- HSeO4 1

H2SeO3(aq) -- .5 SeO4 - HSeO3

Eh (volts) Se 0 H Se(aq) 4 -- 2 SeO3 µ - -- HSe Se3 –.5 25°C 0 2 4 6 8 10 12 14 pH

Figure 9: Eh-pH diagram of selenium (Se-C-S-O-H) for the BC reference porewater system. Assumed activities of dissolved [Se] = 10-8. Database: MOLDATA_R2. Code: The Geochemist's Workbench - 10.0.

SCK•CEN/12201513 Page 40 of 208 Compilation of Technical Notes on less studied elements

However, redox reactions between different Se oxidation states are known to be generally 2- kinetically slow and frequently mediated by microbial activity. Selenate (SeO4 ) is the most 2- oxidised form of Se and behaves chemically similar to sulphate (SO4 ). Thus, inorganic reduction in sediments is known to be extremely slow (or even non-existing) or to occur only on very specific minerals, such as green rust (Myneni et al., 1997) or Fe(0) (Sasaki et al., 2008; 2- Kvashnina et al., 2009). Up to now, no experimental evidence for SeO4 reduction under Boom Clay conditions is available.

2- In contrast, SeO3 , which is stable under mildly oxidising conditions, can be readily reduced in relatively short time periods on Fe(II)-containing minerals such as pyrite, mackinawite, magnetite, siderite (Breynaert et al., 2008; Scheinost and Charlet, 2008; Badaut et al., 2012) or by Fe(II) sorbed on minerals (Chakraborty et al., 2010). Depending on the geochemical conditions of the system, the Se reaction product was either Se(0) (Chakraborty et al., 2010; Badaut et al., 2012) or Se(-I/-II) (Scheinost and Charlet, 2008).

-3 2- Breynaert et al. (2010) contacted 1×10 M SeO3 with pyrite in SBCW for ~ 1 month and investigated the Se solid phase speciation by X-ray absorption spectroscopy. It was found 2- -5 -3 that Se(IV) was fully transformed into Se(0). Upon addition of SeO3 (5×10 – 5×10 M) to a Boom Clay batch system (~ 0.2 kg/L) and an equilibration time of ~ 1 month, a similar reduction was found to take place. In samples with lowest amount of administered Se, the reduction to Se(0) was quasi-complete, while in samples with highest levels of Se, residual Se(IV) was also identified (Figure 10).

Figure 10: X-ray absorption near-edge structure spectra for different Boom Clay fractions (A > 1 µm, 250 nm < B < 1 µm, 40 nm < C 250 nm) equilibrated with an initial Se concentration of 5×10-5 M ("5"), 5×10-4 M ("4") and 5×10-3 M ("3") (Breynaert et al., 2010)

2- As a conclusion, it can be stated that both SeO4 (kinetically hindered reduction) and more reduced forms of Se (Se(0) and Se(-I/-II)) are likely stable in Boom Clay. The presence of 2- SeO4 will depend on the source term: the mechanisms that govern dissolution of the Se- containing waste forms and the redox processes in the near field.

SCK•CEN/12201513 Page 41 of 208 Compilation of Technical Notes on less studied elements

2- Due to its very soluble character, no solubility limit is expected for SeO4 under undisturbed BC conditions in the absence of heavy metals, or large concentrations of earth alkaline cations, such as Ba2+ (De Cannière et al., 2010). Two elemental selenium polymorphs are comprised in MOLDATA_R2, i.e. monoclinic and trigonal selenium. The latter represents the thermodynamically more stable phase in the temperature range from 0 to 494.2 K, while the former phase is metastable and melts at 413 K (Olin et al., 2005).

Table 5: Solubility of Se in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV. Database: MOLDATA. Code: The Geochemist's Workbench- 10.0.

Solubility controlling phases Solubility, [Se], mol/L

Se (mono)1 2.0 × 10-6 Se (trig)1 5.7 × 10-7 1 -7 Fe3Se4 (gamma) 1.4 × 10 2 -9 Ferroselite (FeSe2,cr) 5.5 × 10 Achavalite (FeSe,s)2 1.1 × 10-9 Source data: 1NEA TDB, 2ANDRA TDB

Metal selenides are similar to sulphides characterized by low solubility products (Hummel et al., 2002). The stabilities of the iron selenides, i.e. achavalite [FeSe(s)] and ferroselite [FeSe2] depend on the one hand on the Fe concentration present in the porewater and on the other hand on the redox conditions. The solubilities of these two solids under Boom Clay conditions were calculated to correspond to 1.1×10-9 M and 5.5×10-9 M, respectively. According to Brookins (1988), ferroselite, i.e. Se analogue of pyrite occurs however only rarely in nature.

The reaction constants for the selenium solids comprised in (Table 7) and MOLDATA_R2 are the following:

2- + Se(mono) + O2(aq) + H2O ↔ SeO3 + 2 H log K = 25.06 2+ 2- Fe3Se4(gamma) + 5.5 O2(aq) + H2O ↔ 3 Fe + 4 SeO3 log K = 233.16 2- + Se(trig) + O2(aq) + H2O ↔ SeO3 + 2 H log K = 24.84 2+ 2- FeSe(s) + 1.5 O2(aq) ↔ Fe + SeO3 log K = 67.52 2+ + 2- FeSe2(cr) + 2.5 O2(aq) + H2O ↔ Fe + 2 H + 2 SeO3 log K = 91.22 (before: 90.78)

The solubility of solid iron selenide (FeSe) in synthetic and interstitial Boom Clay water was extensively studied by KU Leuven and AEA Technology (AEAT). The effect of Eh, organic matter, the presence or absence of pyrite and solid BC on the FeSe solubility was tested in different experimental set-ups. For details concerning these experiments, it is referred to the De Cannière et al. (2010). Results of these experiments revealed that FeSe appears to be very sensitive to oxidation, which may result in high apparent solubilities (8×10-6 to 2×10-5 M, AEAT) due to the presence and dissolution of oxidation products at the FeSe surface. By

SCK•CEN/12201513 Page 42 of 208 Compilation of Technical Notes on less studied elements

applying a pre-leaching step at the beginning of the experiments, a decrease of the solubility by several orders of magnitude (5×10-8 to 4.5×10-10 M) could be observed.

Similar experiments as for FeSe(s) were performed with elemental Se(s). The independently determined solubilities of Se(s) by AEAT and KU Leuven range between 4×10-8 to 3×10-7 M and 1.7×10-9 to 8×10-8, respectively. FeSe(s) and Se(s) solubilities appear not to be influenced by the presence of organic matter.

2.1.3.3 Sorption and retardation 2- SeO4 is known to be a non-sorbing compound to argillaceous rocks, including the Boom 2- Clay, at neutral to alkaline pH (De Cannière et al., 2010; Frasca et al., 2014). SeO4 was found neither to interact with dissolved Boom Clay organic matter (Bruggeman et al., 2007) nor with FeS2 (Bruggeman et al., 2002) (Figure 11). However, some authors did report sorption of 2- SeO4 onto hardened cement paste (HCP) and on important constituents of the cement matrix (Bonhoure et al., 2005).

75 2- 75 2- Figure 11: Evolution of SeO3 (a) and SeO4 (b) concentrations in FeS2 suspensions (2.5 g/L and 10 g/L) measured by a combination of ion chromatography and gamma counting (Bruggeman et al., 2002)

2- In contrast, SeO3 uptake on Boom Clay (Figure 12) and its constituting minerals (Figure 11) was experimentally observed (Bruggeman et al., 2002; Bruggeman et al., 2005; Montavon et 2- al., 2009; Breynaert et al., 2010). The uptake pattern of SeO3 on BC shows a clear kinetic behaviour with steady-state concentrations (value depending on initial Se concentration and solid-to-liquid ratio) being reached after approximately 1 month. Hereafter, in some samples a further decrease of Se solution concentrations was observed (Bruggeman et al., 2005). As discussed before, the uptake was followed by reduction of Se(IV) to Se(0) (Breynaert et al.,

SCK•CEN/12201513 Page 43 of 208 Compilation of Technical Notes on less studied elements

2010). Besides sorption, also co-precipitation and/or incorporation of selenite with/in calcite have been observed by different authors (e.g. Heberling et al., 2014 and references threrein). According to the obtained results, close-to equilibrium conditions favour rather sorption, while only supersaturation of the aqueous solution with respect to the bulk calcite-CaSeCO3 solid solution may lead to the incorporation of selenite in the calcite structure by substituting 2- for carbonate. SeO3 has also been observed to interact with dissolved Boom Clay organic matter, forming a colloidal Se phase (Bruggeman et al., 2005; Bruggeman et al., 2007).

2- Figure 12: Time evolution of SeO3 in supernatant solutions of batch experiments with Boom Clay (0.05 kg/L and 0.21 kg/L) at two initial Se concentrations (1×10-6 and 5×10-6 M) (Bruggeman et al., 2005)

Regarding Se(-II), no experiments have been performed to study HSe- uptake on Boom Clay or its constituting phases. Due to the difficulty in preparing (pure) Se(-II) solutions, similar uptake experiments are also rarely found in scientific literature. Naveau et al. (2007) and Liu et al. (2008) studied selenide uptake onto pyrite under reducing conditions. Both authors observed a fairly rapid removal of Se(-II) from solution. The removal process is likely a surface redox reaction in which FeS2 is reduced and Se(-II) oxidised to Se(0) (Liu et al., 2008). Iida et - 2- al. (2011) studied the sorption behaviour of HSe and Se4 onto granodiorite, sandy mudstone, tuffaceous sandstone and their major constituent and accessory minerals between pH 8.5 and 11.5. Sorption was mostly low but significant, with Kd values ranging between 0.2 and 82 L/kg. The dominant sorbing minerals for Se at neutral to alkaline pH were determined to be biotite and pyrite.

2.1.3.4 Transport and diffusion The migration behaviour of selenate in Boom Clay was studied by means of electromigration (Figure 13). The results obtained by Beauwens et al. (2005) confirmed the high mobility of the 2- -11 SeO4 anion, with apparent diffusion coefficients, Dapp, in the range between 1.7×10 and 6.2×10-11 m²/s. Also, no reduction of selenate in Boom Clay was observed.

SCK•CEN/12201513 Page 44 of 208 Compilation of Technical Notes on less studied elements

Figure 13: The electromigration experimental setup (Beauwens et al., 2005)

Figure 14: Experimental results from electromigration experiments with selenate source. Incomplete oxidation of the source (originally in selenite form) causes the simultaneous occurrence of an immobile and a mobile species (Beauwens et al., 2005)

75 2- Two percolation experiments (type C4) have been performed starting from SeO3 in confined Boom Clay cores. Details of these experiments can be found below:

1. Percolation C4 experiment Se75/2/1 (NRM010A) • Initiated 08/03/1995; under Ar • Clay core code R17 6.81-7.14 (06/09/1994) number 14 (vertical) • Initial activity 25 MBq, in chemical form sodium selenite • Se concentration in source solution 1.5×10-3 M • Total clay core length: 72 mm (42 mm "inlet" + 30 mm "outlet"), diameter 38 mm • Percolated solutions followed until 17/04/1997

SCK•CEN/12201513 Page 45 of 208 Compilation of Technical Notes on less studied elements

• Slicing was performed on 17/04/1997 (106 samples, average moisture content 17.9 weight%, calculated dry density 1.61 g/cm³)

2. Percolation C4 experiment Se75/2/3 (NRM010B) • Initiated 08/03/1995; under Ar • Clay core code R17 6.81-7.14 (06/09/1994) number 14 (vertical) • Initial activity 25 MBq, in chemical form sodium selenite • Se concentration in source solution 1.5×10-3 M • Total clay core length: 72 mm (42 mm "inlet" + 30 mm "outlet"), diameter 38 mm • Percolated solutions followed until 05/03/1998 • Slicing was performed on 05/03/1998 (142 samples, average moisture content 16.5 weight%, calculated dry density 1.64 g/cm³) • The hydraulic conductivity, K (m/s), for the two experiments is given in Figure 15. The value for K is in line with those normally observed in Putte Member of Boom Clay (Yu et al., 2013). The two clay cores exhibit approximately the same hydraulic conductivity. Over the course of the experiment, the conductivity appears to be slowly decreasing, but remains overall in the same range as the initial condition. • Calculated hydraulic conductivity, K (in m/s), as function of time in 2 percolation experiments (Se75m2c1 and Se75m2c3).

Hydraulic Conductivity 1.80 E-12 Se75m2c1 1.70 E-12 Se75m2c3

1.60 E-12

1.50 E-12

1.40 E-12

1.30 E-12 Hydraulic conductivity K conductivity (m/s) Hydraulic 1.20 E-12

1.10 E-12

1.00 E-12 0 200 400 600 800 1000 1200 Days since start experiment

Figure 15: Se concentration in outlet (75Se, as Bq/L, recalculated towards start of the percolation) as function of time

SCK•CEN/12201513 Page 46 of 208 Compilation of Technical Notes on less studied elements

Se concentration outlet 30,000,000

25,000,000

20,000,000

15,000,000 75 (Bq/L) Se - Se75m2c1 10,000,000 Se75m2c3

5,000,000

0 200 400 600 800 1000 1200 Days since start experiment

Figure 16: Se concentration in outlet (75Se, as Bq/L, recalculated towards start of the percolation) as function of time

Se concentration outlet

1.00 E-07

8.00 E-08

Se75m2c1 6.00 E-08 Se75m2c3

4.00 E-08 Se concentration (mol/L)

2.00 E-08

0.00 E+00 .0 50.0 100.0 150.0 200.0 250.0 Volume percolated since start of experiment (mL)

Figure 17: Se concentration in outlet (mol/L) as function of percolated volume (mL)

The outlet compositions of both experiments (Figure 16 and Figure 17) exhibit a very rapid breakthrough of a very small fraction of the administered Se in the form of a peak in the first 100 days of the experiment. The maximum Se concentration in the peak is different for the two experiments: 3×10-7 M for Se75/2/3 and 8×10-7 M for Se75/2/4. Although no speciation analysis has been performed on the percolated solutions, it is assumed that these peaks are 75 2- due to the presence of contaminations of SeO4 in the original source solutions 2- (Bruggeman et al., 2002). As SeO4 is not retarded in Boom Clay (Beauwens et al., 2005), the

SCK•CEN/12201513 Page 47 of 208 Compilation of Technical Notes on less studied elements

presence of such anionic species, would indeed result in a rapid outflow from the confined clay cores.

After this peak, a tailing is observed and percolated Se concentrations for the two experiments tend towards a similar value of ~ 3-4×10-9 M. Such concentrations are in line with predicted and observed solubilities of reduced Se phases. Again, no speciation analysis was performed on the outlet solutions, so these interpretations are speculative. It is also unclear whether the percolated Se is associated with organic matter or not.

After stopping the experiment, both clay cores were cut in ~ 0.5-1.0 mm slices and the 75Se bulk activity in the slices was measured. The obtained profiles are shown in Figure 18 and Figure 19. From the profiles it can be observed that the bulk of the administered Se is still located at or near the source position. It is unclear whether this is due to an adsorption or precipitation process. From Figure 19 it appears that the detailed profile over the clay cores shows a quite complex pattern. Apart from the high concentrations at the source position, a bell-shaped diffusion-type profile has developed around the source. At the side of the outlet, Se concentrations also appear to be increasing. The reason for the latter observation is unclear, but might be due to microbial activity in the outlet filters of the percolation experiment, immobilising percolated Se by reduction. The bell-shaped profile around the source could be attributed to a kinetically controlled sorption-reduction of the administered 2- SeO3 .

The information from the outlet concentration evolution and the concentration profile in the core, highlight the complex geochemistry of Se in the Boom Clay (Bruggeman et al., 2005). The occurrence of redox processes in the clay cores (and at the outlet filter) and the Se speciation in the source solution prove to be two main factors which need to be elucidated in order to come to a correct interpretation of the performed experiments. It is clear that the occurrence of multiple redox species hinders the derivation of diffusion coefficients from 2- such experiments. Assuming that the major part of the source solution was in the SeO3 form, the percolation experiments do confirm the immobilisation (likely due to sorption- reduction on solid surfaces) of Se(IV) in the Boom Clay.

SCK•CEN/12201513 Page 48 of 208 Compilation of Technical Notes on less studied elements

Se concentration profile in clay core 12000000

10000000

Se75m2c1 Se75m2c3 8000000

6000000

75 concentration (Bq/g) 75 concentration 4000000 Se -

2000000

0 -40 -30 -20 -10 0 10 20 30 40 50 Distance from source position (mm)

Figure 18: Se concentration profile (in Bq/g) in the two clay cores, obtained after ~ 800-1000 days percolation. The profiles are given relative to the position of the administered 75Se source (in mm). The percolation direction is from left to right.

Se concentration profile in clay core 100000000

10000000

Se75m2c1 1000000 Se75m2c3

100000

10000 75 concentration (Bq/g) 75 concentration 1000 Se -

100

10 -40 -30 -20 -10 0 10 20 30 40 50 Distance from source position (mm)

Figure 19: Se concentration profile (in Bq/g, logarithmic scale) in the two clay cores, obtained after ~ 800-1000 days percolation. The profiles are given relative to the position of the administered 75Se source (in mm). The percolation direction is from left to right.

SCK•CEN/12201513 Page 49 of 208 Compilation of Technical Notes on less studied elements

2.1.3.5 Justification According to the speciation calculations, the hydrogen selenide anion (HSe-/Se-II) represents the dominant aqueous species in BC porewater under undisturbed conditions. Its migration should however be limited by the low solubility of elemental Se and/or iron selenide. As revealed by the different sorption and migration studies, selenium is characterized by a complex geochemistry and the occurrence of multiple redox species. The concomitant - 2- - presence of oxy-anions besides HSe , such as SeO4 and SeO3 - normally stable under more oxidizing conditions - has been referred to redox disequilibrium in the BC porewater and 2- different kinetic reaction/redox rates. While selenate (SeO4 /SeVI) was found to behave as an unretarded element of high mobility, the results of the migration experiments using selenite 2- (SeO3 /SeIV) were less conclusive. While the outlet concentrations seem to indicate a solubility control and complexation with DOM could not be excluded, the slicing of the core and concentration profile rather revealed control of the Se(IV) species by some kinetically controlled sorption-reduction processes. The formerly said, it is clear that the reasoning concerning the grouping is not straightforward. Despite this fact, selenium was associated to the group of anions, mainly based on the speciation.

2.1.3.6 References Badaut, V., Schlegel, M.L., Descostes, M., Moutiers, G. (2012) In situ time-resolved X-ray near-edge absorption spectroscopy of selenite reduction by siderite, Environmental Science & Technology, 46, 10820-10826

Beauwens, T., De Cannière, P., Moors, H., Wang, L., Maes, N. (2005) Studying the migration behaviour of selenate in Boom Clay by electromigration, Engineering Geology, 77, 285-293

Bonhoure, I., Baur, I., Wieland, E., Johnson, C.A., Scheidegger, A.M. (2005) Uptake of Se(IV/VI) oxyanions by hardened cement paste and cement minerals: An X-ray absorption spectroscopy study, Cement and Concrete Research, 36, 91-98

Breynaert, E., Bruggeman, C., Maes, A. (2008) XANES-EXAFS analysis of Se solid-phase reaction products formed upon contacting Se(IV) with FeS2 and FeS, Environmental Science & Technology, 42, 3595-3601

Breynaert, E., Scheinost, A.C., Dom, D., Rossberg, A., Vancluysen, J., Gobechiya, E., Kirschhock, C.E.A., Maes, A. (2010) Reduction of Se(IV) in Boom Clay: XAS solid phase speciation, Environmental Science & Technology¸44, 6649-6655

Bruggeman, C., Vancluysen, J., Maes, A. (2002) New selenium solution speciation method by ion chromatography + gamma counting and its application to FeS2-controlled reducing conditions, Radiochimica Acta, 90, 629-635

Bruggeman, C., Maes, A., Vancluysen, J., Vandemussele, P. (2005) Selenite reduction in Boom Clay: Effect of FeS2, clay minerals and dissolved organic matter, Environmental Pollution, 137, 209-221

Bruggeman, C., Maes, A., Vancluysen, J. (2007) The interaction of dissolved Boom Clay and Gorleben humic substances with selenium oxyanions (selenite and selenate), Applied Geochemistry, 22, 1371- 1379

Brookins D. G. (1988) Eh-pH Diagrams for Geochemistry, Springer Verlag, 176 pp.

SCK•CEN/12201513 Page 50 of 208 Compilation of Technical Notes on less studied elements

Chakraborty, S., Bardelli, F., Charlet, L. (2010) Reactivities of Fe(II) on calcite: selenium reduction, Environmental Science & Technology, 44, 1288-1294

De Cannière, P., Maes, A., Williams, S., Bruggeman, C., Beauwens, T., Maes, N., Cowper, M. (2010) Behaviour of Selenium in Boom Clay. External Report, SCK•CEN-ER-120, 328 pp.

Frasca, B., Savoye, S., Wittebroodt, C., Leupin, O.X., Michelot, J.-L. (2014) Comparative study of Se oxyanions retention on three argillaceous rocks: Upper Toarcian (Tournemire, France), Black Shales (Tournemire, France) and Opalinus Clay (Mont Terri, Switzerland), Journal of Environmental Radioactivity, 127, 133-140

Heberling, F., Vinograd, V.L., Polly, R., Gale, J.D., Heck, S., Rothe, J., Bosbach, D., Geckeis H., Winkler B. (2014) A thermodynamic adsorption/entrapment model for selenium(IV) coprecipitation with calcite. Geochimica et Cosmochimica Acta, 134, 16-38.

Hummel, W., Berner, U., Curti, E., Pearson, F.J., Thoenen, T. (2002) NAGRA/PSI Chemical Thermodynamic DataBase 01/01, NAGRA Technical Report 02-16, Universal Publishers, Parkland, Florida.

Iida, Y., Tanaka, T., Yamaguchi, T., Nakayama, S. (2011) Sorption behavior of selenium(-II) on rocks under reducing conditions, Journal of Nuclear Science and Technology, 48, 279-291

Jörg, G., Bühnemann, R., Hollas, S., Kivel, N., Kossert, K., Van Winckel, S., v. Gostomski, C. L. (2010): Preparation of radiochemically pure 79Se and highly precise determination of ist half-life, Applied Radiation and Isotopes, 68, 2339-2351.

Kvashnina, K.O., Butorin, S.M., Cui, D., Vegelius, J., Puranen, A., Gens, R., Glatzel, P. (2009) Electron transfer during selenium reduction by iron surfaces in aqueous solution: high resolution X-ray absorption study, Journal of Physics: Conference Series, 14th International conference on X-ray absorption fine structure (XAFS14), 190, 012191

Liu, X., Fattahi, M., Montavon, G., Grambow, B. (2008) Selenide retention onto pyrite under reducing conditions, Radiochimica Acta, 96, 473-479

Marivoet, J., Volckaert, G., Labat, S., De Cannière, P., Dierckx, A.., Kursten, B., Lemmens, K., Lolivier, P., Mallants, D., Sneyers, A., Valcke, E., Wang, L., Wemaere, I. (1999) Values for the near-field and clay parameters used in the performance assessment of the geological disposal of radioactive waste in the Boom Clay formation at the Mol site (volume 1 and 2), Report to NIRAS/ONDRAF, Geological disposal of conditioned high-level and long-lived radioactive waste. Contract CCHO-98/332 – KNT 90.98.1042 Task 6.1. SCK•CEN Report R-3344

Montavon, G., Zuo, Z., Lützenkirchen, J., Alhajji, E., Kedziorek, M.A.M., Bourg, A.C.M., Grambow, B. (2009) Interaction of selenite with MX-80 bentonite: Effect of minor phases, pH, selenite loading, solution composition and compaction, Colloids and Surfaces A: Physicochem. Eng. Aspects, 332, 71-77

Myneni, S.C.B., Tokunaga, T.K., Brown Jr., G.E. (1997) Abiotic selenium redox transformations in the presence of Fe(II, III) oxides, Science, 278, 1106-1109

Naveau, A., Monteil-Rivera, F., Guillon, E., Dumonceau, J. (2007) Interactions of aqueous selenium(-II) and (IV) with metallic sulfide surfaces, Environmental Science & Technology, 41, 5376-5382

Olin, A., Nolang, B., Ohman, L., Osadchii, E., Rosen, E. (2005) Chemical Thermodynamics of Selenium, OECD NEA, Issy-les-Moulineaux, France.

Sasaki, K., Blowes, D.W., Ptacek, C.J., Gould, W.D. (2008) Immobilization of Se(VI) in mine drainage by permeable reactive barriers: column performance, Applied Geochemistry, 23, 1012-1022

SCK•CEN/12201513 Page 51 of 208 Compilation of Technical Notes on less studied elements

Scheinost, A.C., Charlet, L. (2008) Selenite reduction by mackinawite, magnetite and siderite: XAS characterization of nanosized redox products, Environmental Science & Technology, 42, 1984-1989.

SCK•CEN/12201513 Page 52 of 208 Compilation of Technical Notes on less studied elements

2.2 Subgroup IIa: Divalent anions

2.2.1 Technical note for Molybdenum (Mo)

2.2.1.1 General Molybdenum (Mo) is a strongly siderophile and chalcophile Group 6 transition metal with an atomic weight of 95.94 and an atomic number of 42 (CRC, 2011). The element is not found in nature as the native metal, but instead predominantly occurs naturally in about 45 Mo- bearing mineral species (Anthony et al., 1990, 1995, 1997, 2000, 2003). The most important ore mineral is molybdenite (MoS2), but wulfenite (PbMoO4) and powellite (Ca(MoW)O4) are also minor ore minerals. Molybdenum also tends to be associated with copper and tungsten and is a by-product of some copper and tungsten mining operations. Mo occurs widely at trace concentrations in association with organic matter and sedimentary sulphide minerals, particularly in black shale. The purified metal is silvery white and very hard, but it has greater ductility than tungsten. Among the commonly used metals, molybdenum has the third highest boiling point, after tungsten and tantalum.

There are seven naturally-occurring isotopes of Mo. The most abundant is 98Mo (24.13%), followed by 96Mo (16.68%), 95Mo (15.92%), 92Mo (14.84%), 100Mo (9.63%), 97Mo (9.55%), and 94Mo (9.25%). All these isotopes are stable, except for 92Mo and 100Mo, which can be treated as though they are stable, owing to their extremely long half-lives of >3 x 1017 years and 1 x 1019 years respectively. In addition there are 30 other isotopes and isomers that do not occur naturally.

The abundance of Mo in the earth’s crust is 1200 ppb by weight (CRC, 2011). In mean seawater Mo has an abundance of 10 µg/l (CRC, 2011), while in contrast terrestrial surface waters have more variable concentrations, reflecting in part the varied rock types within different drainage basins, anthropogenic inputs and varied drainage fluxes. River water from 15 major USA drainage basins is reported to have concentrations ranging from 2 to 1500 μg/L, with a mean concentration of 60 μg/L (WHO, 2011). In contrast De Vos and Tarvainen (2006) state that 807 European river water analyses yielded a median content of 0.22 μg/L, although concentrations range between <0.01 μg/L and 10.1 μg/L (excluding an outlier of 16 μg/L). Levels in groundwater samples from the U.S.A. are reported to range from undetectable to 270 μg/L (WHO, 2011).

Smedley et al. (2014) report Mo concentrations in British groundwaters and surface waters. The Mo concentrations are mostly 2 μg/L or less, with analyses of 11,600 British streamwater samples showing a 10–90th percentile range of 0.08–2.44 μg/L with a medi an value of 0.57 μg/L and a maximum value of 230 μg/L. The higher values were found generally to occur in samples from streams flowing over clay-rich rock formations and rocks with sulphide- mineralisation. Localised urban and industrial contamination also caused higher values in some cases.

Analyses of 1735 groundwater samples from across Britain were found by Smedley et al. (2014) to have a 10–90th percentile range for Mo of 0.035–1.80 μg/L with a median of 0.20 μg/L and maximum observation of 89 μg/L. Relatively high values occurred in some Lower Cretaceous greensand, Carboniferous limestone and mudstone (Coal Measures) aquifers, especially where conditions are anaerobic. It was concluded by Smedley et al. (2014) that

SCK•CEN/12201513 Page 53 of 208 Compilation of Technical Notes on less studied elements

reductive dissolution of Fe- oxides and possibly Mn- oxides could have released Mo, thereby causing the high observed groundwater concentrations. However, rarer samples from aquifers with sulphate-reducing conditions showed lower concentrations of Mo, possibly due to the precipitation of Mo-bearing sulphide minerals.

2.2.1.2 Speciation and solubility The aqueous geochemistry of Mo is complex because the element is redox sensitive and is able to form complex polynuclear complexes (e.g. Duro et al., 2010; Wersin et al., 2014). There are five main oxidation states (+II, +III, +IV, +V and +VI), although only the +VI and +IV states are important under the natural conditions of relevance to a deep repository project. There are few thermodynamic data available and those that are available are mostly of uncertain reliability. In the LLNL V8 R6 "combined" dataset, which is distributed with the Geochemist’s Workbench (GWB) software (Bethke, 1996, 2008) as “thermo.com.V8.R6”, the 2 only aqueous Mo species is MoO4 and the only solid phases are Mo and MoSe2. The Visual MINTEQ database, release 2.40, which is also distributed with GWB also contains data for 2- MoO4 , but also 12 inorganic complexes and 9 organic complexes. This database also contains data for 22 solid Mo phases, including MoS2 and Powellite (CaMoO4).

The aqueous speciation of Mo in BC reference water, calculated using GWB and each of these databases is illustrated in Figure 20. It can be seen from these figures that in the BC 2- porewater (mean Eh = -341 mV and mean pH = 8.31; De Craen et al., 2004), MoO4 is predicted to be the dominant aqueous species. The calculations using the MINTEQ and thermo.com.V8.R6 give MoS2 or MoSe2, respectively as possible stable phases under BC conditions. It should be noted that while the former is not unreasonable, MoSe2 is not a naturally-occurring mineral phase.

Reported calculations of Mo speciation in groundwater using other databases have also concluded that under likely pH-Eh conditions in a deep geological repository for SF, the 2- dominant aqueous Mo species will be MoO4 (e.g. Duro et al., 2010; Wersin et al., 2014). For calculations of Mo solubilities in saline water (ionic strength 0.24 – 0.32) and brine (ionic strength 6.47), such as might occur in the near-field of a future Canadian deep geological repository for spent fuel (SF), Duro et al. (2010) used the YMP database (USDOE, 2007) and the ThermoChimie v.7b/SIT database (Duro et al., 2006; Grivé et al., 2014). Both of these databases support activity coefficient models that are appropriate for very high water salinity. 2- The only aqueous Mo species in the YMP Pitzer database is MoO4 , but 8 solid phases, including Molybdite (MoO3) and Powellite (CaMoO4) are included. In contrast, the ThermoChimie v.7b/SIT TDB has 10 aqueous Mo species, including species with differing 2- 3+ + redox states (e.g. MoO4 (+VI), Mo (+III), and Mo2O5(OH) (+V)) and 16 solid phases, including MoS2 and CaMoO4.

SCK•CEN/12201513 Page 54 of 208 Compilation of Technical Notes on less studied elements

Figure 20: Eh-pH diagrams of molybdenum (in the system Mo-Ca-C-S-Cl-F-O-H for a) to c) and in the same system plus Se for d)) for the BC reference porewater from De Craen et al. (2004), calculated using Geochemist’s Workbench version 7.0 and assuming an activity of dissolved Mo, [Mo] = 10-5 . a) Calculated using the Visual MINTEQ thermodynamic database release 2.40 with no minerals or species suppressed. b) Calculated using the Visual MINTEQ release 2.40 with all minerals suppressed. c) As for a), but calculated using the LLNL V8 R6 "combined" dataset, thermo.com.V8.R6. d) As for b), but calculated using the LLNL V8 R6 "combined" dataset, thermo.com.V8.R6.full and with activity of Se, [Se] = 10-8.

The solubility of solid Mo-phases depends strongly on pH, Eh and aqueous Ca concentration, which, combined with the uncertainties in thermodynamic data, means that quantitative estimates of solubilities under in-situ conditions are challenging (Duro et al., 2010; Wersin et al., 2014).

For saline water (ionic strength 0.24 – 0.32) and brine (ionic strength 6.47), such as might occur in the near-field of a future Canadian deep SF repository, Duro et al. (2010) calculated the solubility of MoO2 and CaMoO4 using the ThermoChimie v.7b/SIT database and the solubility of CaMoO4(s) using the YMP Pitzer database. These workers concluded that MoO2(s) would be the solubility-controlling phase at an Eh of -200 mV and near-neutral pH

SCK•CEN/12201513 Page 55 of 208 Compilation of Technical Notes on less studied elements

for both considered waters. However, at more oxidizing, alkaline conditions, above a pH of 8.6, CaMoO4(s) could become the solubility-controlling phase. Under the reducing in-situ conditions expected in a deep geological repository post-closure, Duro et al. (2010) -8 concluded that MoO2(s) would control concentrations in the order of 10 mol/L in the fresh water and in the order of 10-13 mol/L for the brine.

For a proposed SF repository crystalline rock at Olkiluoto, Finland, Wersin et al. (2014) also used the Thermochimie database to calculate solubilities of Mo. They recommended reference solubilities of 1.3×10-7 mol/L for saline water and 4.7×10-8 mol/L for brackish water at this site.

Figure 21 shows solubility diagrams calculated for Mo in the presence of the BC reference water reported in De Craen et al. (2004) using the Visual MINTEQ thermodynamic database release 2.40 and the LLNL V8 R6 "combined" dataset, “thermo.com.V8.R6”. The former database implies a solubility control by MoS2, whereas the latter suggests that MoSe2 could be a solubility-limiting phase, should Se concentrations be within the range of reasonable solubility controls for that element (see the Technical Note on Se). In the former case, solubilities in the BC porewater, with a pH of 8.3 would be around 10-17 mol/L, whereas the latter case suggest 10-65 mol/L.

Figure 21: Solubility diagrams for molybdenum (in the system Mo-Ca-C-S-Cl-F-O-H for a) and in the same system plus Se for b)) for the BC reference porewater from De Craen et al. (2004), calculated using Geochemist’s Workbench version 7.0. a) Calculated using the Visual MINTEQ thermodynamic database release 2.40 with no minerals or species suppressed. b) As for a), but calculated using the LLNL V8 R6 "combined" dataset, thermo.com.V8.R6.full and with activity of Se, [Se] = 10-8. Note that the only Mo-solids in the database used in b) are Mo and MoSe2; if the latter is suppressed than no minerals are calculated to be stable.

When the database “thermo.com.V8.R6.full” is used the only solids predicted to be capable of forming under conditions similar to those on the BC porewaters (mean Eh = -341 mV and mean pH = 8.31; De Craen et al., 2004) are native Mo metal and MoSe2. However, these solids are not known to occur naturally and therefore are not expected to form in the BC. Nevertheless, Mo has been reported to occur naturally in drysdallite (Mo(S,Se)2), in the oxidation zone of a uranium deposit in a talc schist from Gambia (Anthony et al., 2003). This

SCK•CEN/12201513 Page 56 of 208 Compilation of Technical Notes on less studied elements

illustrates that the concentration of Se in solution may have a bearing on the control of Mo solubility, although there are no thermodynamic data available with which to judge the feasibility of this potential control.

2.2.1.3 Sorption and retardation As explained previously, in aqueous solution under conditions relevant to the BC, Mo is 2- expected to be present almost exclusively as MoO4 . Oxo-anions of this kind are likely to show only weak sorption on mineral surfaces. In recognition of this behaviour, in compiling sorption data for use in SKB’s SR-Site safety assessment, Crawford (2010) recommended a Kd of 0 m3/kg. However, Wersin et al. (2014) reviewed literature that showed variable sorption to clay minerals under the saline groundwater conditions present at the proposed site of a deep geological repository in Olkiluoto in Finland. Generally, the sorption seems to decrease with increasing pH, as is expected for a negatively-charged aqueous species given that surface charges will become less positive with increasing pH. Values for Kd that were recommended by Wersin et al. (2014) for use in Posiva’s TURVA-2012 safety assessment are given in Table 6.

Table 6: Molybdenum (Mo): Kd values recommended values and upper and lower limits for reference and bounding porewaters (m3/kg). From Wersin et al. (2014).

Dilute, Saline Brackish carbonate Brine High Glacial water water rich water alkaline melt water KR20/465/1 KR6/135/8 water KR4/861/1 water KR4/81/1

K d recommended 3 3 3 3 1 1 upper limit 20 20 20 20 20 20 lower limit 0.3 0.3 0.3 0.3 0.3 0.3

2.2.1.4 Transport and diffusion For the bentonite buffer and bentonite-bearing backfill proposed for use in the deep geological repository for SF that is being planned by Posiva for construction at Olkiluoto in Finland, Wersin (2014) proposed that the Deff values obtained for HTO could be used conservatively to model Mo migration. Using this approach for bentonite-bearing backfill at - 25 °C they proposed a best estimate value of the effective diffusion coefficient, Deff, of 9 x 10 11 2 -10 2 m /s and a lower value of Deff of 2 x 10 m /s. A corresponding effective porosity of 0.38 was specified. They pointed out that diffusion coefficients would increase with increasing temperature and proposed the following relationship to relate these parameters:

. (± . ) 2 ( ) = (0 ) , where Deff, is in m /s 𝑜𝑜 𝑜𝑜 0 026 0 03 𝑇𝑇 𝑒𝑒𝑒𝑒𝑒𝑒 𝑒𝑒𝑒𝑒𝑒𝑒 More directly relevant𝐷𝐷 𝑇𝑇to 𝐶𝐶the BC𝐷𝐷 conditions𝐶𝐶 𝑥𝑥 𝑒𝑒are measurements of HTO in diffusion coefficients in mudrocks. Wenk et al. (2008) summarize data from the Callovo-Oxfordian at Bure in France, the Opalinus Clay at Mont Terri in Switzerland, and the Opalinus Clay at Benken in Switzerland. These results are given in Table 7.

SCK•CEN/12201513 Page 57 of 208 Compilation of Technical Notes on less studied elements

Table 7: Measured diffusion coefficients for HTO in mudrocks at three different localities (m2/s). From Wenk et al. (2008).

Callovo- Opalinus Clay Opalinus Clay Oxfordian at Benken at Mont Terri at Bure Effective diffusion coefficient for 2.1 x E-11 5.4xE-12 1.4 x E-11 HTO normal to bedding (m2/s) Effective diffusion coefficient for 3.3 x E-11 3.2 xE-11 5.4 x E-11 HTO parallel to bedding (m2/s)

2.2.1.5 Justification Despite the uncertain reliability of the thermodynamic data for Mo, the latter was predicted by the speciation calculations using different databases to occur as oxo-anion under the BC reference conditions. Sorption was found to be only weak under different groundwater conditions. Furthermore, sorption seemed to decrease with increasing pH, which can be judged as an additional support to consider Mo as a negatively-charged aqueous species (given that surface charges will become less positive with increasing pH  repulsion). Diffusion coefficients for Mo under BC relevant conditions have not been determined up to now and are thus not available.

2.2.1.6 References Anthony J.W., Bideaux R.A., Bladh KW, and Nichols M.C. (Eds.) (1990-2003) Handbook of Mineralogy. Mineralogical Society of America, Chantilly, VA 20151-1110, USA. http://www.handbookofmineralogy.org/.

Bethke C.M. (1996) Geochemical Reaction Modeling, Concepts and Applications. Oxford University Press, 397 pp.

Bethke C.M. (2008) Geochemical and Biogeochemical Reaction Modeling. Cambridge University Press, 547 pp.

CRC (2011) Handbook of Chemistry and Physics, CRC Press, 92nd Edition.

De Vos W. and Tarvainen, T. (eds.) et al (2006) Geochemical Atlas of Europe. Part 2 - Interpretation of geochemical maps, additional tables, figures, maps, and related publications. Geological Survey of Finland, Otamedia Oy, Espoo, 692 pp.

Duro L., Cera E., Grivé M., Domènech C., Gaona X. and Bruno J. (2006) Development of the ThermoChimie thermodynamic database. Janvier 2006, Prepared by Enviros Spain S. L. National Radiactive Waste Management Agency (ANDRA) report C.RP.0ENQ.06.0001, Châtenay-Malabry cedex, France.

Duro L., Montoya V., Colàs E. and García D. (2010) Groundwater equilibration and radionuclide solubility calculations. Nuclear Waste Management Organisation (NWMO) Report NWMO TR-2010-02.

Grivé M., García D., Campos I. and Colàs E. (2014) Release of ThermoChimie. Version 7c: Track changes document Project ANDRA-TDB8. ANDRA Report CCRPFSTRI400I0.

Salminen R. (chief ed.) et al. (2005) Geochemical atlas of Europe. Part 1 - Background information, methodology and maps. Geological Survey of Finland, Otamedia Oy, Espoo, 525 pp. SCK•CEN/12201513 Page 58 of 208 Compilation of Technical Notes on less studied elements

Smedley P.L., Cooper D.M., Ander E.L., Milne C.J. and Lapworth D.J. (2014) Occurrence of molybdenum in British surface water and groundwater: Distributions, controls and implications for water supply. Applied Geochemistry, 40, 144–154.

United States Department of Energy (USDOE) (2007) In-Drift Precipitates/Salts Model. ANL-EBS-MD- 000045 REV 03.

Wenk H.-R., Voltolini M., Mazurek M., Van Loon L.R. and Vinsot A. (2008) Preferred orientations and anisotropy in shales: Callovo-Oxfordian Shale (France) and Opalinus Clay (Switzerland). Clays and Clay Minerals, 56, 285–306.

Wersin P., Kiczka M., Rosch D., Ochs M. and Trudel D. (2014) Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto Radionuclide Solubility Limits and Migration Parameters for the Backfill. Posiva Report POSIVA 2012-40.

World Health Organisation (WHO) (2011) Molybdenum in Drinking-water. Background document for development of WHO Guidelines for Drinking-water Quality. WHO Report WHO/SDE/WSH/03.04/11/Rev/1.

SCK•CEN/12201513 Page 59 of 208 Compilation of Technical Notes on less studied elements

3 GROUP III: Elements characterized by cation exchange sorption

3.1 Subgroup IIIa: Monovalent cations (alkali metals)

3.1.1 Technical note for Rubidium (Rb)

3.1.1.1 General Rubidium belongs to the alkali metal group and in total 26 isotopes of rubidium (atomic number: 37) are known. Naturally occurring rubidium is made of two isotopes, i.e. 85Rb (abundance ~72%) and 87Rb (abundance 28%). The latter is radioactive and has a half-life of 4.9 × 1010 years. It decays to stable 87Sr by emission of a negative beta particle. Rubidium metal has a high reactivity towards oxidation leading to the subsequent formation of the rubidium cation Rb+, which is very stable, and is normally unreactive towards further oxidative or reductive chemical reactions. Rubidium, like sodium and potassium, is almost always in its +1 oxidation state when dissolved in water.

3.1.1.2 Speciation and solubility As can be seen in Figure 22, the speciation of Rb is very simple and the uncomplexed Rb+ cation is the dominant species over the entire Eh and pH range. There is no known solubility- limiting phase for Rb.

Rb+ Eh (volts) Eh 0

–.5

25°C 0 2 4 6 8 10 12 14 pH Figure 22: Eh-pH diagram of rubidium (Rb-C-S-O-H) for the BC reference porewater system. Assumed activity of dissolved [Rb] = 10-8. Diagram a) MOLDATA/NEA TDB. Code: The Geochemist's Workbench - 8.08.

3.1.1.3 Sorption and retardation No experimental sorption data are available for Rb+ on Boom Clay.

Sorption of Rb+ occurs via ion-exchange. For sediments and soils, the selectivity of cations generally follows the lyotropic series. Cations with the same charge are more strongly sorbed when their hydration number is smaller, i.e. when the hydration shell of water molecules is

SCK•CEN/12201513 Page 60 of 208 Compilation of Technical Notes on less studied elements

smaller (Appelo and Postma, 2005). In practice, this means that the sorption of alkali metal cations follows the exchange affinity series (low to higher affinity): Li < Na < K < Rb < Cs. Given the position between K and Cs, sorption of Rb at trace concentrations is thus considered to be mainly occurring on high-affinity sites located at the edges of illite clay minerals.

Bradbury and Baeyens (2000) proposed a 3-site cation exchange model to describe the concentration dependent uptake of Cs on natural argillaceous rock systems, based on the premise that the sorption of Cs is dominated by the illite mineral component in the rock. This illite component is described by fixed site types and site capacities ("reference illite"). In their model, Rb was also incorporated because of the possible competition with Cs. The following set of reactions and model input parameters was put forward:

Table 8: Site types and distributions for "reference illite" (CEC = 0.2 eq/kg)

Site types Site capacities

Frayed edge sites 5×10-4 eq/kg

Type II sites 4×10-2 eq/kg

Planar sites 1.6×10-1 eq/kg

Table 9: Selectivity coefficients for the "reference illite" (CEC = 0.2 eq/kg)

Selectivity coefficients FES Type II sites Planar sites Log Cs K K c 4.6 1.5 0.5 LogK Kc 2.2 0.5 0.5 Log Cs K Na c 7.0 3.6 1.6 K Log Na Kc 2.4 2.1 1.1

The selectivity coefficients are given for the following reaction type:

K-illite + Cs ⇔ Cs-illite + K and are defined via a mass action equation according to the convention given in Gaines and Thomas (1953): N a N γ (K) Cs K = Cs K = Cs K K c N a N γ (Cs) K Cs K Cs where "a" terms represent activities, γ terms aqueous activity coefficients and the terms in parenthesis are molar concentrations. NCs and NK are equivalent fractional occupancies defined as:

SCK•CEN/12201513 Page 61 of 208 Compilation of Technical Notes on less studied elements

Γ N = Cs Cs Q

where ΓCs is the quantity of Cs sorbed (eq/kg) and Q is the sorption site capacity (eq/kg).

This model was used to successfully simulate Cs sorption in Boom Clay (Maes et al., 2008).

3.1.1.4 Transport and diffusion No in-house transport parameters are available for Rb.

3.1.1.5 Justification Attributing Rb to the group of cations seems straightforward. Although no in-house experimental data exist, it is well known that Rb is mainly sorbing via cation exchange, characterized however by a lower affinity than Cs due to the stronger hydration of Rb. This is also reflected by lower selectivity coefficients for Rb as compared to Cs on FES and Type II Log Rb K Cs K c LogK Kc sites ( < ) put forward by Bradbury and Baeyens (2000). As said above, no transport parameters for Rb have been determined up to now, but as for the other monovalent cations enhanced double layer diffusion is assumed to be the dominating playing process.

3.1.1.6 References Appelo, C.A.J., Postma, D. (2005) Geochemistry, groundwater and pollution. 2nd edition. A.A.Balkema Publishers, 649 pp.

Bradbury, M.H., Baeyens, B. (2000) A generalised sorption model for the concentration dependent uptake of caesium by argillaceous rocks, Journal of Contaminant Hydrology, 42, 141-163.

Gaines, G.I., Thomas, H.C. (1953) Adsorption studies on clay minerals: II. A formulation of the thermodynamics of exchange adsorption, Journal of Chemical Physics, 21, 714-718.

Li, Y.-H., Gregory, S. (1974) Diffusion of ions in sea water and deep sea sediments, Geochimica et Cosmochimica Acta, 38, 703-714.

Maes, N., Salah, S., Jacques, D., Aertsens, M., Van Gompel, M., De Cannière, P., Velitchkova, N. (2008) Retention of Cs in Boom Clay: Comparison of data from batch sorption tests and diffusion experiments on intact clay cores, Physics and Chemistry of the Earth, 33, S149-S155

Melkior, T., Yahiaoui, S., Motellier, S., Thoby, D., Tevissen, E. (2005) Cesium sorption and diffusion in Bure mudrock samples, Applied Clay Science, 29, 172-186.

Van Loon, L.R., Wersin, P., Soler, J.M., Eikenberg, J., Gimmi, Th., Hernan, P., Dewonck, S., Savoye, S. (2004) In-situ diffusion of HTO, 22Na+, Cs+ and I- in Opalinus Clay at the Mont Terri underground rock laboratory, Radiochimica Acta, 92, 757-763.

SCK•CEN/12201513 Page 62 of 208 Compilation of Technical Notes on less studied elements

3.2 Subgroup IIIb: Divalent cations (alkali earth metals)

3.2.1 Technical note for Calcium (Ca)

3.2.1.1 General Calcium belongs to the group of alkaline earth elements, has the atomic number 20 and is fifth-most-abundant element by mass in the Earth's crust, of which it forms more than 3%. Calcium is not naturally found in its elemental state. It occurs most commonly in sedimentary rocks in the minerals calcite, dolomite and gypsum. Ca is also comprised in igneous and metamorphic rocks, mainly in the silicate minerals, such as plagioclases, amphiboles, pyroxenes and garnets. Calcium from limestone is an important element in Portland cement. The solubility of the carbonate in water containing carbon dioxide causes the formation of caves with stalactites and stalagmites and is responsible for hardness in water (Handbook of Chemistry and Physics, 1992-1993). Calcium has four stable isotopes, i.e. 40Ca, 42Ca, 43Ca and 44Ca. Ninety-seven percent of naturally occurring Ca is 40Ca. 46Ca and 48Ca have such long half-lives that they are considered also stable. 41Ca is a cosmogenic isotope with a half-life of 103,000 years.

3.2.1.2 Speciation and solubility The Eh-pH diagram for calcium is shown in Figure 23 below. In the equilibrium model for the BC porewater chemistry, calcite represents the solubility controlling mineral for the calcium concentration in the porewater. Therefore, the cross symbol representing the BC reference conditions plots exactly on the boundary line between Ca2+ and calcite. The solubility of the latter corresponds to 5.02 × 10-5 under BC conditions.

The reaction constant of calcite comprised in MOLDATA is:

+ 2+ - CaCO3calcite + H ↔ Ca + HCO3 log K = 1.84

1 1

.5 .5 ++ Ca++ Ca Eh (volts) 0 Eh (volts) 0 Calcite CaCO3(aq) µ µ

–.5 –.5 25°C 25°C 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 pH pH Figure 23: Eh-pH diagram of calcium (Ca-C-S-O-H) for the BC reference porewater system. Activity of dissolved [Ca] = 2.44 × 10-5. Left diagram: aqueous speciation, right diagram: solid phases included in calculation. Database: MOLDATA_R2. Code: The Geochemist's Workbench- 10.0

SCK•CEN/12201513 Page 63 of 208 Compilation of Technical Notes on less studied elements

3.2.1.3 Sorption and Retardation Sorption of Ca on clays occurs predominantly through ion-exchange, hence the CEC can be considered as a ruler for Ca sorption. The CEC is correlated to the smectite content as is shown by Honty et al. (2010). No sorption experiments with Ca were performed on Boom Clay. However, sorption data for strontium on Boom Clay are available.

Since Sr and Ca can be considered as analogues for sorption because in the ion-exchange 2+ 2+ 2+ 2+ 2+ 2+ 2+ + + + + + affinity series (Ra > Ba > Sr > Ca > Ni > Cu > Mg > Ag > Cs > Rb > K > NH4 > Na+, (Helfferich 1962), Ca and Sr are at the same level (Sr having slightly higher affinity). Baeyens (1982) and Bradbury & Baeyens (2005) reported similar selectivity coefficients for Ca and Sr, therefore the data for Sr are considered as being representative for Ca and we refer for more details to the Topical Report on Sr by Maes et al. (2012, 2015).

Sr sorption can be described by a straightforward cation-exchange formalism (e.g. 2 SPNE SC/CE model of B&B, 1997) accounting for the presence of other alkaline and alkaline earth elements via selectivity coefficients. For trace level concentrations of Sr (<10-7 mol/L) in the Boom Clay pore water the logKd is ~2.5.

3.2.1.4 Transport and Diffusion The migration behaviour of Ca was studied by means of electromigration (Figure 24) (see for full description Maes et al. (2012, 2015) and ECOCLAY (2005).

1400 DC power supply Cathode Anode Anode 226 2+ Cathode - + Ra 1200

source position 1000

Porous clay core ceramic filter 800 14 + CH3NH3 600 Electromigration cell Cathode compartment Anode compartment 131I- 400 137Cs+ Bulk Activity [cps/g] Activity Bulk 85 2+ 200 Sr

Peristaltic Acid-Base Peristaltic 0 pump neutralization pump reservoir -40 -30 -20 -10 0 10 20 30 40 Distance from source, x [10-3m]

Figure 24: Left: Schematic representation of the electromigration set-up. Right: Typical electromigration dispersion profiles in the clay obtained for radionuclides with different charges.

In case of Ca, 6 experiments were conducted at increasing electrical fields, but in parallel 2 experiments were conducted withouth electrical field (= pure diffusive experiment) as such a wide span in velocities was covered.

SCK•CEN/12201513 Page 64 of 208 Compilation of Technical Notes on less studied elements

Table 10: The results of electromigration experiments with 45Ca2+ at different electrical fields

i Experiment E Vapp Dapp [V/m] [10-8 m/s] [10-12 m²/s] DIF-Ca1 0 0 8.3 DIF-Ca2 0 0 6.6 EM-Ca9 21 3.2 11.1 EM-Ca7 44 8.4 9.4 EM-Ca1 74 7.1 22.2 EM-Ca6 85 7.8 8.9 EM-Ca8 119 12.7 22.4 EM-Ca2 153 16.5 20.3 -5 -5 Linear Slope αD [m] 8.4 10 (s.d.= 3.3 10 ) -12 -12 Regression Intercept Dapp 7.8 10 (s.d.=2.9 10 )

i The linear relationship between Dapp and Vapp thus becomes:

i -12 -5 Dapp =7.8×10 +8.4×10 ×Vapp (r²=0.528)

The value obtained from the linear relationship coincides with the values at 0 velocity.

Figure 25: Results from the Ca electromigration experiments on Boom Clay conductedi at different electrical fields DDapp= app +α D V app resulting in different Vapp. The line represents the linear relationship .

A similar series of experiments was conducted with Sr2+, which can be considered as an 2+ -12 -12 analogue, and similar results were obtained: Dapp(Sr ) = 7.8×10 (s.d.=0.9 10 ) m²/s.

This corroborates the statement that sorption of Ca can be considered analogues to Sr as discussed in previous paragraph.

SCK•CEN/12201513 Page 65 of 208 Compilation of Technical Notes on less studied elements

The phenomenological description of Ca (and other cationic species) transport in clays poses a problem. In the classical theory on transport through porous media, the following relationship is used, which links the sorption (via R) to the diffusion coefficient: DDD δ D =pore =0 = eff app RRτη2 R

It has been observed for cations diffusing in (micaceous and/or smectite) clays, and especially ion-exchangeable cations, that this relationship might not be correct.

Dpore values calculated on basis of independently measured Dapp and R (Kd) values lead to values which exceed the Dpore value for HTO, which seems irrational. This apparent enhanced diffusion is often referred to as "surface diffusion" and although it is systematically observed, it is still matter of debate. The exact nature of the mechanism is not well understood but is linked to the pore geometry and double layer (DL) phenomena. It is also not clear if it applies only to cationic species that interact with clays via a cation exchange mechanism or also to (inner-and outer-sphere) surface complexes.

With respect to predictive modelling (for changing chemical conditions), this poses a problem as chemical coupled transport models are based on the above classical relation between diffusion and sorption. For Ca, the chemical model is well established but the diffusion model has some flaws. However, transport calculations can be based on experimentally determined and robust Dapp values. Models requiring separate input for Dpore and R instead of Dapp might be used by fixing one value and varying the other ensuring that they combine to the known Dapp value.

3.2.1.5 Justification To associate Ca2+ with the cation group is somehow self-explanatory, despite the fact that no in-house sorption data exist. As reported above, Sr can be considered as an analogue for Ca concerning sorption, as they have a similar ion exchange affinity (Sr slightly higher than Ca).

2+ Therefore, it is considered that the log Kd determined for Sr , i.e. ~2.5 (in suspension and compacted clay) applies also for Ca2+. Same apparent diffusion coefficients determined via electromigration experiments corroborate the similar sorption and transport behaviour for Ca and Sr.

3.2.1.6 References Baeyens B. (1982) Strontium, cesium and europium retention in Boom Clay: a potential repository site for nuclear waste. PhD Thesis, Katholieke Universiteit Leuven, Belgium.

Bradbury M.H and Baeyens B (1997) A mechanistic description of Ni and Zn sorption on Na montmorillonite. Part II: Modelling. Journal of Contaminant Hydrology, 27, 223-248.

Bradbury M., Baeyens B. (2005) Experimental and Modelling investigation on Na-illite: Acid-base behaviour and the sorption of strontium, nickel, europium and uranyl. PSI Bericht 05-02/NAGRA NTB 04-02.

SCK•CEN/12201513 Page 66 of 208 Compilation of Technical Notes on less studied elements

ECOCLAY II (2005) ECOCLAY II: Effects of Cement on Clay Barrier Performance Phase II. Final Report EUR21921, EC, Luxembourg.

Helferrich F. (1962) Ion Exchange. McGraw-Hill, New York.

Honty M. (2010) CEC of the Boom Clay – a review. SCK•CEN-ER-134, SCK•CEN, Mol, Belgium.

Maes N., Salah S., Bruggeman C., Aertsens M., Martens E., Van Laer Liesbeth (2012) Strontium retention and migration behaviour in Boom Clay-Topical report FFD, SCK•CEN-ER-197.

Maes N., Salah S., Bruggeman C., Aertsens M., Martens E., Van Laer Liesbeth (2015) Strontium retention and migration behaviour in Boom Clay-Topical report, Revised version of the report ER-197, in preparation.

SCK•CEN/12201513 Page 67 of 208 Compilation of Technical Notes on less studied elements

3.2.2 Technical note for Radium (Ra)

3.2.2.1 General Radium belongs to the group of earth alkaline elements and its atomic number is 88. Radium is a naturally occurring radionuclide that is part of the uranium and thorium decay series and 4 isotopes, i.e. 228Ra, 226Ra, 224Ra and 223Ra are known to exist in nature, which are all radioactive. In addition to their own radiological properties, three of these radium isotopes present additional environmental and health concerns due to the fact that they decay to 226 radon. The radium isotope considered in performance assessment calculations is Ra (t1/2= 3 230 4 1.6 × 10 years), representing the daughter product of Th (t1/2 = 7.52 × 10 years), which itself is part of the 238U (4n + 2) decay chain/series. The other radium isotopes, such as 228Ra 223 224 (t1/2 = 5.75 years), Ra (t1/2 = 11.68 d) and Ra (t1/2 = 3.64 d) are irrelevant with respect to long-term waste management considerations as they represent short-lived isotopes. The chemical properties of radium are very similar to those of strontium. Radium dissolves as Ra2+ + + + and may form following complexes RaOH , RaCl , RaCO3(aq), RaHCO3 , and RaSO4(aq).

3.2.2.2 Speciation and solubility Aqueous speciation under the BC reference conditions is dominated by the divalent radium cation (Figure 26). At higher pH values (> 9.5), the free Ra cation is replaced by the RaCO3(aq) complex. At [Ra] = 10-8, radium is not solubility limited and only at higher Ra concentrations, radium concentrations might be controlled by either Ra carbonate or sulphate (Table 11). Considering the possibility of co-precipitation of Ra or its incorporation into a solid solution, the above mentioned concentration threshold could be lower as these phases are known to reduce the solubility of the pure endmembers.

Ra++

Eh (volts) 0 RaCO3(aq) µ

–.5 25°C 0 2 4 6 8 10 12 14 pH

Figure 26: Eh-pH diagram of radium (Ra-C-S-O-H) for the BC reference porewater system. Assumed activity of dissolved [Ra] = 10-8. Database: MOLDATA_R2. Code: The Geochemist's Workbench - 10.0

SCK•CEN/12201513 Page 68 of 208 Compilation of Technical Notes on less studied elements

Table 11: Solubility of Ra in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV. Database: MOLDATA_R2. Code: The Geochemist's Workbench- 10.0

Solubility controlling phases Solubility, [Ra], mol/L

-5 RaCO3(s) 6.7 × 10 -6 RaSO4(s) 7.0 × 10

Source data: ANDRA TDB

The reaction constants for the Ra-solids comprised in Table 11 and MOLDATA are the following:

+ 2+ - RaCO3(s) + H ↔ Ra + HCO3 log K = 2.03 2+ 2- RaSO4(s) ↔ Ra + SO4 log K = -10.26

Table 12: Species distribution of Ra in equilibrium with RaSO4(s). Database: MOLDATA_R2. Code: The Geochemist's Workbench-10.0

Aqueous species [mol/L] Percentage [%] Ra2+ 6.6 × 10-6 94.1 + -7 RaHCO3 2.2 × 10 3.1 -7 RaCO3(aq) 1.7 × 10 2.4 -8 RaSO4(aq) 3.1 × 10 0.4

Radium carbonate is considered to represent the most relevant solubility controlling solid for Ra in the far-field.

Role of solid solutions comprising Ra

1) Ra-barite solid solution [(Ba,Ra)SO4])

Most (pure) radium salts are insoluble, especially the sulphate (RaSO4,s) and carbonate (RaCO3,s) salts. According to Langmuir and Riese (1985), Ra concentrations in natural waters, and waters associated with uranium mining and nuclear waste disposal are however rarely high enough to reach saturation with a pure radium solid. Therefore, maximum radium concentrations are rather limited by adsorption or solid solution formation. The latter process has been extensively studied in the last years by many authors and it is well known that co- precipitation of Ra isotopes with barite is a phenomenon that should be taken into account when discussing the solubility limit of this radionuclide. The ionic radius of radium (1.43 Å) is very similar to the one of barium (1.36 Å) due to which it can be accommodated in the barite lattice. Besides this, barite commonly contains minor amounts of strontium (Sr) and lead (Pb). Different studies have shown that the solubility limit for radium in many natural and anthropogenic environments is controlled by barite precipitation. Thus, barite represents an excellent scavenger for this radionuclide. Berner and Curti (2002) calculated the retention of

SCK•CEN/12201513 Page 69 of 208 Compilation of Technical Notes on less studied elements

radium in a nuclear waste repository in Opalinus Clay as host rock by using the Gibbs Energy Minimization (GEM) approach. Their results showed that the the total radium concentration in -8 the porewater was 4.8 × 10 M, if controlled by the pure phase RaSO4(s), while the solubility limit was lowered by few orders of magnitude (8.2 × 10-12) when considering co-precipitation of radium with barite.

Kulik et al. (2004) showed that the radium concentration in solution is dependent on the Ba/Ra activity ratio in the precipitated solid solution. The higher the ratio, the more the solubility is decreased compared to the pure phase. It should be mentioned that the sulphate concentration in undisturbed BC porewater is quite low, i.e. 0.02 mmol/L. Barium concentrations have been determined by De Craen et al. (2004) to range between 10-60 µg/L -8 -7 (7.3×10 - 4.4×10 mol/L). A significant decrease in Ra solubility compared to pure RaSO4(s) can be expected only at Ba/Ra activity ratios >100, which seems however not to be achievable under undisturbed BC conditions. Thus, for trace concentrations of Ra, the "host mineral" solid solution can be considered and must be close to saturation with respect to the groundwater in order to limit the Ra solubility. It should be bared in mind that sulphate concentrations may also be affected by microbial activity/perturbation. The presence of 2- sulphate reducing bacteria could diminish the SO4 concentration in solution, leading to the dissolution of the Ra solid phase (either solid solution or pure phase) and to a remobilization/enhanced mobility of radium.

2) Ra-witherite solid solution [(Ba,Ra)CO3])

Also the formation of a binary solid solution between the two end-members BaCO3 (witherite) and RaCO3 should be considered as a potential process controlling Ra solubility. As already mentioned above, the ionic radii of Ra2+ and Ba2+ are similar due to which Ra can be accommodated also in the witherite lattice. Ra can also be incorporated in strontianite and form a (Sr,Ra)CO3 solid solution.

3) Ra-gypsum solid solution [(Ca,Ra)SO4]x2H2O)

Recently Lestini et al. (2013) studied the uptake/incorporation of radium by gypsum via batch experiments. In previously performed studies, two different distribution coefficients (molar ratio between 226Ra and Ca in solid over molar ratio of 226Ra/Ca in solution) for the interaction of 226Ra and gypsum were determined, i.e. 0.3 (Yoshida et al., 2009) and 0.03 (Gnanapragasam and Lewis, 1995). The purpose of the incorporation experiments performed by Lestini et al. (2013) was to re-examine these distribution coefficients under equilibrium conditions. Results indicated that the 226Ra-activity in solution in contact with gypsum showed no evolution, i.e. decrease over the experimental duration (200 days). This lack of interaction of 226Ra with gypsum was mainly referred to the (significant) difference in the ionic radii of Ra (1.48 Å) and Ca (1.12 Å) with Ra being too big to enter Ca’s site in gypsum. Besides this, the high solubility product of gypsum log Ks = -4.58, as compared to barite (logKs = -9.97) was put forward as another inhibiting factor for the incorporation of 226Ra in gypsum. The general conclusion of the authors was that a solid solution between gypsum and radium sulfate cannot be considered per se and that other sulfate phases, such as celestite (SrSO4) and anglesite

SCK•CEN/12201513 Page 70 of 208 Compilation of Technical Notes on less studied elements

(PbSO4), with intermediate solubility products between gypsum and barite could be more favourable to incorporation of 226Ra.

Besides the formation of so-called binary solid solutions, also ternary systems exist (e.g.: [Ba,Sr,Ra]CO3 or [Ba,Sr,Ra]SO4). For further details concerning the formation of the above described sulphate and carbonate solid solutions, it is referred to Kulik and Curti (2005) and Bruno et al. (2007).

3.2.2.3 Sorption and Retardation No sorption data for Ra on Boom Clay are available.

Sorption is considered to occur via ion-exchange (IE). Hence the CEC can be considered as a ruler for Ra sorption. The CEC is correlated to the smectite content as is shown by Honty et al. (2010). Following the exchange affinity series, Ra has a higher affinity than Sr/Ca and should therefore exhibit higher sorption. Due to the lack of IE selectivity coefficients, no geochemical calculations for the sorption could be made.

Recently, Robinet et al. (2013) investigated the influence of organic ligands on the retention of cationic species on Callovo-Oxfordian Clay (mainly an illitic clay). One of the cations investigated was Ra.

In a system without organic ligands added, Kd-values were ~100 L/kg. In presence of organics, the Kd decreased about 10-40%.

Savoye (Altman et al., 2014) studied sorption of Sr on Callovo-Oxfordian Clay and reported Kd values between 4-6 L/kg which is in line with the expectations based on the exchange affinity series.

3.2.2.4 Transport and Diffusion The migration behaviour of Ra was studied by means of electromigration (Figure 27) (see for full description Maes et al., 2001).

1400 DC power supply Cathode Anode Anode 226 2+ Cathode - + Ra 1200

source position 1000

Porous clay core ceramic filter 800 14 + CH3NH3 600 Electromigration cell Cathode compartment Anode compartment 131I- 400 137Cs+ Bulk Activity [cps/g] Activity Bulk 85 2+ 200 Sr

Peristaltic Acid-Base Peristaltic 0 pump neutralization pump reservoir -40 -30 -20 -10 0 10 20 30 40 Distance from source, x [10-3m]

Figure 27: Left: Schematic representation of the electromigration set-up. Right: Typical electromigration dispersion profiles in the clay obtained for radionuclides with different charges.

SCK•CEN/12201513 Page 71 of 208 Compilation of Technical Notes on less studied elements

In case of Ra, 5 experiments were conducted at increasing electrical fields. EM-Ra6 is as an outlier (marked in italic in Table 13 and open symbols in Figure 28).

Table 13: The results of electromigration experiments with 226Ra2+ at different electrical fields

i Experiment E Vapp Dapp [V/m] [10-8 m/s] [10-12 m²/s] EM-Ra5 62 0.7 1.2 EM-Ra4 72 1.0 1.7 EM-Ra6 102 1.2 3.0 EM-Ra2 105 1.2 1.2 EM-Ra1 124 1.4 1.7 -5 -5 Linear Regression Slope αD [m] 5 10 (s.d.= 5 10 ) -13 -13 Intercept Dapp 9.1 10 (s.d.= 6 10 )

i i -13 -5 The linear relationship between Dapp and Vapp becomes: Dapp = 9.10×10 + 5.08×10 *Vapp (r=0.55).

In line with the ion-exchange affinity series, Ra retention should be higher than for Sr/Ca, which should result in a lower Dapp which is observed by these experiments.

-10 -10 1 As the D0 values for Ra (8.90×10 m²/s) and for Ca/Sr (respectively 8.08×10 m²/s and 7.90×10-10 m²/s) are similar, and all three cations are predominantly retained by ion exchange processes, we consider that the difference in Dapp values is solely due to a difference in retardation (R) value.

3,5E-12

3,0E-12 [m²/s] a i 2,5E-12 y = 5,08E-05x + 9,10E-13 r2 = 2,99E-01 2,0E-12

1,5E-12

1,0E-12

Apparent dispersion coefficient, D coefficient, dispersion Apparent 5,0E-13 y = 1,54E-04x + 1,19E-13 r2 = 9,95E-01

0,0E+00 0,0E+00 2,0E-09 4,0E-09 6,0E-09 8,0E-09 1,0E-08 1,2E-08 1,4E-08 1,6E-08 Apparent convection velocity, Va [m/s]

Figure 28: The linear relationship between the apparent convection velocity and the apparent dispersion coefficient gives the apparent diffusion coefficient (intercept). The squares and circles are respectively for 226Ra2+ and 137Cs+. Open symbols denote outliers.

The phenomenological description of Ra (and other cationic species) transport in clays poses a problem. In the classical theory on transport through porous media, following relationship is used which links the sorption (via R) to the diffusion coefficient:

SCK•CEN/12201513 Page 72 of 208 Compilation of Technical Notes on less studied elements

DDD δ D =pore =0 = eff app RRτη2 R

It has been observed for cations diffusing in (micaceous and/or smectite) clays, and especially ion-exchangeable cations, that this relationship might not be correct.

Dpore values calculated on basis of independent measured Dapp and R (Kd) values lead to values which exceed the Dpore value for HTO, which seems irrational. This apparent enhanced diffusion is often referred to as "surface diffusion" and although it is systematically observed, it is still matter of debate. The exact nature of the mechanism is not well understood but is linked to pore geometry and double layer (DL) phenomena. It is also not clear if it applies only to cationic species that interact with clays via a cation exchange mechanism or also to (inner-and outer-sphere) surface complexes.

However, transport calculations can be based on experimentally determined and robust Dapp values. Models requiring separate input for Dpore and R instead of Dapp might be used by fixing one value and varying the other ensuring that they combine to the known Dapp value.

3.2.2.5 Justification As for calcium, the reasoning to associate Ra to the subgroup IIIb, is quite straightforward as cation exchange represents the main retention process with Ra even showing a higher affinity than Sr and Ca. This is also reflected in the lower apparent diffusion coefficient (Dapp around 10 times lower than for Sr/Ca) observed in the electromigration experiments performed with 226Ra. Other derived transport data revealed that also for Ra the phenomenon of “surface diffusion” was observed supporting the grouping.

3.2.2.6 References Altmann S., Van Laer L., Montoya V., Kupcik T., Tournassat C., Glaus M., Schaefer T., Bruggeman C., Maes N., Aertsens M., Frick, Van Loon L., Gaboreau S., Robinet J.-C., Savoye S., Appelo T. Processes of cation migration in clayrocks (CatClay) - Final Scientific Report of the Catclay European Project.2015. .

Maes N., Moors H., Dierckx A., Aertsens M., Wang L., De Cannière P., Put M. (2001) Studying the migration of radionuclides in Boom Clay by electromigration. In: EREM2001 3rd symposium and status report on electrokinetic remediation (Czurda C., Haus R., Kappeler C., Zorn R., eds.). Schriftenreihe Angewandte Geologie Karlsruhe 63, 35/1-35.

Honty M. (2010) CEC of the Boom Clay – a review. SCK•CEN-ER-134, SCK•CEN, Mol, Belgium.

Robinet J.-C., Tournassat C., Dagnelie R., Landesman C., Gehin A. (2013) Caractérisation de la rétention de cations représentatifs sur l'argilite du Callovo-Oxfordien en présence de complexants organiques. Rapport ANDRA CG.NT.ASTR.13.0019A.

SCK•CEN/12201513 Page 73 of 208 Compilation of Technical Notes on less studied elements

4 GROUP IV: Elements characterized by DOM linked transport

4.1 Subgroup IVa: Transition metals (+ Be and Sn)

4.1.1 Technical note for Silver (Ag)

4.1.1.1 General Silver (Ag) is a metallic element (Group 11) that has an atomic number of 47. The most common oxidation number for silver is 1 (Winter, 2014). Silver is very ductile and malleable, it has the highest electrical and thermal conductivity of all metals, and it possesses the lowest contact resistance (Winter, 2014). A number of silver compounds exist, including: fluorides, chloride, iodide, oxide, sulphides and selenide (Winter, 2014). In nature, silver is found as free metal and as sulphide minerals (Winter, 2014). Silver has a crustal abundance of 80 ppb, a seawater abundance of 0.1 ppb (Winter, 2014), and steam water typically has a silver abundance of 0.3 ppb (Drever, 1997). In dilute oxygenated groundwater, silver concentrations typically range between 9.26 and 92.6×10-11 mol/dm3 (UK data taken from Edmunds et al. 1989; cited by Duro et al., 2006). In several groundwater samples from Poços de Caldas (a site often considered as a natural analogue for radioactive waste disposal systems) the content of -6 3 silver can be as high as 0.5×10 mol/dm (Nordstrom et al. 1991).

4.1.1.2 Speciation and solubility In their review for the Swedish ‘SR-Can’ assessment, Duro et al. (2006) consider AgCl to be a candidate solubility-limiting solid phase for the reference groundwater compositions included in the assessment, with aqueous speciation being dominated by silver chloride complexes. A solubility of 4.4×10-6 mol/dm3 is given for the reference water compositions considered in the assessment.

With regard to controls on dissolved silver concentrations, native silver and AgCl have been considered as solubility-limiting phases for the Opalinus Clay (Berner, 2002) with a calculated silver concentration of 2.8×10-6 mol/dm3 (generated using the MINEQL database). In this - calculations, the dominant complexes in solution are silver chlorides (AgCl(aq), 6 %; AgCl2 ,60 2-, 3- %; AgCl3 18 %; AgCl4 , 16 %).

In the recent Finnish ‘TURVA-2012’ assessment for spent fuel disposal, the solubility of silver was considered to be controlled by AgCl(cr) (according to predictions made using the Andra/ThermoChimie database) assuming AgS does not form (Wersin et al., 2014a). It is noted by Wersin et al. (2014a) that the AgS would result in a much lower dissolved silver concentrations, but because of the uncertainties in the actual sulphide contents and the well- established solubility of AgCl, the possibility of AgS formation was conservatively ignored (Wersin et al., 2014a). The formation of native silver was considered difficult to defend due to slow reaction rates. Wersin et al., (2014a) note that in their calculations, aqueous speciation is dominated by chloride complexes, except in a reference glacial melt water composition, for which Ag+ predicted to be the dominant species (Wersin et al., 2014a). For solubility limits associated with waste canisters, reference silver concentrations of 9.9×10-6 and 5.1×10-6

SCK•CEN/12201513 Page 74 of 208 Compilation of Technical Notes on less studied elements

mol/dm3 were adopted for saline and brackish water compositions, respectively (Wersin et al., 2014a). An upper limit of 2.4×10-4 mol/dm3 was adopted for a reference brine composition.

Using MOLDATA, AgHS(aq) has been predicted to be the dominant silver species under Boom Clay conditions (Salah and Wang, 2014). If this species is suppressed in the calculations, AgCl(aq) is the prevalent species. Salah and Wang (2014) show that under Boom Clay conditions, native silver has the lowest solubility of a number of silver compounds (3.7×10-13 3 -13 3 - mol/dm ), followed by acanthite (Ag2S) (5.4×10 mol/dm ), chlorargyrite (AgCl(s)) (2.5×10 5 3 -4 3 mol/dm ) and Ag2CO3(s) (5.6×10 mol/dm ).

Using Geochemist’s Workbench® (Bethke, 2008) and the ‘MINTEQ’ (thermo_minteq.dat) database, solubility diagrams have been calculated for silver (assuming the same mineral- fluid equilbria as in the model of the Boom Clay porewater described by Salah and Wang, 2014; Figure 29). In agreement with the calculations presented by Salah and Wang (2014), AgHS(aq) is predicted to be the dominant aqueous species under conditions associated with the Boom Clay and native silver is predicted to be the most stable (least soluble) solid silver phase, followed by acanthite.

–2

–4

–6 Ag metal

+ –8

- -- –10 H2CO3* (aq) HCO3 CO3

log a Ag log –12

–14

–16 AgCl (aq) AgHS (aq) AgS- –18 16°C –20 0 2 4 6 8 10 12 14 pH

–2

–4

–6 Acanthite

+ –8

- -- –10 H2CO3* (aq) HCO3 CO3

log a Ag log –12

–14

–16 AgCl (aq) AgS- AgHS (aq) –18 16°C –20 0 2 4 6 8 10 12 14 pH

Figure 29: Solubility diagrams for silver, assuming Ca2+ activity buffered by calcite, sulphate buffered by pyrite, 2+ - Fe buffered by siderite, log a Cl = -3.155, log f CO2(g) = -2.44, log f O2(g) = -71.4 (Eh = -0.28 mV at pH 8.36, 16 °C, 1 bar). Diagrams generated using ‘Act2’ module of Geochemist’s Workbench® (Bethke, 2008), speciated for carbonate. Upper diagram includes all solids. Native silver is suppressed in the lower diagram.

SCK•CEN/12201513 Page 75 of 208 Compilation of Technical Notes on less studied elements

4.1.1.3 Sorption and retardation As part of a review undertaken of silver chemistry for the Opalinus Clay assessment, Bradbury and Baeyens (2003a) cite Pleysier and Cremers (1975) who undertook an extensive study on the cation exchange of a silver-thiourea complex onto Na-, Ca- and Al-montmorillonites. In this work, the exchange of silver on a Na montmorillonite (Camp Berteau) was measured at 25°C and a selectivity coefficient of 0.8 was reported. Bradbury and Baeyens (2003a) describe modelling of sorption of silver to MX-80 using the code ‘MINSORB’ and the data from -7 3 -1 Pleysier and Cremers (1975) which resulted in the calculation of a low Kd value of 10 m kg , primarily due to the high sodium background concentration and the strong tendency of silver to form chloro complexes (which are assumed to be non-sorbing). However, Bradbury and Baeyens (2003a) note that silver compounds tend to be photosensitive, and that in a gamma field, there could be reduction to native silver. With regard to the Opalinus Clay itself, Bradbury and Baeyens (2003b) note that data for silver sorption to minerals is rare and that the only data they found is that reported by Legoux et al. (1992). However, the values measured were considered to be very high and Bradbury and Baeyens (2003b) rejected them on the basis that they could reflect precipitation rather than sorption processes. Given that in the water compositions considered, silver would probably form anionic complexes with chloride, or would be present as a neutral species, it was assumed that silver would not sorb.

In a review of silver sorption to clays for the TURVA-2012 assessment, Hakanen et al. (2014) note the lack of available data and cite the study of Daniels and Rao (1983) who measured silver sorption on kaolinite in AgNO3 solutions. Hakanen et al. (2014) give an Rd value of -3 3 -3 4.24×10 m /kg (in a 5×10 M AgNO3 solution) and this is considered to be in agreement with the results obtained for by Khan et al. (1995), for silver sorption to bentonite where the -2 3 -9 3 Rd value was found to b 3.2×10 m /kg (silver concentration of 1×10 mol/dm at pH 6.5 and a bentonite cation exchange capacity of 77 meq/100g. For the TURVA-2012 assessment, Kd values for silver sorption to kaolinite and illite were generated for a number of different water compositions (differing degrees of silver complexation with chloride), ‘best estimate’ values for kaolinite range from 6.4×10-8 to 2×10-3 m3/kg, whereas values for illite range from 2.1×10- 7 to 6.7×10-3 m3/kg (Hakanen et al., 2014).

In a review of generic partition coefficients for metals in soils, surface water and wastes, Allison and Allison (2005) report a mean silver Kd for soil/soil water partitioning of 2.6 L/kg with a standard deviation of 0.8 and a range of 1 to 4.5 L/kg (n = 21). For sediment/porewater, a mean of 3.6 L/kg with a standard deviation of 1.1 and a range of 2.1 to 5.8 L/kg are given (Allison and Allison, 2005).

Praus et al. (2008) studied the sorption of Ag+ to Na-montmorillonite and found that a Langmuir model provided the best fit to their data. Furthermore, it is suggested that Ag+ forms a monolayer on clay surfaces, with part of the Ag+ being present in clay interlayers. The montmorillonite samples saturated with Ag+ were treated with the sodium borohydride solutions to produce native silver nanoparticles which were assumed to be located at the layer surfaces and crystallite edges.

Jacobson et al. (2005) measured sorption isotherms for trace levels of silver and thallium onto three illite-rich mineral soils from central New York and it was found that silver sorbed more strongly than thallium to all the soils. The ‘peaty-muck’ soil described by Jacobson et al. (2005) sorbed silver more strongly than the mineral soils included in the study, suggesting

SCK•CEN/12201513 Page 76 of 208 Compilation of Technical Notes on less studied elements

that silver sorption to soils is dominated by soil organic matter, either through ion exchange or complexation reactions.

4.1.1.4 Transport and diffusion -11 2 In the TURVA-2012 assessment, silver was assumed to have Deff and η values of 9×10 m /s and 0.38 for bentonite backfill, assuming behaviour similar to that of HTO (Wersin et al., 2014b).

-11 2 For the Opalinus Clay, silver was assigned ‘non anionic’ Deff values of: 1×10 m /s (reference value perpendicular to bedding); 1×10-10 m2/s (‘upper pessimistic’ value perpendicular to bedding); and 5×10-11 m2/s (reference value parallel to bedding). A value of 0.12 was assigned to η (Bradbury and Baeyens, 2003b).

4.1.1.5 Justification Silver represents a transition metal ″sensu strictu″, meaning a d-block element of the periodic table. As such, it seems logic to associate it to this group. Under the reference BC conditions, neutral AgHS(aq) species were calculated to represent the dominant complexes in BC porewater. It should be mentioned however that the speciation of Ag depends not only on the redox, but also on its concentration. Taking into account the uncertainty on the redox, silver chloride complexes (being neutral or negative) could also be encountered in the groundwater. Having this in mind, sorption is not expected to be high. Unfortunalely neither in-house sorption nor transport data are available for Ag, so that the reasoning is purely based on literature data. Sorption coefficients reported in the review part (4.1.1.3) are quite scattered, but generally low (highest logKd=1.5). This was mainly referred to the fact that Ag under the different studied groundwater conditions either occurred as neutral and/or negative species (see Figure 29). Assuming that silver may be present as free cation (Ag+) in natural environments (and thus associating it to the cation group) seems highly unlikely. Not only due to the speciation, but also due to the low solubility of silver, high sorption coefficients are not expected and should thus be considered with caution as they may reflect rather precipitation than sorption processes. From mechanistic point of view, cation exchange is reported for Ag+ sorption under lab conditions. In presence of organic matter, also surface complexation might play a role. As said in section 1.5, we do not have experimental evidence that transport of Ag might be organic matter facilitated. Getting more insight in the migration behaviour of Ag would therefore be useful.

4.1.1.6 References Allison, J.D. and Allison, T.L. (2005) Partition coefficients for metals in surface water, soil, and waste. United States Environmental Protection Agency, EPA/600/R-05/074.

Berner, U. (2002) Project Opalinus Clay: Radionuclide concentrations limits in the near-field of a repository for spent fuel and vitrified high-level waste. Nagra Technical Report NTB 02-10, Wettingen, Switzerland.

Bethke, C.M. (2008) Geochemical and Biogeochemical Reaction Modeling. Cambridge University Press.

Bradbury, M.H. and Baeyens, B. (2003a) Near-Field Sorption Data Bases for Compacted MX-80 Bentonite for Performance Assessment of a High-Level Radioactive Waste Repository in Opalinus Clay Host Rock. Nagra Technical Report 02-18. Wettingen, Switzerland.

SCK•CEN/12201513 Page 77 of 208 Compilation of Technical Notes on less studied elements

Bradbury, M. and Baeyens, B. (2003b) Far Field Sorption Data Bases for Performance Assessment of a High-level Radioactive Waste Repository in an Undisturbed Opalinus Clay Host Rock. PSI Report 03- 08/Nagra Report NTB-02-19. Wettingen, Switzerland.

Daniels, E. and Rao, S.M. (1983) Silver adsorption by kaolinite. International Journal of Applied Radiation and Isotopes 34, 981-984.

Drever, J. I. (1997) The geochemistry of natural waters: surface and groundwater environments (3rd edition). Prentice-Hall, London.

Duro, L., Grivé, M., Cera, E., Gaona, X., Domènech, C. and Bruno, J. (2006) Determination and assessment of the concentration limits to be used in SR-Can. SKB Technical Report, TR-06-32. Swedish Nuclear Fuel and Waste Management Company. Stockholm, Sweden.

Edmunds, W.M., Cook, J.M., Kinniburg, D.G., Miles, D.G, and Trafford. J.M. (1989) Trace-element occurrence in British groundwaters. British Geological Survey. Research Report SD/89/3. Nottingham, UK.

Hakanen, M., Ervanne, H., Puuko, E. (2014) Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto. Radionuclide Migration Parameters for the Geosphere. Posiva Report 2012-41. Posiva Oy, Olkiluoto, Finland.

Jacobson, A.R., McBride, M.B., Baveye, P. and Steenhuis, T.S. (2005) Environmental factors determining the trace-level sorption of silver and thallium to soils. Science of the Total Environment 345, 191-205.

Khan, S.A., Riaz-ur-Rehman, Khan, M.A. (1995) Adsorption of chromium (III), chromium (VI) and silver (I) on bentonite. Waste Management 15, 271-282.

Legoux, Y., Blain, G., Guillaumont, R., Ouzounizian, G., Brillard, L. and Hussonnois, M. (1992) Kd measurements of activation, fission and heavy elements in water/solid phase systems. Radiochimica Acta 58/59, 211-218.

Nordstrom, D.K., Smellie, J.A.T., Wolf, M. (1991) Chemical and isotopic composition of groundwaters and their seasonal variability at the Osamu Utsumi and Morro do Ferro analogue study sizes, Poços de Caldas, Brazil. SKB. Poços de Caldas Report 6.

Pleysier, J. and Cremers, A. (1975) Stability of silver-thiourea complexes in montmorillonite clay. Journal of the Chemical Society, Faraday Transactions 17, 1256-264.

Praus, P., Turicová, M. and Valáškova, M. (2008) Study of silver adsorption on montmorillonite. Journal of the Brazilian Chemical Society 19 (3), 549-556.

Salah, S. and Wang, L. (2014) Speciation and solubility calculations for waste relevant radionuclides in Boom Clay. First Full Draft. External Report SCK•CEN-ER-19814/Ssa/P-16.

Wersin, P., Kiczka, M., Rosch, Gruner, A.G. (2014a) Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto Radionuclide Solubility Limits and Migration Parameters for the Canister and Buffer. Posiva Report 2012-39. Posiva Oy, Olkiluoto, Finland.

Wersin, P., Kiczka, M., Rosch, D., Ochs, M. and Trudel, D. (2014b) Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto Radionuclide Solubility Limits and Migration Parameters for the Backfill. Posiva Report 2012-40. Posiva Oy, Olkiluoto, Finland.

Winter, M. (2014) Webelements. University of Sheffield and Webelements Ltd. http://www.webelements.com/silver

SCK•CEN/12201513 Page 78 of 208 Compilation of Technical Notes on less studied elements

4.1.2 Technical note for Beryllium (Be)

4.1.2.1 General Beryllium (Be) belongs to the group of alkaline earth metal group with the atomic number 4 and represents one of the most toxic elements in the periodic table. Beryllium has one stable isotope, i.e. 9Be and eleven known radioisotopes. Being one of the lightest known structural metals Be is used in a wide variety of both nuclear and non-nuclear applications. Be, due to its unique combination of structural, chemical and neutron absorption cross-section characteristics has been successfully used in nuclear reactors as neutron reflector and neutron moderator (i.e. reduces the energy of neutrons).

4.1.2.2 Speciation and solubility Beryllium's chemical behaviour is largely a result of its small atomic and ionic radii. It is likely more chemically similar to aluminium (having a similar charge-to-radius ratio) than to other alkaline earth metals. Beryllium exhibits only the +2 valence state in solution and the Be2+ ion has the smallest size (r = 0.31Å) of all the metal cations, and because of this, it has the unusually low hydration number of 4 (Baes and Mesmer, 1976). Be is relatively immobile in natural waters at neutral pH (Edmunds, 2011). Under acidic (pH < 4-5) and alkaline conditions (pH > 11-12), it is more soluble and due to its toxicity becomes an environmental problem when it is mobilized. Both, mono- and polynuclear hydrolysis products are rapidly and reversibly formed in acid solution (Baes and Mesmer, 1976). It should be mentioned that the former are dominating at metal concentrations < 10-3 molal, while the latter become only important beyond that concentration.

Hem (1985) reviewed the environmental behavior of Be and summarized the following: although nominally included in the alkaline-earth group, the element beryllium has little in its chemistry that is in common with the typical alkaline-earth metals. Beryllium ions are small enough to replace silicon in igneous-rock minerals. One of the more important of the minerals in which beryllium is an essential constituent is beryl, a silicate of aluminum and berryllium that is found most commonly in pegmatites. Other silicates or hydroxy-silicates may also be important sources of beryllium. Berylium is rather a rare element; its abundance in crustal rocks is similar to that of cesium. Beryllium sulfate and carbonate compounds appear to be too soluble to be important controls, but the oxide and hydroxides have very low solubilities (comment: not under BC conditions). Beryllium may form anionic fluoride complexes that could increase its aqueous mobility.

Thermodynamic data for Be are scarce. In MOLDATA_R2 only 6 species are comprised (i.e. 2 aqueous species, 1 gas, 3 minerals). It can be seen in Figure 30, that the speciation of Be 2- under BC conditions is dominated by the negatively charged BeO2 species (copied from LLNL TDB).

The only solid Be-phase comprised in MOLDATA_R2 is bromellite (BeO). The solubility of the latter phase under BC conditions was calculated to be very low, i.e. 8.4×10-15 M. The species 2- distribution in equilibrium with bromellite was calculated to correspond to 95% of BeO2 and 5% of Be2+.

SCK•CEN/12201513 Page 79 of 208 Compilation of Technical Notes on less studied elements

The reaction constant of BeO comprised in MOLDATA is:

+ 2+ BeO(s) + 2 H ↔ Be + H2O logK = 1.1237

Since the previously reported calculations were made Salah and Wang, 2014), it was discovered that in the MINTEQ TDB release 2.40, 17 inorganic species, 6 minerals and 29 organic Be-species are included. The inorganic species and minerals were copied to MOLDATA_R2 and the consistency checked for the carbonate and sulphate complexes by - 2- - 2- comparing the logK values of the HS /SO4 and HCO3 /CO3 reactions. While for the former redox couple no difference exists, for the latter a slight difference is recognizable (logK MINTEQ: 10.3289 versus logK MOLDATA: 10.3267). This difference is however considered to be negligible and not to affect the speciation and solubility results. While with the previously 2- performed calculations BeO2 was predicted to represent the main aqueous species under BC conditions, according to the new calculations Be(OH)2(aq) represents the dominant one (Figure 30 c). It should be mentioned that MINTEQ comprises only the 2nd hydrolysis species, while data for the first, third and fourth hydrolysis constant are not provided therein.

a) b) 1 1

.5 .5 Be++ Be++

Bromellite Eh (volts) Eh (volts) 0 -- 0 BeO2 -- BeO2 µ µ

–.5 –.5 25°C 25°C 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 pH pH

c) d) 1 1

.5 .5 Be++ Be++

Bromellite Be(OH)2 (aq) Eh (volts)

Eh (volts) 0 0 -- BeO-- BeO2 2 µ µ –.5 –.5 25°C 25°C 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 pH pH

Figure 30: Eh-pH diagram of beryllium (Be-C-S-O-H) for the BC reference porewater system. The assumed activity of dissolved [Be] = 10-8. Diagram a) aqueous speciation, b) solid phases included in calculation. Database MOLDATA_R2, diagram c and d) MOLDATA_R2_adapted for Be species according to MINTEQ v2.4. Code: The Geochemist's Workbench- 10.0.

SCK•CEN/12201513 Page 80 of 208 Compilation of Technical Notes on less studied elements

Distribution diagrams provided by Maes and Mesmer (1976) for Be-concentrations of 0.1 and 10-5 M suggest that the onset of hydrolysis for both concentrations starts already at pH 2 with ~100% of BeOH+ (Figure 31).

Figure 31: Distribution of Be-hydrolysis products (x,y) at (a) I = 1 m Be(II) and (b) 10-5 m Be(II) (copied from Baes and Mesmer, 1976)

As the MINTEQ TDB does not comprise bromellite, and the other minerals included therein are less stable/more soluble under BC conditions, the diagram calculated with the adapted TDB (including solids) is not different from the previous one (i.e. diagram b) and predicts this mineral to be the only stable one using an activity of [Be] of 10-8. As can be seen in Table 14, amorphous Be-hydroxide, as well as the alpha and beta polymorphs are characterized by higher solubilities, i.e. ranging around 10-5 M.

In this context it should be mentioned that the solubility of the Be-hydroxides is generally reported to be low (i.e. ~10-7 M; Baes and Mesmer, 1976), which is not reflected by the recent calculations. This suggests that the neutral species Be(OH)2(aq), representing the dominant equilibrium species cannot be nearly as stable as considered in the MINTEQ TDB.

Table 14: Solubility of Be in the BC reference porewater system at 25°C, pH 8.355 and Eh -281 mV. Database MOLDATA_R2 (adapted for Be-species from MINTEQ v2.4. Code: The Geochemist’s Workbench

Solubility controlling phases Solubility [Be], mol/L

-5 Be(OH)2(alpha) 5.737 × 10 -5 Be(OH)2(beta) 2.284 × 10 -5 Be(OH)2(am) 1.145 × 10 Bromellite 9.732 × 10-11

Approximately 50 beryllium minerals occur in nature and over half of these minerals are silicates. The common beryllium silicates are highly insoluble in aqueous solution and are resistant to chemical weathering. According to data compiled by Baes and Mesmer (1976), -7 the solubility of beryllium hydroxide (Be(OH)2(s)) reaches a minimum of about 10 moles/L of Be2+ near pH 8.5. This is equivalent to 0.9 µg/L. In systems in which the dissolved species Be2+, + - BeOH , Be(OH)2(aq), and Be(OH)3 are the only significant forms, the equilibrium solubility at

SCK•CEN/12201513 Page 81 of 208 Compilation of Technical Notes on less studied elements

pH 6.0 would be near 100 µg/L in dilute solutions and would be higher in highly mineralized waters and brines. A potential for concentrations exceeding 1 mg/L evidently exists in acid waters such as might occur in mine drainage or industrial-waste streams.

Within the safety case for the disposal of spent nuclear fuel at Olkiluoto (Finland), Wersin et al. (2014) took thermodynamic data for Be from the Minteq v.4 database and implemented these in the Andra Thermochimie v.7b database. According to this database, Be(OH)2(beta) is the least soluble mineral under all water conditions. Taking into account kinetic considerations, a less soluble X-ray amorphous hydroxide Be(OH)2(am) was selected as the solubility limiting phase instead of the crystalline alpha and beta form. Speciation and solubility of Be under in situ (Olkiluoto) conditions are controlled by pH via the hydrolysis of - - 2- 2- Be. Complexation of Be by F , Cl , CO3 or SO4 only plays a subordinate role. A reference solubility value of 2.5×10-6 M was put forward for saline water.

4.1.2.3 Retention and retardation Studies on sorption of Be are relatively scarce in literature. No experimental sorption data are available for Be2+ on Boom Clay.

You et al. (1989) performed an extensive investigation of Be partition between solid and water using laboratory batch experiments. They observed that Be is strongly held by the solid particles in natural environments under neutral conditions. Among others, they reported Be sorption data for illite and montmorillonite in river and seawater. Kd values determined for clay minerals in river water (pH ~ 7.8) was about 2×105 L/kg. In common sediments such as 5 soil, mud, sand and silt a Kd of ~10 was measured. The solid-liquid distribution is most strongly dependent on the pH of the system (Figure 32), with about 4 orders of magnitude increase in Kd between pH 2 and 6. In seawater, Kd values for illite are the same as those observed in river water. However, beryllium sorption on montmorillonite in sea water was 4 found to be lower with an experimental Kd around 5×10 L/kg.

The sorption behavior of Be on kaolinite and silica was studied by Takahashi et al. (1999) with a focus on the effect of humic and fulvic acids. At pH above 5, Be sorbs very strongly to both silica and kaolinite in absence of humic and fulvic acids. When humic and fulvic acids were introduced, a large portion of Be is soluble. The finding was however explained as a solubility increasing effect. In any case, the study seems to suggest that natural humic substances may enhance the mobility of Be in presence of sorbents like silica and kaolinite.

Taylor et al. (2003) reviewed in detail the environmental behavior of Be. Beryllium is highly mobile in acidic, organic rich continental river waters, whereas the estuarine-ocean mixing zone exhibits a significant scavenging effect. As a result, the beryllium content of ocean waters is approximately three orders of magnitudes less than that of river waters. Beryllium ions can freely substitute for other ions in the lattice of non-beryllium minerals, especially micas. The small size and tetrahedral coordination of the beryllium ion, combined with its tendency to form covalent bonds, suggest that it behaves geochemically more like zinc relative to other alkaline earth ions, although it should be noted that beryllium is more oxophilic than zinc.

SCK•CEN/12201513 Page 82 of 208 Compilation of Technical Notes on less studied elements

Figure 32: Dependence of measured Kd values on the pH value of the solution. Three different systems (illite, kaolinite and river mud) show the same trend in river water (You et al., 1989).

Glazoff and Rashkeev (2010) studied the sorption of Be on alumina and demonstrated that γ- and ε-aluminas (transition Al2O3 polytypes with defect spinel structure) can effectively capture beryllium atoms. Although the bulk crystal structures of these two oxides are characterized only by slight differences in cation vacancy distributions, the interactions of Be with the two polytypes are different. For γ-Al2O3, the Be adsorption energy is high (similar to 5 eV per atom) and all Be atoms are captured and trapped at the surface - all attempts to move Be in the subsurface region result in its expulsion back to the surface. On the other hand, for ε- alumina, Be atoms can be captured both at the surface and in octahedrally coordinated subsurface cation vacancies

Taylor et al. (2012) investigated the cosmogenic fallout radionuclide beryllium-7 as a tracer to estimate soil redistribution rates, residence times and relative contributions of surface material to fluvial systems. Experiments have been carried out, using four representative soils, to assess the geochemical mobility of 7Be under various chemical conditions and the rate and extent of sorption of stable Be. Time-dependent uptake of stable Be over 10 days was complete (i.e. >90% removal) within 0.1 h. No detectable 7Be was found in artificial rainwater solutions following a 24-hour extraction period. Soil-sorbed 7Be was predominantly associated (42-62%) with the reducible fraction of the soil. The exchangeable fraction held 27-37% and <21% was associated with the oxidisable fraction. Incubation of the soils over 53 days indicated that Be was associated with stronger binding sites within the soil matrix as the proportion of exchangeable Be decreased and the proportion of reducible and oxidisable Be increased. The results imply that 7Be is held on binding sites of differing energy and that stronger binding occurs with time of contact.

Yang et al. (2013) investigated the adsorption characteristics of 7Be onto micro-particle surfaces (marine suspended particulate matter, kaolinite, Al2O3, SiO2, CaCO3, Fe2O3, MnO2, and chitin) and the effects of macromolecular organic compounds (Aldrich humic acid, acid polysaccharides and proteins) in natural seawater (< 1 kDa, pH 8.1). Batch adsorption

SCK•CEN/12201513 Page 83 of 208 Compilation of Technical Notes on less studied elements

experiments were performed at 40 mg/L final particulate concentration and total initial 7Be was 150 Bq. Final organic compound concentrations were 1 mg/L. Equilibration time was 2h.

7 Figure 33: Variations in partition coefficients (in log Kd) of Be with different particle types in the presence or absence of model macromolecular organic compounds, i.e. humic acid (HA), acid polysaccharide (APS), and protein (BSA) (Yang et al., 2013)

7 Solid-liquid distribution coefficients, log Kd, ranged from 3.57 to 4.65 for Be. The highest and lowest log Kd value was measured, in general, on Fe2O3 and CaCO3, respectively. Nevertheless, the log Kd values varied little between particle types regardless of the presence or absence of macromolecular compounds (see Figure 33).

In case of lack of reliable experimental sorption data, Ochs et al. (2011) discussed the chemical analogy between Be and Al and suggested that the two elements may be similar in sorption behavior. Because of the much larger first hydrolysis constant compared with other radionuclides of group II, Wersin et al. (2014) put forward that sorption via cation exchange is not very likely for Be.

4.1.2.4 Diffusion and migration Two percolation experiments (type C4) have been performed with Be in confined Boom Clay cores. Details of these experiments can be found below:

1. Percolation C4 experiment Be7/3/9 (NRM023A)

• Initiated 03/07/1997; under Ar

SCK•CEN/12201513 Page 84 of 208 Compilation of Technical Notes on less studied elements

• Clay core code Andra 12/96 – 2; R A10 2.05-2.40 (02/12/1996) (vertical) • Initial activity 69 MBq, in chemical form beryllium chloride • Be concentration in source solution 8.46×10-7 M • Total clay core length: 72 mm (40 mm "inlet" + 32 mm "outlet"), diameter 38 mm • Percolated solutions followed until 22/12/1998 • Slicing was performed on 22/12/1998 (134 samples, average moisture content 16.3 weight%, calculated dry density 1.66 g/cm³)

2. Percolation C4 experiment Be7/3/10 (NRM023B)

• Initiated 03/07/1997; under Ar • Clay core code Andra 12/96 – 2; R A10 2.05-2.40 (02/12/1996) (vertical) • Initial activity 69 MBq, in chemical form beryllium chloride • Be concentration in source solution 8.46×10-7 M • Total clay core length: 72 mm (45 mm "inlet" + 27 mm "outlet"), diameter 38 mm • Percolated solutions followed until 15/01/1999 • Slicing was performed on 11/02/1999 (143 samples, average moisture content 16.4 weight%, calculated dry density 1.71 g/cm³)

The hydraulic conductivity, K (m/s), for the two experiments is given in Figure 34. The value for K is in line with those normally observed in Putte Member of Boom Clay (Yu et al., 2013). The two clay cores exhibit approximately the same hydraulic conductivity. Over the course of the experiment, the conductivity appears to be slowly decreasing, but remains overall in the same range as the initial condition.

Hydraulic Conductivity 2.00 E-12 Be7m3c9 Be7m3c10 1.80 E-12

1.60 E-12

1.40 E-12 Hydraulic conductivity K conductivity (m/s) Hydraulic

1.20 E-12

1.00 E-12 0 100 200 300 400 500 600 Days since start experiment

Figure 34: Calculated hydraulic conductivity, K (in m/s), as function of time in 2 percolation experiments (Be7m3c9 and Be7m3c10)

SCK•CEN/12201513 Page 85 of 208 Compilation of Technical Notes on less studied elements

Be concentration outlet 90,000

80,000

70,000

60,000

50,000 7 (Bq/L) -

Be 40,000 Be7m3c9 30,000 Be7m3c10

20,000

10,000

0 20 40 60 80 100 120 140 160 Days since start experiment

Figure 35: Be concentration in outlet (7Be, as Bq/L, recalculated towards start of the percolation) as function of time

The outlet composition of both experiments (Figure 35 and Figure 36) exhibits a very rapid breakthrough of a very small fraction of the administered Be. The concentration is well below the value for amorphous Be(OH)2 but above the value predicted for bromellite. Also, Be concentrations are observed to increase with time (one order of magnitude in 150 days). Due to the rapid decay of 7Be, no Be was observed beyond the period indicated in the graph (although the experiments were continued for another 150 days, approximately). The rapid breakthrough is in contrast with high sorption (retardation) commonly associated with Be in soils.

Be concentration outlet

1.00 E-12

9.00 E-13

8.00 E-13

7.00 E-13 Be7m3c9 6.00 E-13 Be7m3c10 5.00 E-13

4.00 E-13

Be concentration (mol/L) Be concentration 3.00 E-13

2.00 E-13

1.00 E-13

0.00 E+00 .0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 Volume percolated since start of experiment (mL)

Figure 36: Be concentration in outlet (mol/L) as function of percolated volume

SCK•CEN/12201513 Page 86 of 208 Compilation of Technical Notes on less studied elements

After stopping the experiment, both clay cores were cut in ~ 0.5 mm slices and the 7Be bulk activity in the slices was measured. The obtained profiles are shown in Figure 37. From the profiles it can be observed that the bulk of the administered Be is still located at or near the source position. It is unclear whether this is due to an adsorption or precipitation process since no further attempt was made to perform speciation measurements. Besides the high Be concentration at the source position, the profile over the entire clay core typically shows a quite symmetrical shape which is expected from the diffusion of a mobile species.

Be concentration profile in clay core 100000000

10000000 Be7m3c9 Be7m3c10

1000000

100000 7 concentration (Bq/g) 7 concentration - Be

10000

1000 -50 -40 -30 -20 -10 0 10 20 30 40 Distance from source position (mm)

Figure 37: Be concentration profile (in Bq/g) in the two clay cores, obtained after ~ 280 days percolation. The profiles are given relative to the position of the administered 7Be source (in mm).

Thus, taking together the information from the outlet concentration evolution and the concentration profile in the core, it appears that two separate processes are dominating Be transport in a confined Boom Clay core. The first process is an immobilization mechanism which is either due to strong adsorption or to precipitation of a relatively insoluble phase. The second process is a mobilization mechanism which results in a rapid outflow of part of the Be inventory from the core. Since inorganic Be species are assumed to sorb strongly, this rapid outflow is coupled to a kinetically controlled (colloidal) transport process of Be associated with (mobile) dissolved organic matter. Thus, the phenomenological model for Be transport is similar to that for, e.g., technetium(IV) (Bruggeman et al., 2010).

4.1.2.5 Justification As revealed above, the geochemistry of Be is quite complex and the grouping not straightforward. Based on the recently performed calculations, neutral Be(OH)2(aq) hydrolysis products represent the dominant species under BC conditions. No in-house data for sorption are available at SCK•CEN. According to literature data, the sorption of Be on different substrates (e.g. illite, montmorillonite, soils) measured in various media (e.g. seawater, riverwater) is reported to be generally strong (logKd-values ~3.5-6) and highly pH dependent,

SCK•CEN/12201513 Page 87 of 208 Compilation of Technical Notes on less studied elements

especially up to pH 6. This suggests that Be seems to be mainly sorbed via surface/inner- sphere complexation and/or ligand exchange on edge sites of clays. Wersin et al. (2014) also put forward that sorption via cation exchange seems not very likely for Be, as it is strongly hydrolyzing already in the acid pH range. As such, Be shows rather a similar behavior to Al (having also a similar charge-to-radius ratio) than to zinc, with the latter generally starting to hydrolyze only at higher pH (pH > 6). Sorption studies on soils revealed however that both mechanisms may play a role in the uptake of Be with a redistribution of Be over time towards stronger sorption sites. In presence of organic acids, the sorption was observed to decrease due to the complexation of Be with organic ligands. Percolation experiments (C4-type) revealed on the one hand a rapid breakthrough of Be, which seems contradictory to the high sorption/retardation reported above. At the same time the concentration profiles showed that the bulk of the administered Be was (still) located at the source position, suggesting immobilization either by sorption and/or precipitation. The latter explanation seems consistent with both, i.e. the strong Be-uptake and the low solubility of Be oxides and hydroxides. As the affinity of Be to organics has been reported in literature, explaining the rapid breakthrough observed in the percolation experiments to a DOM linked transport process is obvious and represents the main reason to associate Be to GroupIV elements.

4.1.2.6 References Baes, C.F., and Mesmer, R.E. (1976) The hydrolysis of cations. Wiley, 489 pp.

Bruggeman, C., Maes, N., Aertsens, M., Govaerts, J., Martens, E., Jacops, E., Van Gompel, M., Van Ravestyn, L. (2010) Technetium retention and migration behaviour in Boom Clay, Topical Report, First Full Draft, SCK•CEN-ER-101, 102 pp.

Edmunds, W.M. (2011) Beryllium: Environmental Geochemistry and Health Effects. Reference Module in Earth Systems and Environmental Sciences, 293-301.

Glazoff, M. V. and Rashkeev, S. N. (2010) Beryllium Adsorption at Transition Aluminas: Implications for Environmental Science and Oxidation of Aluminum Alloys, Journal of Physical Chemistry C, 114, 14208- 14212.

Hem, J. D. (1985) Study and interpretation of the chemical characteristics of natural water, Third Edition, U. S geological survey water-supply paper 2254.

Ochs, M., Vieille-Petit, L., Wang, L. and Mallants, D. (2011) Additional sorption parameters for the cementitious barriers of a near-surface repository, Project near-surface disposal of category A waste at Dessel, NIROND–TR 2010–06 E V1, 9 March 2011.

Takahashi, Y., Minai, Y., Ambe, S., Makide, Y., Ambe, F. (1999) Comparison of adsorption behavior of multiple inorganic ions on kaolinite and d-silica in the presence of humic acid using the multitracer technique, Geochimica et Cosmochimica Acta, 63, 815-836.

Taylor, A., Blake, W. H., Couldrick, L., Keith-Roach, M. J. (2012) Sorption behaviour of beryllium-7 and implications for its use as a sediment tracer, Geoderma, 187-188, 16-23.

Taylor, T. P., Ding, M., Ehler, D. S., Foreman, T. M., Kaszuba, J. P., and Sauer, N. N. (2003) Beryllium in the environment: A review, Journal of environmental science and health, Vol-A 38, 439-469.

SCK•CEN/12201513 Page 88 of 208 Compilation of Technical Notes on less studied elements

Wersin P., Kiczka M., Rosch D., Ochs M., Trudel D. (2014) Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto – Radionuclide solubility limits and migration parameters for the backfill, POSIVA 2012-40, 166 pp.

Yang, W., Guo, L., Chuang, C.-Y., Schumann, D., Ayranov, M., Santschi, P.H. (2013) Adsorption characteristics of 210Pb, 210Po and 7Be onto micro-particle surfaces and the effects of macromolecular organic compounds, Geochimica et Cosmochimica Acta, 107, 47-64

You, C.-F., Lee, T., Li, Y-H. (1989) The partition of Be between soil and water, Chemical Geology, 77, 105-118

SCK•CEN/12201513 Page 89 of 208 Compilation of Technical Notes on less studied elements

4.1.3 Technical note for Niobium (Nb)

4.1.3.1 General Niobium (Nb) is a shiny, soft ductile metal with an atomic weight of 92.90638 and an atomic number of 41 (CRC, 2011). The metal oxidizes to form a bluish coating when exposed to air at room temperature for a prolonged period.

The element occurs in niobate (also called columbite, (Fe, Mn)Nb2O6), niobite-tantalite ((Fe,Mn)(Ta,Nb)2O6), pyrochlore ((Ca,Na)2Nb2O6(OH,F)) and euxenite ((Y,Ca,Ce,U,Th)(Nb,Ta, Ti)2O6). Large deposits occur in association with carbonatites (carbonate-silicate rocks), as a constituent of pyrochlore.

There is only one natural isotope of Nb, 93Nb. However, 47 other isotopes have been recognized in products from artificially-induced nuclear reactions. The radioisotopes 94Nb (half-life 2.03 x 104 years) and 93mNb (half-life 16.1 years) are produced by activation of the metallic cladding of fuel elements by irradiation. The former radio-isotope is therefore considered to be important for geosphere transport calculations, but 93mNb is much less important.

4.1.3.2 Speciation and solubility There is relatively little thermodynamic data available for Nb aqueous species and solids and no data are contained in the following commonly used databases: the NEA-TDB (Östhols and Wanner, 2000); the LLNL V8 R6 "combined" dataset, which is distributed with the Geochemist’s Workbench (GWB) software (Bethke, 1996, 2008) as “thermo.com.V8.R6; the Visual MINTEQ database, release 2.40, which is also distributed with GWB; and the YMP database (USDOE, 2007). However, there are data in the Thermochimie v.9/SIT database (Duro et al., 2006a; Grivé et al., 2014), the Nagra-PSI thermodynamic database (Hummel et al., 2002) and the thermodynamic database “JAEA-TDB” produced by the Japan Atomic Energy Agency (JAEA) (Kitamura et al., 2010a,b):

Nb is not redox-sensitive under the range of relevant conditions and the only oxidation state of concern is Nb(V) (Crawford, 2010). Thermochimie, v.9 contains data for the following aqueous Nb(V) species:

+2 • Nb(OH)3 + • Nb(OH)4

• Nb(OH)5 - • Nb(OH)6 -2 • Nb(OH)7 + • NbO2(H3Cit)

• NbO2(HOx) - • NbO2(HOx)2 - • NbO2(Ox) The database also contains data for three solid Nb phases:

SCK•CEN/12201513 Page 90 of 208 Compilation of Technical Notes on less studied elements

• NaNbO3(s) • Nb(cr) • Nb2O5(s)

The data in the Thermochimie database are the same as those in the Nagra-PSI database, except that Thermochimie gives estimated uncertainties on the data. Ervanne et al. (2013) used PHREEQC and the Thermochimie database, v.7b (which contains the same Nb data as version 9) to calculate the distribution of Nb among different aqueous species over a range of pH (6 to 11) in four different reference water compositions representing groundwaters from Olkiluoto in Finland: fresh groundwater; saline groundwater; glacial meltwater; and brackish groundwater. The results are shown for the saline reference water in Figure 38 similar results were obtained for the other reference waters. However, Ervanne et al. (2013) also found that measurements of Nb sorption on kaolinite and illite could not be modelled accurately across the full pH range considered (6 to 11) using only the species in the Thermochimie database. Their results implied that other species could be important in certain groundwater conditions, possibly Ca-niobates or niobium carbonates.

Figure 38: Aqueous speciation of niobium (10-8 M) in saline Olkiluoto reference water as pH is varied between 6 and 11, after Ervanne et al. (2013). The curves were calculated by PHREEQC and ThermoChimie database v 7b (note that Nb data in this version of the thermodynamic database are the same as those in v 9 used to calculate the results in Table 15 below). Similar speciation results were obtained for all the Olkiluoto reference waters.

In contrast to the Thermochimie database, the JAEA-TDB contains data for only two aqueous - species, i.e. Nb(OH)6 and Nb(OH)5 and a single solid phase (Nb2O5(s)). For near-neutral conditions these two species would probably account for most of the dissolved Nb, but by comparison with the Thermochimie database (Figure 38) it seems likely that a substantial 2- proportion could be accounted for by other species, notably Nb(OH)7 . Furthermore, the modelling with Thermochimie implies that Nb(OH)5 will account for decreasing proportions - of Nb at increasingly alkaline pH, and Nb(OH)6 will also decrease in abundance with

SCK•CEN/12201513 Page 91 of 208 Compilation of Technical Notes on less studied elements

increasing pH above about 8. An implication is that the JAEA-TDB may give inaccurate results if applied to alkaline solutions.

Crawford (2010), when reporting data for use by SKB in its SR-Site safety assessment considered that the dominant aqueous species under repository conditions at Forsmark – would be NbO3 and Nb(OH)5. At near-neutral conditions, these species were considered by Crawford (2010) to have roughly equal activities, but the former was thought to become progressively more important with increasing pH. The speciation of Nb is reported by Crawford (2010) to be based on “literature data”, but the origins of this data are not specified. However, other sources state that SKB has used data from the Nagra-PSI thermodynamic - database 01/01 (Hummel et al., 2002), which has NbO3 as the master species for Nb (Duro et al., 2006b). The thermodynamic data for Nb aqueous species and solid phases in the Nagra- PSI thermodynamic database are based on data reported in Wagman et al. (1982) and Lothenbach (1999).

The solubilities of Nb, as controlled by equilibrium with the three solid phases present in the Thermochimie version 9.0 thermodynamic database (Duro et al., 2006a; Grivé et al., 2014), were calculated for BC porewater using PHREEQC Interactive v 2.18.3514 (Table 15). These calculations show that the least soluble of the considered phases is Nb2O5(s), which would constrain the solubility of Nb(V) to be 2.54×10-6 molal.

Table 15: Speciation of Nb(V) when concentrations are constrained by equilibrium with NaNbO3(s), Nb(cr) or Nb2O5(s) in the presence of BC reference porewater (composition from De Craen et al., 2004). The speciation and saturation indices were calculated using PHREEQC Interactive v2.18.3514 and the Thermochimie version 9.0 thermodynamic database (Duro et al., 2006a; Grivé et al., 2014) for a temperature of 16 ̊C. Note that Nb data in this version of the thermodynamic database are the same as those in v.7c used to calculate the results in Figure 38 above).

Total Nb Activities of Nb Species Saturation Indices of Nb Phases Concentration +2 + - -2 Molal Nb(OH)3 Nb(OH)4 Nb(OH)5 Nb(OH)6 Nb(OH)7 NaNbO3(s) Nb(cr) Nb2O5(s)

2.54E-06 2.73E-24 1.13E-16 6.60E-10 1.66E-06 8.79E-07 -3.9951 -119.8014 0

5.00E+03 1.00E+03 1.00E+03 1.00E+03 1.00E+03 1.00E+03 94.0286 0 203.7645

3.12E-02 3.85E-20 1.29E-12 6.99E-06 1.89E-02 1.24E-02 0 -115.7752 8.0514

Kitamura et al. (2010a) similarly considered that Nb2O5(s) could be a solubility-limiting phase for Nb(V), on the basis of experiments carried out by Yajima (1994) and Yajima et al. (1992) (Figure 39). For a pH similar to the BC reference porewater (8.5) it can be seen that their experimental data imply a solubility consistent with that calculated using PHREEQC and the Thermochimie thermodynamic database, when Nb2O5(s) is specified to control solublity (Table 15).

Kitamura et al. (2010a) carried out experiments to independently check the results of Yajima (1994) and Yajima et al. (1992) and obtained equilibrium constants for the following reactions:

- + Nb2O5(s) + 7 H2O = 2 Nb(OH)6 + 2H (1)

SCK•CEN/12201513 Page 92 of 208 Compilation of Technical Notes on less studied elements

- + Nb(OH)5(aq) + H2O(l) = Nb(OH)6 + H (2)

For (1) logK = -28.9130 and for (2) logK = -6.758, values which concurred with the earlier work of Yajima (1994) and Yajima et al. (1992). However, although Kitamura et al. (2010a) treated the solubility of Nb2O5(s) as an upper limit, they acknowledged that the solubility data for this phase was the only they had identified. Hence the solubility was considered to be uncertain, particularly at high pH, although solubility clearly increases with increasing pH. Kitamura et al. (2010a) quoted work by Wu et al. (2009) that indicates Nb2O5(s) might transform to Na8Nb6O19·13H2O(s) in alkaline solutions. Kitamura et al. (2010) also referred to work by Talerico et al., (2004) concerning the solubility of Nb in the presence of cement, indicating that Nb(V) solubility decreases with increasing Ca concentration.

Figure 39: Solubility of Nb(V) under anaerobic conditions, as measured by Yajima (1992) and Yajima et al (1994), as plotted in Kitamura et al. (2010a). Plots with mark “�” were taken from the first run (analyzed by ICP-OES) and others were taken from a second run (analyzed by ICP-MS). The solid line is a least-square regression assuming - that dissolved Nb is in the form Nb(OH)5(aq) and Nb(OH)6 , with the solubility limiting phase being Nb2O5(s); dashed lines show the contribution from each aqueous species.

Vuorinen and Snellman (1998) also highlight the strong pH-dependence of Nb solubility for alkaline pH values. They quoted experimental data reported by Kulmala and Hakanen (1993) that showed Nb solubility to increase by around 3 to 4 orders of magnitude between pH = 7 and pH = 13.

For use in a safety assessment for a HLW/SF repository by Nagra during Project Opalinus Clay, Wersin and Schwyn (2004) proposed a reference Nb(V) solubility under reducing (in-situ repository) conditions (pH=7.25) of 3×10-5 mol/L, with a minimum value of 1×10-8 mol/L and an upper limit of 1×10-4 mol/L. For oxidizing conditions a solubility of 3×10-5 mol/L was recommended.Nb2O5 was specified to be the solubility-controlling phase.

- Speciation calculations using MOLDATA indicate that the niobiate anion Nb(OH)6 is the prevalent species under undisturbed BC conditions (Figure 40). In Table 17, the species SCK•CEN/12201513 Page 93 of 208 Compilation of Technical Notes on less studied elements

- 2- distribution in equilibrium with Nb2O5(s) is summarized. Besides Nb(OH)6 , Nb(OH)7 represents the second most important aqueous species.

At [Nb] = 10-8, no phase is predicted to be stable under BC conditions. By increasing the niobium activity by 3 orders of magnitude in the calculations, Nb2O5(s) appears in the Pourbaix diagram and as can be seen in Figure 40 b is stable between pH = 0-9 over the entire Eh range. NbO2(s) is very soluble under BC conditions.

a) b)

1 1

++ Nb(OH)+3 .5 Nb(OH)4 .5 Nb(OH)5 Nb2O5(s) - Nb(OH)6 Eh (volts) Eh Eh (volts) Eh 0 0 Nb(OH)-- -- 7 Nb(OH)7 µ µ

–.5 –.5

25°C 25°C 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 pH pH

Figure 40: Eh-pH diagram of niobium (Nb-C-S-O-H) for the BC reference porewater system. Diagram a) activity of dissolved [Nb] = 10-8, b) minerals included in calculation, activity of dissolved [Nb] = 10-5. Code: The Geochemist's Workbench -8.08.

Table 16: Solubility of Nb in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV. Database: MOLDATA. Code: The Geochemist's Workbench- 8.08.

Solubility controlling phases Solubility, [Nb], mol/L

2 -6 Nb2O5 (s) 2.4 × 10 3 NbO2(cr) very soluble

Source data: 2ANDRA TDB, 3NAGRA/PSI

The reaction constants of the Nb minerals comprised in Table 16 and MOLDATA are the following:

- + Nb2O5(s) + 7 H2O ↔ 2 Nb(OH)6 + 2 H log K = -28.38 - + NbO2(cr) + 3.5 H2O + 0.25 O2(aq) ↔ Nb(OH)6 + H log K = 16.68

Table 17: Species distribution of Nb in equilibrium with Nb2O5(cr). Database: MOLDATA. Code: The Geochemist's Workbench- 8.08.

SCK•CEN/12201513 Page 94 of 208 Compilation of Technical Notes on less studied elements

Aqueous species Nb [mol/L] Percentage [%]

- -6 Nb(OH)6 1.65 × 10 69

2- -7 Nb(OH)7 7.52 × 10 31

Total 2.36 × 10-6 100

4.1.3.3 Sorption and retardation There is some uncertainty about the sorption mechanism for Nb and Bradbury et al. (2010) state that the sorption mechanism is unknown. However, Linklater (2003) pointed out that Nb shows a strong tendency to hydrolyse and that such behaviour is often associated with strong sorption, particularly on oxide/hydroxide solid surfaces are involved. This suggests that Nb sorption should be strong, at least when oxides / hydroxides are present in the solid phase. Crawford (2010) considered that Nb probably sorbs by an inner-sphere surface complexation mechanism. In this second case, sorption should be relatively insensitive to the ionic strength of the solution. In contrast, the increasing dominance of negatively charged - -2 species at increasingly alkaline pH (e.g. Nb(OH)6 , then Nb(OH)7 ) should mean that there is a decreasing tendency to sorb in more alkaline solutions, when mineral surfaces tend to acquire an increasingly negative charge (Stenhouse, 1995; Linklater, 2003; Bradbury et al., 2010). However, Bradbury and Baeyens (2003a) report a review of sorption data for Nb on various mineral phases that showed no conclusive evidence for a change in Kd across the pH range 6.6 to 8. This was attributed to the dominance of the neutral hydoxy species Nb(OH)5 under these pH conditions, although this is inconsistent with the thermodynamic data described in the previous sections. For Nb sorption in MX-80 bentonite across this pH range Bradbury and Baeyens (2003a) recommended a Kd of 30 m3/kg, with an uncertainty factor of 5. This Kd is much higher than the “realistic” value of 1 m3/kg recommended for bentonite in earlier work by Stenhouse (1995), based partly on using Zr as an analogue. Stenhouse (1995) recommended the same “realistic” value for crystalline rock, but a smaller value of 0.5 m3/kg for marl. Stenhouse (1995) gave conservative values of 0.1 m3/kg in bentonite and crystalline rock and 0.05 m3/kg in marl. These values were considered appropriate for groundwater with ionic strength in the range 0.01 to 0.2.

For Opalinus Clay, again based on published literature, Bradbury and Baeyens (2003b) recommended a Kd of 4 m3/kg for pH between 6.3 and 6.8. However, they gave a large uncertainty factor of 5.

More recently Bradbury et al. (2010) have recommended Kd values for Nb in Swiss argillaceous rocks based on an assessment of experimental data reported by Legoux et al. (1992). In view of uncertainties about the speciation of Nb over the relevant pH range (6-8) and its consequent impact upon sorption, Bradbury et al. (2010) adopted a conservative approach giving rise to Kd values similar to those recommended by Stenhouse (1995). That is, a Kd value of 1 m3/kg was recommended for argillaceous rock systems in which clay contents are c.10 wt% and pH lay in the range 6-10 (i.e. similar to those investigated experimentally by Legoux et al. (1992)). For any systems where pH > 8, Nb(V) was conservatively considered not to sorb owing to being dominantly in the form of hydroxyl anions. Conservatively, the Kd was assumed not to increase for 2:1 clay contents >10 wt%. However, for 2:1 clay contents less than this value, it was recommended to scale the Kd in proportion to the clay (e.g. argillaceous rock with 5 wt% 2:1 clay would have Kd of 0.5 m3/kg.

SCK•CEN/12201513 Page 95 of 208 Compilation of Technical Notes on less studied elements

For use in SKB’s SR-Site safety assessment, Crawford (2010) proposed lower values than those given by Stenhouse; a best estimate Nb Kd value of 1.98×10–2 m3/kg was recommended for groundwater at both Forsmark and Laxemar for all groundwater compositions (ranging from fresh to saline, with ionic strength up to c. 2). The recommended uncertainty distribution is log-normal, with log10 Kd = –1.70±0.64. These values were derived from experimental results reported by Kulmala and Hakanen (1993). In this study, sorption was measured on two crushed rock samples taken from different sites. These were a tonalite (Olkiluoto) crushed to a particle size specified as less than 3 mm, and a Rapakivi granite (Hästholmen) crushed to a particle size specified as less than 2 mm. The sorption experiments were carried out under oxic conditions using natural groundwater samples native to each site.

For use in a safety assessment by Nagra during Project Opalinus Clay, Wersin and Schwyn (2004) recommended a reference Kd based on the work of Bradbury and Baeyens (2003b). For Nb(V) in a bentonite buffer Wersin and Schwyn (2005) recommended a value of 30 m3/kg, with a lower limit of 1 m3/kg and an upper limit of 900 m3/kg. An uncertainty factor of 30 m3/kg was specified. In contrast, for the Opalinus Clay itself, a reference case value of 4 m3/kg was recommended, with lower and upper values of 0.1 m3/kg and 100 m3/kg respectively. An uncertainty factor of 30 was again given.

Vilks (2011) reviewed sorption data for Nb(V) in order to deduce appropriate values for saline groundwater. He found Kd-measurements for only relatively narrow ranges of salinities and pH, across which Kd-values did not vary significantly. In the presence of water with ionic 3 strength of 0.72 and pH in the range 7.1 to 7.5 Kd-values were found to range from 1.2 kg/m to 1.8 m3/kg with a geometric mean of 1.5 m3/kg. For shale in the presence of similar porewater, the Kd-values were found to range from 1.4 m3/kg to 2.8 m3/kg with a geometric mean of 1.7 m3/kg.

SCK•CEN/12201513 Page 96 of 208 Compilation of Technical Notes on less studied elements

4.1.3.4 Transport and diffusion For its SR-Site safety assessment of the proposed SF repository in crystalline rock at -10 2 Forsmark, SKB used median effective diffusion coefficient values, Deff of 1.33×10 m /s, 1.59×10-10 m2/s and 2.0×10-14 m2/s for the buffer, backfill and host rock respectively (SKB, 2010). Corresponding diffusion-accessible porosities were 0.45 and 0.46 and 1.8×10-3 for buffer, backfill and host rock respectively. For flow path averaged Deff, the following log- normal distribution was used:

log = 14.2 ± 0.5

10 𝑒𝑒𝑒𝑒𝑒𝑒 These transport parameters were derived for𝐷𝐷 the −SR-Site assessment from laboratory measurements. The uncertainty on the diffusivities was taken to be log-normally distributed, with different characteristics for anions and cations. Nb was taken to be in anionic form.

The geosphere accessible porosity is an estimated flow-path average, obtained by scaling laboratory measurements by a factor of 0.8 to allow for the possible bias introduce by mechanical damage due to drilling and core extraction. The scaling factor was based on comparison of water saturation porosities with microcrack volume (porosity) measurements made in triaxial compression tests using site-specific rock samples.

For use in a safety assessment by Nagra during Project Opalinus Clay, Wersin and Schwyn (2004) recommended an effective diffusion coefficient of 2×10-10 m2/s and an effective porosity of 0.36 for Nb(V) in a bentonite buffer. For the Opalinus Clay itself, these authors recommended a reference case value of 1×10-11 m2/s, with and upper pessimistic limit of 1×10-10 m2/s. An effective porosity of 0.12 was specified.

4.1.3.5 Justification Neither in-house sorption nor migration data are available, due to which the lines of evidence for the grouping are limited and also debatable. Based on the speciation calculations, Nb - occurs predominantly as negatively charged Nb(OH)6 species under the BC reference conditions. Despite this fact, it is generally considered that strongly hydrolyzing elements are also sorbing strongly. This reasoning was also adopted here due to which we associated Nb rather to the group of transition metals than to the group of anions (as initially done). It should be mentioned, that there exists however quite some uncertainty concerning the exact sorption mechanism for Nb. As revealed by speciation calculations, Nb is forming increasingly negative hydrolysis species with increasing pH (and increasing negative surface). Consequently sorption is expected to decrease with increasing pH and to be rather low at the BC reference pH of 8.36. Association of Nb with natural organic matter and DOM linked transport of Nb has not been observed yet and only put forward based on analogy with other multivalent metals. Getting more insight into the retention and migration behaviour of Nb is certainly needed.

4.1.3.6 References Bethke C.M. (1996) Geochemical Reaction Modeling, Concepts and Applications. Oxford University Press, 397 pp.

SCK•CEN/12201513 Page 97 of 208 Compilation of Technical Notes on less studied elements

Bethke C.M. (2008) Geochemical and Biogeochemical Reaction Modeling. Cambridge University Press, 547 pp.

Bradbury M.H. and Baeyens B. (2003a) Near-field sorption data bases for compacted MX-80 bentonite for performance assessment of a high-level radioactive waste repository in Opalinus Clay host rock. Nagra Technical Report NT-02-18.

Bradbury M.H. and Baeyens B. (2003b) Far-field sorption data bases for performance assessment of a high-level radioactive waste repository in an undisturbed Opalinus Clay host rock. Nagra Technical Report NTB-02-19.

Bradbury M.H., Baeyens B. and Thoenen T. (2010) Sorption data bases for generic Swiss argillaceous rock systems. Nagra Technical Report NTB 09-03.

CRC (2011) Handbook of Chemistry and Physics, CRC Press, 92nd Edition.

De Craen M., Wang L., Van Geet M. and Moors H. (2004) The geochemistry of Boom Clay pore water at the Mol site, status 2004. SCK•CEN-BLG-990.

Crawford J. (2010) Bedrock Kd data and uncertainty assessment for application in SR-Site geosphere transport calculations. SKB Report R-10-48.

Duro L., Cera E., Grivé M., Domènech C., Gaona X. and Bruno J. (2006a) Development of the ThermoChimie thermodynamic database. Janvier 2006, Prepared by Enviros Spain S. L. National Radiactive Waste Management Agency (ANDRA) report C.RP.0ENQ.06.0001, Châtenay-Malabry cedex, France.

Duro L., Grivé M., Cera E., Domènech C. and Bruno J. (2006b) Update of a thermodynamic database for radionuclides to assist solubility limits calculation for performance assessment. SKB Technical Report TR-06-17.

Ervanne H., Puukko E. and Hakanen M. (2013). Modeling of sorption of Eu, Mo, Nb, Ni, Pa, Se, Sn, Th and U on kaolinite and illite in Olkiluoto groundwater simulants. Posiva Working Report 2013-31.

Grivé M., García D., Campos I. and Colàs E. (2014) Release of ThermoChimie Version 7c: Track changes document Project ANDRA-TDB8. ANDRA Report CCRPFSTRI400I0.

Hummel W., Berner U., Curti E., Pearson F.J. and Thoenen T. (2002). The Nagra / PSI Chemical Thermodynamic Data Base 01/01. Nagra Technical Report NTB 02-16.

Kitamura A., Fujiwara K., Doi R., Yoshida Y., Mihara M., Terashima M. and Yui M. (2010a) JAEA Thermodynamic database for Performance Assessment of geological diposal of high-level radioactive and TRU wastes. JAEA-Data/Code 2009-024.

Kitamura A., Fujiwara K. and Yui M. (2010b) JAEA Thermodynamic Database for Performance Assessment of Geological Disposal of High-level and TRU Wastes: Refinement of Thermodynamic Data for Trivalent Actinoids and Samarium. JAEA-Review 2009-047.

Kulmala S., Hakanen M. (1993) The solubility of Zr, Nb and Ni in groundwater and concrete water, and sorption on crushed rock and cement. Report YJT-93-21, Nuclear Waste Commission of Finnish Power Companies.

Legoux Y., Blain G., Guillaumont R., Ouzounizian G., Brillard L. and Hussonnois M. (1992) Kd measurements of activation, fission and heavy elements in water/solid phase systems. Radiochimica Acta, 58/59, 211-218.

SCK•CEN/12201513 Page 98 of 208 Compilation of Technical Notes on less studied elements

Linklater C.M., Moreton A.D. and Tweed C.J. (2003) Analysis and Interpretation of Geosphere Sorption Data for a Nirex Performance Assessment. United Kingdom Nirex Limited, Nirex Report no. N/083.

Lothenbach B., Ochs M., Wanner H. and Yui M. (1999) Thermodynamic data for the speciation and solubility of Pd, Pb, Sn, Sb, Nb and Bi in aqueous solution. LNC TN8400 99-011, Japan Nuclear Cycle Development Institute, Ibaraki, Japan.

Östhols E. and Wanner H. (2000) TDB-0: The NEA Thermochemical Database Project. OECD/NEA, Paris.

Vuorinen U., Kulmala S. and Hakanen M. (1998) Solubility database forTILA-99. Posiva Report POSIVA 98-14.

SKB (2010) Radionuclide transport report for the safety assessment SR-Site. SKB Technical Report TR- 10-50.

Stenhouse M.J. (1995). Sorption databases for crystalline, marl and bentonite for performance assessment. Nagra Technical Report 93-06.

Talerico C., Ochs M. and Giffaut E. (2004). Solubility of niobium(V) under cementitious conditions: importance of Ca-niobate, Materials Research Society Symposium Proceedings, 824, pp.CC8.31.1– CC8.31.6.

United States Department of Energy (USDOE) (2007) In-Drift Precipitates/Salts Model. ANL-EBS-MD- 000045 REV 03.

Vilks (2011) Sorption of selected radionuclides on sedimentary rocks in saline conditions – literature review. NWMO Report NWMO TR-2011-12.

Vuorinen U. and Snellman M. (1998). Finnish reference waters for solubility, sorption and diffusion studies. Posiva Working Report 98-61.

Wagman D.D., Evans W.H., Parker V.B., Schumm R.H., Halow I., Bailey S.M., Churney K.L. and Nuttall R.L. (1982). The NBS tables of chemical thermodynamic properties: Selected values for inorganic and C1 and C2 organic substances in SI units. Journal of Physical and Chemical Reference Data, 11, Supplement No. 2, 1-392.

Wersin P. and Schwyn B. (2004) Project Opalinus Clay: Integrated approach for the development of geochemical databases used for safety assessment. Nagra Technical Report NTB-03-06.

Yajima T. (1994) Solubility measurements of uranium and niobium. Report of Yayoi Kenkyukai, UTNL-R 0331, University of Tokyo, pp. 127–144 (1994) [in Japanese].

Yajima T., Tobita S. and Ueta S; (1992) Solubility measurements of niobium in the system Nb-O-H under CO2-free condition”, presented at 1992 Fall Meeting of Atomic Energy Society of Japan, F33, p. 341 [in Japanese].

Wu S.Y., Zhang W. and Chen X.M. (2009) Formation mechanism of NaNbO3 powders during hydrothermal synthesis. Journal of Materials Science: Materials in Electronics, 21, 450-455.

SCK•CEN/12201513 Page 99 of 208 Compilation of Technical Notes on less studied elements

4.1.4 Technical note for Nickel (Ni)

4.1.4.1 General Nickel (Ni) is a Group 10 metallic element with an atomic number of 28 (Winter, 2014). Nickel can exist as a number of compounds, including oxides and sulphides, and it is found in a number of rock-forming minerals and sulphide mineral deposits (Winter, 2014; Duro et al., 2006). Nickel commonly has an oxidation number of 2 (Winter, 2014; Duro et al., 2006). Nickel has an abundance of ~0.2 ppb in seawater, whereas in stream water it is ~0.3 ppb, and in crustal rocks, it is ~90 ppm (Winter et al., 2014).

4.1.4.2 Speciation and solubility NiCO3(cr) has been identified as a potential solubility-limiting phase for Opalinus Clay conditions giving a dissolved nickel concentration of 3.1×10-5 mol/dm3 (noting that the limiting solid changes to Ni(OH)2(cr) below log pCO2 = -3.5 (Berner, 2002). For the spent fuel disposal assessment by Posiva (‘TURVA-2012’), Ni(OH)2(s) was adopted as a solubility-limiting phase in the near field (water associated with canister/bentonite buffer, Wersin et al., 2014a).

Assuming a nickel activity of 10-8, the aqueous speciation under Boom Clay conditions is, according to the MOLDATA database, dominated by carbonate complexes of the 2- composition Ni(CO3)2 (Salah and Wang, 2014). Solubility calculations show that Ni-oxide (NiO), Ni-hydroxide [β-Ni(OH)2], Ni-silicate (Ni2SiO4), as well as the Ni-carbonates NiCO3(cr) and NiCO3∙5.5 H2O(cr) are characterized by quite high solubilities under Boom Clay conditions, while the sulphides, millerite (β-NiS), violarite (FeNi2S4) and vaesite (NiS2) are less soluble (Table 18).

Table 18: Solubility of Ni in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV. Database: MOLDATA. Code: The Geochemist's Workbench (from Salah and Wang, 2014).

Solubility controlling phases Solubility, [Ni], mol/L

Bunsenite NiO (cr)1 3.0 × 10-2 1 -3 Ni2SiO4 (oliv) 2.7 × 10 1 -4 Nickel hydroxide (Ni[OH]2,beta) 7.2 × 10 2 -5 Trevorite (NiFe2O4) 1.8 × 10 1 -5 Gaspéite (NiCO3,cr) 2.6 × 10 1 -7 Vaesite NiS2 (cr) 6.2 × 10 Millerite (β-NiS)1 1.8 × 10-8 2 -9 Violarite (FeNi2S4) 9.5 × 10 1 NiCO3 × 5.5 H2O(cr) very soluble

It should be noted that trevorite (NiFe2O4) belongs to the spinel group of minerals and it been suggested that it is less likely to form under low temperature conditions (Bruno et al., 2001). Suppressing trevorite in the calculations, results in NiCO3(cr) being a candidate

SCK•CEN/12201513 Page 100 of 208 Compilation of Technical Notes on less studied elements

solubility-limiting solid. Ni(OH)2 is also considered to be a candidate solubility-limiting phase and it has been suggested that nickel-bearing olivine and sulphide minerals are more likely to form under elevated temperature conditions (Salah and Wang, 2014). Bunsenite (NiO) is reported to form either due to Ni(OH)2 dehydration at 440-558 K (Hummel et al., 2002) or by hydrothermal decomposition of gaspéite (NiCO3) and is therefore discounted as a solubility- limiting phase (Salah and Wang, 2014).

When nickel concentration is set by Ni(OH)2(beta) solubility under Boom Clay conditions, 2- speciation calculations predict that 98.9% of the dissolved Ni will be present as Ni(CO3)2 , + 2+ with very small amounts of NiCO3(aq) Ni(HCO3) and Ni (Salah and Wang, 2014).

Using Geochemist’s Workbench® (Bethke, 2008) and the ‘MINTEQ’ (thermo_minteq.dat) database, solubility diagrams calculated for nickel by QUINTESSA are illustrated in Figure 42. The most stable solid in the MINTEQ database is NiS(gamma), followed by NiS (beta) and then NiCO3. The MINTEQ database does not appear to include some of the minerals in MOLDATA, such as violarite and trevorite. The MINTEQ database predicts that NiCO3(aq) will be the dominant nickel species under Boom Clay conditions, as the database does not 2- include Ni(CO3)2 .

0 (a)

–5 NiS (gamma) ++

- -- –10 H2CO3* (aq) HCO3 CO3 Ni++ log a Ni log

Ni(OH)- –15 3 NiCO (aq) Ni(OH)3 2 (aq)

16°C –20 0 2 4 6 8 10 12 14 pH

0 (b)

Ni(OH)2 (c) NiS (beta) –5 ++

- -- –10 H2CO3* (aq) HCO3 CO3 Ni++ log a Ni log - Ni(OH)3 –15 NiCO (aq) Ni(OH)3 2 (aq)

16°C –20 0 2 4 6 8 10 12 14 pH

SCK•CEN/12201513 Page 101 of 208 Compilation of Technical Notes on less studied elements

0 c)

NiCO3 Ni(OH)2 (c) –5 ++

- -- –10 H2CO3* (aq) HCO3 CO3 Ni++ log a Ni log Ni(OH)- NiCO3 (aq) 3 Ni(OH)2 (aq) –15

16°C –20 0 2 4 6 8 10 12 14 pH

Figure 42: Solubility diagrams for nickel, assuming Ca2+ activity buffered by calcite, sulphate buffered by pyrite, 2+ - Fe buffered by siderite, log f CO2(g) = -2.44, log a Cl = -3.155, log f O2(g) = -71.4 (Eh = -0.28 mV at pH 8.36, 16 °C, 1 bar). Solute activity buffers are the same as those assumed for models of the Boom Clay porewater discussed by Salah and Wang (2014). Diagrams generated using ‘Act2’ module of Geochemist’s Workbench® (Bethke, 2008) and speciated for carbonate. Diagram (a) includes all solids; (b) excludes NiS(gamma); (c) excludes NiS(gamma) and NiS(beta).

4.1.4.3 Sorption and retardation Within the Catclay project the sorption of Ni on Silver Hill Illite (IMt-2), Illite du Puy and Boom Clay was studied under various conditions, i.e. pH (sorption edges), pCO2 (isotherms) and concentrations (isotherms). Part of the experiments was performed in presence of different concentrations of organic matter, added as purified BCHA. In Figure 43 the experimental results for the Ni-sorption on Silver Hill Illite are illustrated. Under atmospheric pCO2 conditions (0.04% CO2, no OC), the Ni-sorption can be considered as quasi constant over the -9 -5 studied concentration range (10 to 10 M) with log Kd-values differing by maximum 0.5 log units (Figure 43a). Under in-situ BC conditions (i.e. 0.4% CO2, but no OC), the logKd-values are -7 similar to the ones measured at 0.04% CO2, but from Ceq ~10 M, the sorption coefficients are decreasing with increasing Ni equilibrium concentrations (Figure 43c). A possible 2- explanation for this observation could be the enhanced formation of soluble Ni(CO3)2 complexes at higher Ni concentrations. Comparing the CO2-poor (pCO2 = atm = 0.04%) or CO2-free systems (Figure 43a), it becomes clear that organic carbon plays an important role in the Ni-sorption behaviour. For a TOC concentration of 30 mg C/L (no IC), log Kd-values are significantly lowered (by at least 0.5 log units) compared to the system without OC. This can be explained by the formation of organic Ni-colloids/complexes (< 0.22µm) in general reducing the sorption. It is however remarkable that in these experiments (with OC), sorption (logKd-values) is increasing with increasing Ni equilibrium concentrations (Figure 43a and c). In the experiments with IC of 0.4 % CO2 (Figure 43d) the positive slopes of the sorption isotherms are even steeper than for the batch systems without IC (Figure 43b). Also here it is interesting to observe that for the system with 10 mg C/L (and additional IC), Ni sorption on illite is higher than for the system without OC. Until now, no explanation was found for these observations. The formation of binary (Ni-HA) and ternary (Ni-HA-CO3) surface complexes could be a possible hypothesis in absence and presence of inorganic carbon, respectively.

SCK•CEN/12201513 Page 102 of 208 Compilation of Technical Notes on less studied elements

a) c)

b) d)

Figure 43: Sorption isotherms and log Kd-values as function of the Ni-equilibrium concentrations for Ni on Silver Hill Illite at various conditions. In the left graphs the isotherms measured at pCO2atm = 0.04% and in CO2-free atm with different amounts of organic carbon (0-30 mg/L) are shown. The right graphs show the sorption results measured at in-situ pCO2 of 0.4 % and different amounts of organic carbon (0-40 mg/L).

A sorption edge for Ni on Silver Hill Illite was not determined, but the pH dependency of the Ni sorption on Illite du Puy was examined. The sorption edge (Figure 44) exhibits a typical "edge" form with lower sorption in the acid pH range to higher sorption in the alkaline pH range with maximum logKd-values of 4 at ~pH 8-9. This latter value corresponds to the log Kd observed for Silver Hill Illite at pH 8.4, determined in the isotherm experiment.

The sorption of Ni on Boom Clay was measured in two different background solutions, i.e. Real Boom Clay Water (RBCW) and Synthetic Boom Clay Water (SBCW), which has a similar composition as RBCW, but does not comprise organic matter.

The log Kd-values measured in SBCW (Figure 45) show that the sorption behaviour is constant over a broad concentration range. The log Kd-values in RBCW could be determined only in a narrower range due to the background concentration of Ni in RBCW of ± 6.5×10-7 M and are slightly increasing with increasing equilibrium concentration. The lower Kd-values determined in RBCW compared to SBCW can be referred to the higher DOC concentrations in the RBCW suspensions, leading to the formation of soluble organic Ni-complexes/colloids, reducing the sorption of Ni onto the solid clay surfaces.

SCK•CEN/12201513 Page 103 of 208 Compilation of Technical Notes on less studied elements

3 Ni (L/kg) d 2 log K 1

0 2 3 4 5 6 7 8 9 10 11 12 pH

Figure 44: Sorption edge for Ni on Illite du Puy (S/L:1 g/L) at IS 0.1 M NaClO4. Error bars (2σ) calculated by error propagation

4 (L/kg)

d 3 log K log SBCW 2 RBCW 1 -9 -8 -7 -6 -5 log C (mol/L) eq Figure 45: Log Kd-values of Ni on Boom Clay determined in SBCW and RBCW. Error bars (2σ) were calculated by error propagation.

In general, like other transition metals, nickel has been found to sorb strongly to oxides and hydroxides (particularly of iron and manganese) and silicate minerals including clays (Linklater et al., 2003).

Sorption of nickel onto oxides/hydroxides shows strong pH dependence and is believed to occur via surface complexation reactions (Linklater et al., 2003). Surface complexation reactions are likely to contribute to nickel sorption onto clay minerals; most of them have some component of pH-dependent surface charge related to exposed silanol (Si-OH) and

SCK•CEN/12201513 Page 104 of 208 Compilation of Technical Notes on less studied elements

aluminol (Al-OH) sites on the mineral surface; however, ion exchange may also play a role (Linklater et al., 2003). A wide range in reported sorption coefficients for rocks and soil suggests that nickel sorption is very sensitive to changes in the rock/water chemistry (Linklater et al., 2003).

A large body of work has been produced on the sorption of nickel (and other metals) to clays such as montmorillonite (Bradbury and Baeyens, 1997a, 1997b; 1999; 2005; 2011) and illite (Bradbury and Baeyens, 2009a, 2009b), the mechanisms by which sorption occurs to clay ‘edge sites’ have been explored (e.g. Dähn et al., 2003).

For the Opalinus Clay assessment, Bradbury and Baeyens (2003a) give an in situ sorption value of 2.3×10-1 m3/kg for MX-80 bentonite. At the bounding pH values of 6.9 and 7.9, in situ sorption values of 1.4×10-1 m3/kg and 5.8×10-1 m3/kg respectively, were chosen. The data given by Bradbury and Baeyens (2003a) were used to derive in situ Kd values for the near-field in the TURVA-2012 assessment. In the TURVA-2012 assessment, Kd values range from 0.11 to 3.15 m3/kg for the different reference water compositions (Wersin et al., 2014a).

Bradbury and Baeyens (2003b) recommend a far field in situ sorption value for nickel of 9.3×10-1 m3/kg for the reference system at pH 7.24. At the bounding pH values of 6.3 and 7.8, sorption values of 2.9×10-1 m3/kg and 1.9 m3/kg, respectively, were chosen. Bradbury et al. -1 3 (2010) give Rd values in the range of 1.5×10 to 3.2 m /kg for ‘generic’ Swiss argillaceous rocks.

3 In the Belgian SAFIR 2 assessment, a Kd value of 0.008 m /kg was used for nickel (data set 2 value, converted from retardation factor of 50 by Wersin and Schwyn, 2004, from ONDRAF/NIRAS, 2001). The association of nickel with organic matter under Boom Clay conditions has been acknowledged by Bruggeman and Maes (2016).

For the TURVA-2012 assessment, Hakanen et al. (2012) reviewed data on nickel sorption to clay minerals, in particular the data on sorption of nickel to kaolinite given by Puukko and Hakanen (1998). The experiments by Puukko and Hakanen (1998) indicate a strong dependence of sorption on pH and ionic strength. Kd values for Ni sorption to kaolinite in the presence of different water compositions given by Hakanen et al. (2014) are summarised in Table 19. A summary of Kd values for nickel sorption to illite in the presence of different water compositions is given in Table 20.

3 Table 19: Calculated Kd values (m /kg) for nickel sorption to kaolinite KGa-1b under Finnish reference water conditions (from Hakanen et al., 2014).

SCK•CEN/12201513 Page 105 of 208 Compilation of Technical Notes on less studied elements

3 Table 20: Calculated Kd values (m /kg) for nickel sorption to illite-IMt1 under Finnish reference water conditions (from Hakanen et al., 2014).

4.1.4.4 Transport and diffusion When the dissolved nickel concentration is set by Ni(OH)2(beta) solubility under Boom Clay conditions, speciation calculations predict that 98.9% of the dissolved nickel will be present 2- + 2+ as Ni(CO3)2 , with very small amounts of NiCO3(aq) Ni(HCO3) and Ni (Salah and Wang, 2014). Therefore, the assumption could be made that dissolved nickel will be an anionic species in the Boom Clay.

-11 2 For the Opalinus Clay, nickel was assigned ‘non anionic’ Deff values of: 1×10 m /s (reference value perpendicular to bedding); 1×10-10 m2/s (‘upper pessimistic’ value perpendicular to bedding); and 5×10-11 m2/s (reference value parallel to bedding). A value of 0.12 was assigned to η (Bradbury and Baeyens, 2003b).

-11 In the TURVA-2012 assessment, nickel was assumed to have Deff and η values of 9×10 m2/s and 0.38 for bentonite backfill, assuming behaviour similar to that of HTO (Wersin et al., 2014b).

4.1.4.5 Justifications Based on speciation calculations with MOLDATA, the dominant species under BC conditions 2- would be an anionic Ni(CO3)2 species, while using MINTEQ predicts the neutral NiCO3(aq) species being the major complex. With respect to sorption, the data are quite scattered. Whereas sorption experiments performed within the frame of the CATCLAY project reveal strong sorption for Ni with logKd-values ranging between 2-4. Also the literature review by QUINTESSA supports this range of values for different substrates and conditions. Main argument to associate Ni to the group of transition metals was the observed interaction of Ni with organic matter in the sorption experiments. As discussed above, the presence of DOM seems to decrease the sorption compared to experiments performed in absence of DOM, but beyond a cetain threshold value of Ni-concentration, DOM may also increase the sorption (most likely through formation of binary Ni-HA and/or ternary Ni-HA-CO3 complexes). A clear mechanistic view is currently however lacking. Unfortunately, no in-house migration/diffusion data are available and colloid mediated transport is currently only assumed to be the main process for the Ni-transport through BC. In order to get a clearer view on the influence of

SCK•CEN/12201513 Page 106 of 208 Compilation of Technical Notes on less studied elements

OM on the retention and migration behaviour of nickel, it would be very beneficial to perform also diffusion experiments in future.

4.1.4.6 References Berner, U. (2002) Project Opalinus Clay: Radionuclide concentrations limits in the near-field of a repository for spent fuel and vitrified high-level waste. Nagra Technical Report NTB 02-10, Wettingen, Switzerland.

Bethke, C.M. (2008) Geochemical and Biogeochemical Reaction Modeling. Cambridge University Press.

Bradbury M. H. and Baeyens B. (1997a) A mechanistic description of Ni and Zn sorption on Na- montmorillonite. Part I. Titration and sorption measurements. Journal of Contaminant Hydrology, 27, 199-222.

Bradbury M. H. and Baeyens B. (1997b) A mechanistic description of Ni and Zn sorption on Na- montmorillonite. Part II: Modelling. Journal of Contaminant Hydrology, 27, 223–248.

Bradbury M. H. and Baeyens B. (1999) Modelling the sorption of Zn and Ni on Ca-montmorillonite. Geochimica et Cosmochimica Acta, 63, 325–336.

Bradbury, M.H. and Baeyens, B. (2003b) Far-Field Sorption Data Bases for Performance Assessment of a High-Level Radioactive Waste Repository in Opalinus Clay Host Rock. Nagra Technical Report 02-19.

Bradbury, M. H. and Baeyens, B. (2005) Modelling the sorption of Mn(II), Co(II), Ni(II), Cd(II), Eu(III), Am(III), Sn(IV), Th(IV), Np(V) and U(VI) on montmorillonite: Linear free energy relationships and estimates of surface binding constants for some selected heavy metals and actinides. Geochimica et Cosmochimica Acta, 69, 875– 892.

Bradbury, M. H. and Baeyens, B. (2009a) Sorption modelling on illite Part I: Titration measurements and the sorption of Ni, Co, Eu and Sn. Geochimica et Cosmochimica Acta, 73, 990-1003.

Bradbury, M. H. and Baeyens, B. (2009b) Sorption modelling on illite. Part II: Actinide sorption and linear free energy relationships. Geochimica et Cosmochimica Acta, 73, 1004-1013.

Bradbury, M.H. and Baeyens, B. (2011) Predictive sorption modelling of Ni(II), Co(II), Eu(IIII), Th(IV) and U(VI) on MX-80 bentonite and Opalinus Clay: A “bottom-up” approach. Applied Clay Science 52, 27-33.

Bradbury, M.H., Baeyens, B. and Thoenen, T. (2010) Sorption Data Bases for Generic Swiss Argillaceous Rock Systems. Nagra Technical Report 09-03.

Bruggeman, C. and Maes, N. (2016) Radionuclide migration and retention in Boom Clay. External Report, SCK•CEN-ER-0345, SCK•CEN, Mol, Belgium.

Bruno J., Duro L. and Grivé M. (2001) The applicability and limitations of the geochemical models and tools used in simulating radionuclide behaviour in natural waters. Lessons learned from the Blind predictive Modelling exercises performed in conjunction with Natural Analogue studies. SKB Technical Report, TR-01-20.

SCK•CEN/12201513 Page 107 of 208 Compilation of Technical Notes on less studied elements

Dähn, R., Scheidegger, A.M., Manceau, A., Schlegel, M.L., Baeyens, B., Bradbury, M., Chateigner, D. (2003) Structural evidence for the sorption of Ni(II) atoms on the edges of montmorillonite clay minerals: a polarized X-ray absorption fine structure study. Geochimica et Cosmochimica Acta, 67, 1-15.

Hakanen, M., Ervanne, H., Puuko, E. (2014) Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto. Radionuclide Migration Parameters for the Geosphere. Posiva, 2012-41.

Hummel W., Berner U., Curti E., Pearson F.J. and Thoenen T. (2002) NAGRA/PSI Chemical Thermodynamic DataBase 01/01. NAGRA Technical Report 02-16.

Linklater, C.M., Moreton, A.D. and Tweed, C.J. (2003) Analysis and interpretation of geosphere sorption data for a Nirex performance assessment. UK Nirex Report N/083. Harwell, United Kingdom.

ONDRAF/NIRAS (2001) SAFIR 2 Safety Assessment and Feasibility Interim Report 2.

Puukko, E. and Hakanen, M. (1998) Characterisation of kaolinite and the sorption of nickel on kaolinite. Posiva Oy, Helsinki, Finland. Posiva Working report 98-60.

Salah, S. and Wang, L. (2014) Speciation and solubility calculations for waste relevant radionuclides in Boom Clay. First Full Draft. EXTERNAL REPORT SCK•CEN-ER-19814/Ssa/P-16.

Wersin, P. and Schwyn, B. (2004) Project Opalinus Clay. Integrated Approach for the Development of Geochemical Databases Used for Safety Assessment. Nagra Technical Report 03-06.

Winter, M. (2014). Webelements. University of Sheffield and Webelements Ltd. http://www.webelements.com/nickel/

SCK•CEN/12201513 Page 108 of 208 Compilation of Technical Notes on less studied elements

4.1.5 Technical note for Palladium (Pd)

4.1.5.1 General Palladium belongs to the group of platinum (Pt) metals (besides rhodium, ruthenium, iridium and osmium) and its atomic number is 46. The most stable and waste relevant isotope is 107Pd with a half-life of 6.5 million years. It is produced during nuclear fission, but only in low percentages. Due to its low decay energy (30 KeV), Pd does not represent a very hazardous isotope. Pd forms rare minerals, such as cooperite [(Pt,Pd,Ni)S] and polarite [(Pd)Bi,Pb),cr]. Additionally, Pd sulphide and arsenide minerals are known (Hummel et al., 2002). Examples are Vysotskite (PdS,cr), PdS2(s) and Pd4S(s), the latter being high-T solids (Hummel et al., 2002).

4.1.5.2 Speciation and solubility Palladium occurs in 3 oxidation states, i.e. 0, +2 and +4. The hydrolysis of the Pd2+ ion is reported to start at very low pH values, i.e. pH ~0.7 (Hummel et al., 2002). Besides this, Pd has a tendency to form polynuclear species and colloids, due to which the hydrolysis behaviour of Pd is controversely discussed.

-8 Speciation calculation performed with MOLDATA at [Pd] = 10 show that, Pd(OH)2(aq) is the dominant species under BC conditions (Figure 46). The same speciation is predicted by calculations using the ANDRA or NAGRA/PSI TDB, respectively. It should be mentioned that the aqueous species PdO(aq) was suppressed in the calculations, as the reaction constant was found to be wrong due to erroneous data in the LLNL TDB from which this species was copied to MOLDATA (Wang, pers. communication).

Solubility calculations were performed for Pd(IV) oxide and hydroxide, as well as for elemental Pd and Pd sulphide. The results are summarized in Table 21. Elemental Pd and PdS(s) are characterized by an extremely low solubility under BC conditions. According to ANDRA (Dossier 5), these phases represent quite unprobable solubility limiting phases and it seems more reasonable to retain the oxide and hydroxide phases in SA evaluations.

.5 PdCl2

Pd(OH)2 Eh (volts) Eh 0 --- Pd(OH)3 µ 4

–.5

25°C 0 2 4 6 8 10 12 14 pH Figure 46: Eh-pH diagram of palladium (Pd-C-S-O-H) for the BC reference porewater system. Assumed activity of dissolved [Pd] = 10-8. Database: MOLDATA_R2. Code: The Geochemist's Workbench - 10.0.

SCK•CEN/12201513 Page 109 of 208 Compilation of Technical Notes on less studied elements

Table 21: Solubility of Pd in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV. Database: MOLDATA. Code: The Geochemist's Workbench- 8.08.

Solubility controlling phases Solubility, [Pd], mol/L

-6 Pd(OH)2 (s) 4.0 × 10 PdO (s) 1.5 × 10-10 Pd (cr) 2.2 × 10-30 (insoluble) PdS (s) 7.9 × 10-35 (insoluble)

Source data: all solid phase data were taken from the ANDRA TDB

The reaction constants for the Pd solids comprised in Table 21 and MOLDATA are the following:

+ 2+ Pd(OH)2(s) + 2 H ↔ Pd + 2 H2O log K = -1.61 + 2+ PdO(s) + 2 H ↔ Pd + H2O log K = -6.02 + 2+ Pd(cr) + 0.5 O2(aq) +2 H ↔ Pd + H2O log K = 9.96 2+ 2- PdS(s) + 2 O2(aq) ↔ Pd + SO4 log K = 91.42

Based on the solubility and speciation calculations, the following precipitation/formation sequence could be anticipated:

At [Pd] = 10-8:

1) PdO(s) kinetically favoured phase → 2) Pd(cr) → 3) PdS(s) most stable phase and at [Pd] = 10-5:

1) Pd(OH)2(s) kinetically favoured phase → 2) PdO(s) → 3) Pd(cr)

→ 4) PdS(s) most stable phase

4.1.5.3 Sorption and retardation Sako et al. (2009) investigated the sorption of Pd, Rh, and Pt in sediment and soil samples. Sorption isotherms, time-dependent adsorption and surface complexation modeling studies were used to investigate the post-depositional mobility of three of the platinum group- elements (Pd, Rh, and Pt) in semi-arid soil and sediment samples with varying surface properties. The acidity constants (LogK(a1) and LogK(a2)), optimized from batch titration data, ranged from 4.69 to 5.34 for LogK(a1) and from -6.51 to -7.61 for Log K(a2), suggesting the occurrence of both protonation and deprotonation reactions on the solid surfaces. Partition coefficients and removal rates of the metals had a general trend of Pd > Pt > Rh. The sediment sample, with the highest clay content and exchangeable cation concentrations, also had the highest affinity for the metals. The times required for sediment to adsorb 63% of the metals were 2.63h, 4.08h and 10.64h for Pd, Pt and Rh, respectively. The FITEQL program successfully optimized the conditional binding constants of the metals on the solids from batch adsorption data. The constants decreased in the order of Pd > Rh > Pt, which was consistent with the observed high affinity of the solids for Pd. The modeling results also

SCK•CEN/12201513 Page 110 of 208 Compilation of Technical Notes on less studied elements

showed that aqueous Pd was the least sensitive to pH followed by Rh and Pt. However, metal adsorption below the points of zero net proton charges (ca. pH 6.7) is attributable to the involvement of permanent negatively charged binding sites in the adsorption process. Notably, partition coefficients, removal rates and conditional binding constants all showed a high affinity of Pd for the solids. A similarity between the model outputs and the batch adsorption data indicates the suitability of the model for describing the mobility and retention of the three metals in semi-arid soils and sediments.

In terms of chemical analogy, Ochs et al. (2011) suggested to use Ni and Pb to estimate the sorption value of Pd when experimental sorption data of Pd are lacking.

4.1.5.4 Transport and diffusion No transport/diffusion data are available for Pd.

4.1.5.5 Justification According to the speciation calculations, the neutral Pd(OH)2(aq) represents the dominant aqueous species under BC conditions. As revealed by the Pourbaix diagram, the onset of hydrolysis starts at very low pH and the neutral 1,2 hydrolysis species shows a large stability field. Therefore, also sorption of Pa is expected to be high, which has been confirmed by the literature data. The affinity of Pd on soil and sediment samples was reported to be high with cation exchange and surface complexation playing a role in the retention. The association of Pd to the group of transition metals was mainly based on the strong hydrolysis and reported sorption behaviour. No information is currently available on the transport behaviour of Pd and interaction with OM. Getting further insight into the latter would therefore be desirable.

4.1.5.6 References Hummel W., Berner U., Curti E., Pearson F.J. and Thoenen T. (2002): NAGRA/PSI Chemical Thermodynamic DataBase 01/01. NAGRA Technical Report 02-16. Universal Publishers, Parkland, Florida.

Ochs, M. Vieille-Petit, L., Wang, L. and Mallants, D. 2011 Additional sorption parameters for the cementitious barriers of a near-surface repository, Project near-surface disposal of category A waste at Dessel, NIROND–TR 2010–06 E V1, 9 March 2011.

Sako, A. Lopes, L and Roychoudhury, A. N. 2009, Adsorption and surface complexation modeling of palladium, rhodium and platinum in surficial semi-arid soils and sediment. Applied Geochemistry, 24, 86-95.

Tagirov et al. (2013): The speciation and transport of palladium in hydrothermal fluids: Experimental modeling and thermodynamic constraints.

SCK•CEN/12201513 Page 111 of 208 Compilation of Technical Notes on less studied elements

4.1.6 Technical note for Tin (Sn)

4.1.6.1 General Tin (Sn) is a silvery-white metal that is malleable, somewhat ductile, and has a highly crystalline structure. It has an atomic number of 50 and is in Group 14 (Winter, 2014). Ten stable isotopes are known with atomic masses ranging between 112 and 124. A number of tin compounds exist, common oxidation states are (IV) and (II) and in nature it is found as the ore mineral cassiterite (Sn(IV)O2) (Winter, 2014). Tin also forms solid solutions with various ferromagnesian silicates such as pyroxenes, micas and amphiboles due to the substitution of tin (ionic radius 0.71Å) for titanium (ionic radius 0.68 Å) and Fe(III) (ionic radius 0.64Å). Tin concentrations in biotite and muscovite may be as high as 1250 ppm (Duro et al., 2006). Tin also occurs in sulphides such as stannite (Cu2FeSnS4), canfieldite (Ag8SnS6) and teallite -9 3 (PbSnS2) (Duro et al., 2006). Measured dissolved Sn concentrations are often ~10 mol/dm in groundwaters and seawater (Duro et al., 2006).

4.1.6.2 Speciation and solubility Tin oxide or hydroxide (SnO, Sn(OH)4) with different degrees of crystallinity has been assumed to act as solubility-limiting phases for tin (e.g. Berner, 2002; Linklater et al., 2003; Wersin et al., 2014a; Hakanen et al., 2014) and at pH 11 and higher, the solubility has been considered to be controlled by CaSn(OH)6(s) (Hakanen et al., 2014).

Calculations performed by Salah and Wang (2014) with MOLDATA (and the ANDRA ThermoChimie database) reveal that the aqueous speciation of tin under Boom Clay - conditions is dominated by Sn(OH)5 (Figure 47 b). The species distribution in equilibrium - 2- with SnO2(am) was calculated to correspond to 73% Sn(OH)5 , 26% Sn(OH)4(aq) and 1% Sn(OH)6 (Salah and Wang, 2014). Duro et al. (2006) found similar speciation in their calculations for - the SR-Can assessment, with Sn(OH)4(aq) dominating up to ~pH 8, with Sn(OH)5 from ~pH 8- 2- 10, then Sn(OH)6 from pH 10 upwards. Under Boom Clay conditions, metastable SnO2(am) has a calculated solubility of 1.9×10-7 mol/dm3, whereas cassiterite has a solubility of 3.4×10-8 mol/dm3 (Salah and Wang, 2014).

a) b) 1 1

+ SnCl3 + .5 .5 Sn(OH)3

Sn(OH)4 Sn(OH)4(aq) - Eh (volts) Eh Eh (volts) Eh Sn(OH) 0 Sn++ 0 ++ 5 Sn -- Sn(OH)6 µ µ

–.5 –.5

25°C 25°C 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 pH pH

SCK•CEN/12201513 Page 112 of 208 Compilation of Technical Notes on less studied elements

c) d) 1 1

++ .5 Sn .5 SnOH+

++ Sn(OH)2 Sn Eh (volts) Eh Eh (volts) Eh 0 0 - Sn(OH)3 µ µ

–.5 –.5

25°C 25°C 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 pH pH

e) f) 1 1

+ + SnCl SnCl3 + 3 Sn(OH)3 + .5 .5 Sn(OH)3

Cassiterite(tetragonal) Cassiterite(tetragonal) Eh (volts) Eh

Eh (volts) Eh - Sn(OH) ++ - 0 ++ 5 0 Sn Sn(OH)5 Sn -- -- Sn(OH)6 Sn(OH)6 SnS(s) µ SnS(s) µ –.5 –.5 25°C 25°C 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 pH pH

g) 1

SnCl+ Sn(OH)3 + .5 3

SnO2(am) - no graph - Eh (volts) Eh - 0 Sn++ Sn(OH)5 -- Sn(OH)6 Ottemannite SnS(s) µ

–.5

25°C 0 2 4 6 8 10 12 14 pH

Figure 47: Eh-pH diagram of tin (Sn-C-S-O-H) for the BC reference porewater system. Assumed activity of dissolved [Sn] = 10-8. Diagram a) LLNL TDB, b) ANDRA (ThermoChimie) and MOLDATA TDB, c) NAGRA/PSI TDB, d) NEA TDB (not meaningful, as Sn2+ is the only aqueous species), e) MOLDATA TDB solid phases included, f) same diagram as e, but [Sn] = 10-6, g) same diagram as f, but cassiterite suppressed in calculation. Code: The Geochemist's Workbench (from Salah and Wang, 2014).

4.1.6.3 Sorption and retardation 4 3 In their review, Linklater et al. (2003) found Rd values for tin in the range 200 - 10 cm /g in sedimentary rocks and soils. Linklater et al. (2003) note the lack of data on sorption

SCK•CEN/12201513 Page 113 of 208 Compilation of Technical Notes on less studied elements

mechanisms but consider surface complexation to occur, with the possibility that tin could take part in ion exchange reactions with calcium.

The sorption of tin on kaolinite (KGa-1b) is described by Hakanen et al. (2014) for fresh ALLMR (fresh mildly reducing granitic reference water), glacial OLGA (synthetic glacial anoxic meltwater) and OLGO (synthetic glacial oxic) water compositions and on illite (IMt-1) and on chlorite (CCa-2) in ALLMR water (for the TURVA-2012 assessment). The sorption of tin on IMt-1 illite was determined in fresh ALLMR water at pH 7–10.4. For pH 7–9, the lowest reported Rd value is 115 m3/kg and most values are higher than 150 m3/kg (Hakanen et al., 2014). The best estimate Rd value is proposed to be 150 m3/kg for all waters (Hakanen et al., 2014). The sorption of tin on chlorite (ripidolite CCa-2) from the Clay Mineral Society was determined in fresh ALLMR water and it was found that Rd values decrease as the pH increases in a linear fashion (Hakanen et al., 2014). Hakanen et al. (2014) give reference Rd values (all reference water compositions) of 1×103 m3/kg for kaolinite, 1.5×102 m3/kg for illite (all reference water compositons) and a range of 6.1×101 to 2.3×102 m3/kg for sorption to chlorite for the different reference waters.

3 For the Opalinus Clay, Bradbury and Baeyens (2003a) give a Sn(IV) Rd value of 810 m /kg for sorption to MX-80 bentonite in the near field at pH 7.25. For Opalinus Clay itself, Bradbury 3 and Baeyens (2003b) give a Sn(IV) Rd value of 110 m /kg, which is taken from Lauber et al. (2000).

For ‘generic’ Swiss argillaceous rocks, Bradbury et al. (2010) give Rd values in the range 4.6 (low ionic strength, pH 9, ‘Argl9’ composition) to 70 m3/kg (pH 6, ‘low’ and ‘high’ ionic strength ‘Argh6’ and ‘Argl6’ compositions).

Kedziorek et al. (2007) describe experiments on the adsorption of Sn(IV) by MX-80 bentonite and Callovo-Oxfordian claystone and found that within the range of experimental conditions, sorption was independent of suspension pH, and of the nature of both the sorbent and the electrolyte. Kedziorek et al. (2007) suggest that the sorbate in the experiments was Sn(OH)4, and that the clays present (most likely smectite minerals) in both solids (bentonite and claystone) probably control adsorption.

4.1.6.4 Diffusion and transport Calculations performed by Salah and Wang (2014) with MOLDATA (and the ANDRA database) reveal that the aqueous speciation of tin under Boom Clay conditions is dominated - by Sn(OH)5 (Figure 47 b).

-11 2 For the Opalinus Clay, tin was assigned ‘non anionic’ Deff values of: 1×10 m /s (reference value perpendicular to bedding); 1×10-10 m2/s (‘upper pessimistic’ value perpendicular to bedding); and 5×10-11 m2/s (reference value parallel to bedding). A value of 0.12 was assigned to η (Bradbury and Baeyens, 2003b).

-11 2 In the TURVA-2012 assessment, tin was assumed to have Deff and η values of 9×10 m /s and 0.38 for bentonite backfill, assuming behaviour similar to that of HTO (Wersin et al., 2014b).

SCK•CEN/12201513 Page 114 of 208 Compilation of Technical Notes on less studied elements

4.1.6.5 Justification As for palladium, the association of Sn to the trace metal group is mainly based on the fact that tin represents – based on the speciation calculations with MOLDATA - also a strongly hydrolyzing species with hydrolysis starting already at very low pH. In contrast to Pd, Sn(IV) is - predicted to be present in the porewater as a negatively charged species, i.e. Sn(OH)5 under undisturbed BC conditions. As such, sorption/retardation at pH > 8 is expected to be rather low/limited. In this context, it should however be mentioned that the speciation might be quite different depending on which TDB is used for the calculations. Based on the NAGRA/PSI database, Sn would occur as neutral (SnII) species, i.e. Sn(OH)2(aq) in the pH range 4-10, but according to the LLNL TDB, Sn(IV) is predicted to be the prevalent species as Sn(OH)4(aq), and using NEA data Sn would prevail as divalent cation (Sn2+) over the entire pH range. Unfortunately, no in-house sorption, neither transport data are available. The literature data wrt sorption reported by QUINTESSA reveal generally high sorption for Sn,with Kd-values ranging between ~60-104 m3/kg, but they are difficult to evaluate as the exact experimental conditons (pH, Eh, ionic strength) are only partly or not given at all by the authors. From mechanistic point of view, surface complexation seems to be the dominant process, although also cation exchange processes (e.g. between tin and calcium) have been reported to be possible. Although also for Sn(IV) DOM linked transport has been put forward in analogy to Tc(IV), no experimental evidence exists up to now to support this hypothesis. More work on Sn is surely needed to get a better insight into its migration/retention, but also redox behaviour.

4.1.6.6 References Berner, U. (2002) Project Opalinus Clay: Radionuclide concentrations limits in the near-field of a repository for spent fuel and vitrified high-level waste. Nagra Technical Report NTB 02-10, Wettingen, Switzerland.

Bradbury, M.H. and Baeyens, B. (2003b) Far-Field Sorption Data Bases for Performance Assessment of a High-Level Radioactive Waste Repository in Opalinus Clay Host Rock. Nagra Technical Report 02-19.

Bradbury, M.H., Baeyens, B. and Thoenen, T. (2010) Sorption Data Bases for Generic Swiss Argillaceous Rock Systems. Nagra Technical Report 09-03.

Hakanen, M., Ervanne, H., Puuko, E. (2014) Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto. Radionuclide Migration Parameters for the Geosphere. Posiva, 2012-41.

Kedziorek, M.A.M., Bourg, A.C.M. and Giffaut, E. (2007) Hydrogeochemistry of Sn(IV) in the context of radioactive waste disposal: Solubility and adsorption on MX-80 bentonite and Callovo-Oxfordian argillite. Physics and Chemistry of the Earth, Parts A/B/C, 32, Issues 8–14, 568–572.

Lauber, M., Baeyens, B., Bradbury, M.H. (2000) Physico-chemical characterisation and sorption measurements of Cs, Sr, Ni, Eu, Th, Sn and Se on Opalinus clay from Mont Terri. PSI Bericht Nr. 00-10,

SCK•CEN/12201513 Page 115 of 208 Compilation of Technical Notes on less studied elements

Paul Scherrer Institut, Villigen, Switzerland and Nagra Technical Report NTB 00-11, Nagra, Wettingen, Switzerland.

Linklater, C.M., Moreton, A.D. and Tweed, C.J. (2003) Analysis and interpretation of geosphere sorption data for a Nirex performance assessment. UK Nirex Report N/083. Harwell, United Kingdom.

Salah, S. and Wang, L. (2014) Speciation and solubility calculations for waste relevant radionuclides in Boom Clay. First Full Draft. EXTERNAL REPORT SCK•CEN-ER-19814/Ssa/P-16.

Winter, M. (2014) Webelements. University of Sheffield and Webelements Ltd. http://www.webelements.com/tin/

SCK•CEN/12201513 Page 116 of 208 Compilation of Technical Notes on less studied elements

4.1.7 Technical note for Zirconium (Zr)

4.1.7.1 General Zirconium (Zr) is a Group 5 metallic element, with an atomic number of 40 (Winter, 2014). Zirconium predominantly occurs as Zr(IV) (Duro et al., 2006; Wersin et al., 2014a). Zircon (ZrSiO4) is the main ore mineral for zirconium and a number of zirconium-bearing minerals exist (Winter, 2014).

4.1.7.2 Speciation and solubility Under pH values of 6-10, Zr(OH)4(aq) is often the dominant aqueous species (Wersin et al., - 2014a). The formation of Zr-carbonate complexes only influences solubility at HCO3 concentrations above 5 mmol/dm3 (Wersin et al, 2014a). Duro et al. (2006) give a zirconium -9 3 concentration of 9.7×10 mol/dm (Zr(OH)4 solubility) for the SR-Can reference water under different redox conditions (at pH 7). Wersin et al. (2014a) give zirconium concentrations of 1.7×10-8 mol/dm3 and 1.8×10-8 mol/dm3 for saline (pH = 7.2) and brackish (pH 7.12) reference water compositions, respectively. Hakanen et al. (2014) cite the measured and modelled data on Zr-hydroxide solubility by Sasaki et al. (2006, 2008), who found that the -14 -9 3 solubility of ZrO2(OH)4 is 1×10 to 9×10 mol/dm under carbonate free conditions. Hakanen et al. (2014) cite Pouchon et al. (2001) who report an increase in solubility from 10- 10 -8 3 -6 -5 3 3 -10 mol/dm in carbonate-free water to 10 -10 mol/dm in 50 mmol/dm NaHCO3 solution.

Zirconium hydroxide (Zr(OH)4) has been considered to be a solubility-limiting phase for Zirconium in High Level Waste(HLW)/Spent Fuel (SF) repositories with bentonite buffers in hard rock (Duro et al., 2006; Wersin et al., 2014a), with more soluble amorphous forms being considered to be conservative (Wersin et al., 2014a). ZrO2 has also been considered as a potential solubility-limiting phase in claystone (Berner, 2002).

Under Boom Clay conditions, Salah and Wang (2014) show that Zr(OH)4(aq) is the likely dominant zirconium species using speciation calculations undertaken with the LLNL, ANDRA, NEA and MOLDATA databases (Figure 48a, b, d and e). In contrast, using the NAGRA/PSI TDB, - Zr(OH)5 is predicted to be the prevalent aqueous species (Figure 48c) in the near-neutral and - alkaline pH range. Note that Berner (2002) identified Zr(OH)5 as the dominant zirconium species in Opalinus Clay porewater (pH 7.25, log pCO2 = -2.2) using the NAGRA/PSI TDB.

Salah and Wang (2014) note that zirconium has a strong tendency to form polymeric species and colloids (Baes and Mesmer, 1976; Brown et al., 2005) and Zr-NOM colloid association has been put forward in the newly developed phenomenological model (Bruggeman et al., in prep.).

Salah and Wang (2014) identify Zr(OH)4 as a potential solubility-limiting phase, and calculate solubility limits for both fresh and aged amorphous Zr(OH)4 under Boom Clay conditions (1.11×10-4 mol/dm3 and 1.82×10-8 mol/dm3, respectively). Baddeleyite has a much lower solubility (6.43×10-10 mol/dm3) as has zircon (8.56×10-14 mol/dm3). Note that for the SAFIR-2 assessment, the ‘best estimate’ value was 1×10-6 mol/dm3 (range from 10-9 to 10-3 mol/dm3, ONDRAF/NIRAS, 2001).

SCK•CEN/12201513 Page 117 of 208 Compilation of Technical Notes on less studied elements

b) a) 1

Zr(OH)+++ .5

Zr(OH)4(aq) Eh (volts) Eh 0

- Zr(OH)Zr(OH) 5 µ 6 2

–.5

25°C 0 2 4 6 8 10 12 14 pH

c) 1 d) 1

Zr(OH)++ .5 ZrOH+++ .5 2

Zr(OH)4

Zr(OH)4(aq) Eh (volts) Eh Eh (volts) Eh - 0 0 Zr(OH)5 -- Zr(OH)6 µ µ

–.5 –.5

25°C 25°C 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 pH pH

e) 1

Zr(OH)++ .5 2

- no graph - Zr(OH)4(aq) Eh (volts) Eh 0

Zr(OH)-- µ 6

–.5

25°C 0 2 4 6 8 10 12 14 pH

Figure 48: Eh-pH diagrams of zirconium (Zr-C-S-O-H) for the BC reference porewater system. Assumed activity of dissolved Zr = 10-8. Diagram a) LLNL TDB, b) ANDRA TDB c) NAGRA/PSI TDB, d) NEA TDB, e) MOLDATA Code: The Geochemist's Workbench (reproduced from Salah and Wang, 2014).

SCK•CEN/12201513 Page 118 of 208 Compilation of Technical Notes on less studied elements

Using Geochemist’s Workbench® (Bethke, 2008) and the ‘MINTEQ’ (thermo_minteq.dat) database, a solubility diagram has been calculated for zirconium (Figure 49). In agreement with the calculations presented by Salah and Wang (2014), Zr(OH)4(aq) is predicted to be the dominant aqueous species under conditions associated with the Boom Clay.

–1

–2 ZrO2 –3

–4 ++++ - -- –5 H2CO3* (aq) HCO3 CO3 ZrOH+++ –6 Zr(OH)- log a Zr log 5 –7 Zr(OH)4 (aq) –8

–9 16°C –10 0 2 4 6 8 10 12 14 pH

Figure 49: Solubility diagrams for zirconium, assuming Ca2+ activity buffered by calcite, sulphate buffered by 2+ pyrite, Fe buffered by siderite, log f CO2(g) = -2.44, log f O2(g) = -71.4 (Eh = -0.28 mV at pH 8.36, 16 °C, 1 bar), log a Cl- = -3.155. Solute activity buffers are the same as those used for model Boom Clay porewater described by Salah and Wang (2014). Diagrams generated using ‘Act2’ module of Geochemist’s Workbench® (Bethke, 2008). Upper diagram includes all solids.

4.1.7.3 Sorption and retardation Linklater et al. (2003) provide a compilation of Rd values for zirconium sorption to both rocks and sediments. In marine sediment, values range between 102-104 cm3/g.

In the Finnish ‘TURVA-2012’ assessment, thorium was used as an analogue for zirconium under near-field conditions (Wersin et al., 2014a). This results in a best estimate Kd for zirconium of 63 m3/kg (lower limit 5 m3/kg) for all in-situ porewaters (Wersin et al., 2014a). Hakanen et al. (2014) also use thorium as an analogue for zirconium sorption onto clay, Kd values for sorption to kaolinite and illite at Olkiluoto are 200 and 20 m3/kg respectively, for a range of reference water compositions.

3 A Kd value of 8 m /kg has been used by Andra for thorium sorption in the Callovo-Oxfordian 3 claystone (Andra, 2005). In the SAFIR 2 assessment, a Kd value of 0.08 m /kg was used for thorium (data set 2 value, converted from retardation factor of 500 by Wersin and Schwyn, 2004, whereas a retardation factor of 400 was given for zirconium in the SAFIR 2 assessment (ONDRAF/NIRAS, 2001).

Bradbury and Baeyens (2003a) could not identify sorption data for MX-80 bentonite and used 3 tin (which is likely to be present as a neutral hydroxy species) as an analogue (Rd = 81 m /kg). Tin was also used as an analogue to define sorption coefficients for ‘generic’ Swiss argillaceous rocks (values ranging from 0.1 to 42 m3/kg, Bradbury et al., 2010). Tin is also

SCK•CEN/12201513 Page 119 of 208 Compilation of Technical Notes on less studied elements

3 used as an analogue for Opalinus Clay (Rd = 10.9 m /kg at pH 7.24, Bradbury and Baeyens, 2003b).

4.1.7.4 Transport and diffusion Under Boom Clay conditions, Salah and Wang (2014) demonstrate that Zr(OH)4(aq) represents the dominant aqueous species, as revealed by calculations using the LLNL, ANDRA, NEA and MOLDATA TDB, respectively (Figure 48 a, b, d and e). In contrast, using the NAGRA/PSI TDB, - Zr(OH)5 is predicted to be the prevalent aqueous species (Figure 48c) in the near-neutral and alkaline pH range. These calculations therefore suggest that zirconium will be present as a neutral (or possibly negatively-charged) species under Boom Clay conditions.

-11 2 For the Opalinus Clay, zirconium was assigned ‘non-anionic’ Deff values of: 1×10 m /s (reference value perpendicular to bedding); 1×10-10 m2/s (‘upper pessimistic’ value perpendicular to bedding); and 5×10-11 m2/s (reference value parallel to bedding) (Bradbury and Baeyens, 2003b). A value of 0.12 was assigned to η (Bradbury and Baeyens, 2003b).

-11 In the TURVA-2012 assessment, zirconium was assumed to have Deff and η values of 9×10 m2/s and 0.38 for bentonite backfill, assuming behaviour similar to that of HTO (Wersin et al., 2014b).

4.1.7.5 Justification As palladium and tin, zirconium is strongly hydrolyzing and predicted to occur as neutral Zr(OH)4(aq) species under BC conditions. Sorption/retardation is therefore assumed to be high. According to the literature review of QUINTESSA, the reported Kd- and Rd-values for different clays range between 0.08-200 m3/kg. Unfortunately, no in-house data are available concerning the sorption affinity of Zr onto Boom Clay. Also the knowledge on its transport behavior is limited. Due to the fact, that also the formation of polymeric species has been described by different authors, colloid – colloid association between Zr(IV) and dissolved organic matter (DOM) – in analogy to Tc(IV) – has been put forward. Due to the aforementioned reasoning, Zr was associated to the group of transition metals. Providing support for the grouping by performing experiments under BC representative conditions is however considered to be necessary.

4.1.7.6 References Andra (2005) Dossier Argile Safety Evaluation of a Geological Repository. Andra, France

Baes, C. F. Jr. and Mesmer, R. E. (1976) The hydrolysis of cations. John Wiley and Sons, New York.

Bethke, C.M. (2008) Geochemical and Biogeochemical Reaction Modeling. Cambridge University Press.

SCK•CEN/12201513 Page 120 of 208 Compilation of Technical Notes on less studied elements

Bradbury, M.H. and Baeyens, B. (2003b) Far-Field Sorption Data Bases for Performance Assessment of a High-Level Radioactive Waste Repository in Opalinus Clay Host Rock. Nagra Technical Report 02-19.

Bradbury, M.H., Baeyens, B. and Thoenen, T. (2010) Sorption Data Bases for Generic Swiss Argillaceous Rock Systems. Nagra Technical Report 09-03.

Brown, P.L., Curti, E., and Grambow, B. (2005) Chemical Thermodynamics of Zirconium. Vol. 8, OECD NEA, Databank. Elsevier.

Bruggeman, C. and Maes, N. (2016) Radionuclide migration and retention in Boom Clay. External Report, SCK•CEN-ER-0345, SCK•CEN, Mol, Belgium.

Hakanen, M., Ervanne, H., Puuko, E. (2014) Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto. Radionuclide Migration Parameters for the Geosphere. Posiva, 2012-41.

Linklater, C.M., Moreton, A.D. and Tweed, C.J. (2003) Analysis and interpretation of geosphere sorption data for a Nirex performance assessment. UK Nirex Report N/083. Harwell, United Kingdom.

ONDRAF/NIRAS (2001) SAFIR 2 Safety Assessment and Feasibility Interim Report 2.

Pouchon, M.A., Curti, E., Degueldre, C., Tobler, L. (2001) The influence of carbonate complexes on the solubility of zirconia: new experimental data. Progress in Nuclear Energy 38, 443-446.

Salah, S. and Wang, L. (2014) Speciation and solubility calculations for waste relevant radionuclides in Boom Clay. First Full Draft. EXTERNAL REPORT SCK•CEN-ER-19814/Ssa/P-16.

Sasaki, T., Kobayashi, T., Takagi, I., Moriyama, M. (2006) Solubility measurement of zirconium(IV) hydrous oxide. Radiochimica Acta 94, 489–494.

Sasaki, T., Kobayashi, T., Takagi, I., Moriyama, H. (2008) Hydrolysis Constant and Coordination Geometry of Zirconium(IV). Journal of Nuclear Science and Technology 45, 735–739.

Wersin, P. and Schwyn, B. (2014) Project Opalinus Clay. Integrated Approach for the Development of Geochemical Databases Used for Safety Assessment. Nagra Technical Report 03-06.

Winter, M. (2014). Webelements. University of Sheffield and Webelements Ltd. http://www.webelements.com/zirconium/

SCK•CEN/12201513 Page 121 of 208 Compilation of Technical Notes on less studied elements

4.2 Subgroup IVb: Trivalent lanthanides and actinides

4.2.1 Technical note for Actinium (Ac)

4.2.1.1 General Actinium (Ac) is an actinide with an atomic weight of 227 and an atomic number of 89 (CRC, 2011). When pure, Ac is a soft silver-coloured metal that has similar characteristics to rare earth elements such as lanthanum. If there is no other light, Ac glows blue as result of its intense radioactivity, which excites the air around it. The element is a strong α-emitter and β- emitter.

Only trace quantities of Ac are found naturally and are produced by the radioactive decay of 235U. Therefore, no distinct Ac minerals exist. A tonne of pitchblende (essentially a variety of UO2) contains around 150 mg of Ac.

Actinium has thirty-six isotopes all of which are radioactive (CRC, 2011). Among these isotopes the only one comprising natural Ac is 227Ac, which has the longest half-life of 21,773 years. This isotope is about a hundred and fifty times as radioactive as radium, making it valuable as a neutron source. All the remaining radioactive isotopes have half-lives of less than ten hours, and most of them have half-lives of less than a minute. 227Ac decays mainly to 227Th (18.72-day half-life), 223Ra (11.4 day half-life) and several shorter-lived isotopes including isotopes of Rn, Bi, Po and Pb.

Although available information is limited (e.g. Kitamura et al., 2010a and references therein), it appears that the chemical behaviour of Ac is similar to that of the rare earth elements (REE), especially La. However, Kitamura et al. (2010a) noted that the crystal ionic radius of Ac is larger (1.12 Å and 1.26 Å in 6-fold and 8-fold coordination respectively) than for Am (0.980 Å and 1.106 Å in 6-fold and 8-fold coordination respectively). They surmised that this difference, together with the higher radioactivity of Ac compared to Am, may explain a much larger experimental solubility (4 orders of magnitude greater) reported for Ac(OH)3(s) than for Am(OH)3(am). There is, however, considerable uncertainty about the significance of this difference as the solubility of Ac(OH)3(s) is based on only a single measurement (reported by Ziv and Shestakova, 1965).

4.2.1.2 Speciation and solubility Kitamura et al. (2010a) supported the validity of treating aqueous Am complexes as analogues of aqueous Ac species, by showing that stability constants for bromide and chloride complexes of Ac(III) are similar to (within 0.2 of) stability constants for corresponding complexes of Am. To make this comparison these researchers quoted stability constants for these Ac(III) complexes that were determined experimentally by Fukasawa et al. (1982).

There are no thermodynamic data for Ac in most of the commonly used thermodynamic databases: the NEA-TDB (Östhols and Wanner, 2000); the LLNL V8 R6 "combined" dataset, which is distributed with the Geochemist’s Workbench (GWB) software (Bethke, 1996, 2008) as “thermo.com.V8.R6; the Visual MINTEQ database, release 2.40, which is also distributed with GWB; the YMP database (USDOE, 2007); and the ThermoChimie v.7b/SIT database (Duro et al., 2006; Grivé et al., 2014). However, the thermodynamic database “JAEA-TDB” produced

SCK•CEN/12201513 Page 122 of 208 Compilation of Technical Notes on less studied elements

by the Japan Atomic Energy Agency (JAEA) does contain data for a number of Ac species and solid phases (Kitamura et al., 2010a,b):

• Ac+++; ++ + • AcOH ; Ac(OH)2 ; Ac(OH)3(aq); ++ + • AcF ; AcF2 ; ++ + • AcCl ; AcCl2 ; + - • AcSO4 ; Ac(SO4)2 ; ++ ++ ++ • AcN3 ; AcNO2 ; AcNO3 ; ++ • AcH2PO4 ; + - --- ++ • AcCO3 ; Ac(CO3)2 ; Ac(CO3)3 ; AcHCO3 ; ++ • AcSiO(OH)3 ; • AcSCN++; + - + • Organic complexes: AcOx ; AcOx2 ; AcCi(aq); AcHCi ; • Solid phases: Ac(OH)3(am); Ac2(CO3)3(am); AcCO3OH(am); AcPO4(am,hydr).

Here the notation used is the same as that in JAEA’s thermodynamic database; Ox represents oxalate; Ci represents citrate.

When constructing this database, JAEA considered, like other researchers, that Am is a good chemical analogue for Ac (Kitamura et al., 2010a). Hence, the data for the aqueous Ac species and Ac solid phases in JAEA’s thermodynamic database are those of Am. These Am data were taken mostly from the NEA-TDB (Östhols and Wanner, 2000). However, larger uncertainties (0.2 orders of magnitude) were assigned to equilibrium constants for the included Ac-species than for the Am-species used as analogues. Similarly, although solubility products for amorphous solid actinium(III) compounds (Ac(OH)3(am), AcPO4(am,hydr), Ac2(CO3)3(am) and AcCO3OH(am)) were set to the same values as those for Am(III) analogues, larger uncertainties (4 order of magnitude greater) than for the Am compounds were assigned. No data was included for crystalline Ac(III) solids owing to the supposed influence of self- irradiation by α–radiation on the solubilities of these compounds (based on the comparison between Am(OH)3(am) and the single reported solubility of Am(OH)3(c) reported by Ziv and Shestakova (1965)).

-3 Using Am as an analogue for Ac implies that in the BC porewater, AcCO3 would be the dominant aqueous species (Figure 50).

SCK•CEN/12201513 Page 123 of 208 Compilation of Technical Notes on less studied elements

Figure 50: Eh-pH diagrams of actinium (in the system Ac-C-S-Cl-F-O-H) to for the BC reference porewater from De Craen et al. (2004), calculated using Geochemist’s Workbench version 7.0 and the JAEA-TDB version 140331, and assuming an activity of dissolved Ac, [Ac] = 10-5 and a temperature of 16 ̊C. In a) all aqueous species and solid phases in the database are allowed to form; in b) solid phases are suppressed.

When defining transport parameters for use in SKB’s SR-Site assessment, Crawford (2010) - also concluded that Ac would be complexed dominantly by CO3 in the far-field groundwater at Forsmark, on the basis of treating Am and Eu as analogues for Ac. However, rather than -3 + Ac(CO3)3 , AcCO3 was considered by Crawford (2010) to be the dominant aqueous Ac- species.

Kitamura et al. (2010a) took the Gibbs free energy of formation for Ac3+ (-640.152 ± 25.104 kJ·mol-1) from Fuger and Oetting (1976). Gibbs free energies of formation of complex Ac(III) species were then estimated from this Gibbs free energy, the Gibbs free energies of formation of the other complex-forming ions, and from logK° values of the complexation reactions. However, the enthalpies, entropies and heat capacities of these species were not included in the JAEA-TDB owing to lacking data.

Performance Assessments for radioactive waste repositories have either specified no solubility control for Ac (RWMD, 2010), or else is constrained by equilibrium with a solid phase for which Am equilibrium constants are assigned (e.g. Berner, 2003; Wersin and Schwyn, 2004; Kitamura et al., 2010a).

RWMD assumed that Ac would not be solubility-limited in a reference case for a generic PA (RWMD, 2010). In contrast, for Nagra’s Project Opalinus Clay assessment, Berner (2002) and Wersin and Schwyn (2004) recommended that the same Ac solubility limits as for Am should be used for bentonite in the near field, on the grounds that Ac and Am have similar chemical characteristics. These recommended solubilities are given in Table 22. Similarly, JNC (2000) considered AcCO3OH(cr) to be a solubility-limiting phase in a fresh reducing high-pH (FRHP) reference water (ionic strength 0.02 molal, pH 8.5, Eh -2.81 V), the solubility being 2×10-7 mol/L at 25 C.̊ In this case, a crystalline phase is used to constrain Ac, in contrast to the more recent assessment of Kitamura et al. (2010a), who considered that amorphous phases should be used (see previous section).

SCK•CEN/12201513 Page 124 of 208 Compilation of Technical Notes on less studied elements

Table 22: Solubilities of Ac recommended by Berner (2002) and Wersin and Schwyn (2004) for use in NAGRA’s Project Opalinus Clay safety assessment. The reference conditions are for an Na-Cl dominated porewater with an ionic strength of 0.323, pH of 7.25 and Eh of -193.6 mV.

Porewater Reference Case Lower Limit Upper Limit Oxidising Case Value (mol/L) (mol/L) (mol/L) Value (mol/L)

ILW repository far- 2 x 10-6 2 x 10-7 2 x 10-5 field

Reference Porewater for a SF / HLW 1 x 10-6 5 x 10-8 3 x 10-5 1 x 10-6 repository far-field

Figure 51 shows a solubility diagram calculated for Ac using data from JAEA’s thermodynamic database, JAEA-TDB. This database treats Am data as analogous to Ac data, such that this solubility diagram is similar to one generated for Am using the same database; “Ac” could be replaced by “Am” in all the illustrated species and solid phases.

Figure 51: Solubility diagram for actinium (in the system Ac-C-S-Cl-F-O-H) for the BC reference porewater from De Craen et al. (2004), calculated using Geochemist’s Workbench version 7.0 and the JAEA-TDB version 140331.

4.2.1.3 Sorption and retardation As for speciation and solubility estimates, most reported PA has applied Am data to model Ac sorption (e.g. Bradbury and Baeyens, 2003a,b; Bradbury et al., 2010; Crawford, 2010; NDA, 2010). However, JNC (1999, 2000) in a reference PA case used Sm data as an analogue for Ac, 3 and accordingly selected a Kd-value of 1 m /kg for Ac sorption in bentonite buffer material. In contrast the same source used a value of 10 m3/kg for Am.

SCK•CEN/12201513 Page 125 of 208 Compilation of Technical Notes on less studied elements

Based on an analogy with Am(III), the tendency for Ac(III) is expected to be towards strong to very strong sorption where surface complexation is the uptake mechanism (Bradbury et al., 2010). It follows from this expectation that the sorption of Ac(III) would depend strongly upon the pH, and not so much on the solution’s ionic strength. However, the sorption would -3 depend on speciation and mineralogy. As noted above, AcCO3 is thought likely to be the dominant aqueous Ac specie in BC porewater over a wide range of pH, in which case sorption is unlikely to vary significantly within the BC.

Bradbury and Baeyens (2003a) identified no sorption data for Ac in the open literature. They recommended that the same Kd-values as have been determined experimentally for Am should be used for sorption in bentonite. Based on this approach, an in-situ sorption value for Ac of 26.8 m3/kg was recommended for an MX-80 bentonite reference system, with Na-Cl dominated porewater having an ionic strength in the range 0.31 to 0.35. At the bounding pH 3 3 values of 6.9 and 7.9 they recommended in-situ Kd-values of 6.6 m /kg and 63 m /kg respectively.

Similarly, using the same approach, for the far-field Opalinus Clay Bradbury and Baeyens (2003b) recommended an in situ sorption value for Ac of 17 m3/kg for the OPA reference system at pH = 7.24. At the bounding pH values of 6.3 and 7.8 sorption values of 1.2 m3/kg and 63 m3/kg respectively were recommended.

In a generic safety assessment Nirex (2003) used minimum, best estimate and maximum -3 3 1 3 4 3 values of Kd for clay-rich rocks of 1.0×10 m /kg, 5.0×10 m /kg and 1.0×10 m /kg respectively.

For use in a generic safety assessment RWMD specified Kd values for Ac sorption onto sandstone in the far field, using Am as an analogue (NDA, 2010). The best estimate Kd value was 3.2 m3/kg, with a range from 1×10-2 m3/kg to 1×103 kg/m3 . For sorption onto cement in the near-field of an ILW repository, RWMD gave a 50 percentile value of 1×10-1 kg/m3 and an upper value of 3×100 m3/kg. However, they also used the concept of Sorption Reduction Factors (SRF), a scaling factor to account for the influence of organic complexants on the sorption behaviour of radionuclides. Lower bound, upper bound and best estimates for the SRF to be applied to Ac for a cellulose loading of 10% were 2.2×10-3, 4.6×10-2, and 1.0×100 respectively. Corresponding values for a cellulose loading of 1% were 1.1×10-2, 1.3×10-1 and 1.5×100.

For the SR-Site safety assessment of SKB, Crawford (2010) recommended a best estimate for Ac sorption of 1.48×10–2 m3/kg, with lower and upper limits of 5.74×10-4 m3/kg and 3.83×10– 1 m3/kg, respectively. Combined data for Am and Eu were used to deduce these values, both these elements being considered chemical analogues of Am.

-1 3 0 IAEA (2010) quotes values of Kd for mineral soils ranging from 4.5×10 m /kg to 2.4×10 m3/kg, with a mean value of 1.2×100 m3/kg (3 samples). A single value of 5.4×100 m3/kg is also given for organic-rich soils. However, the provenance of all these values is unclear.

4.2.1.4 Diffusion and transport For the bentonite buffer and bentonite-bearing backfill proposed for use in the SF repository being planned by Posiva for construction at Olkiluoto in Finland, Wersin (2014) proposed that

SCK•CEN/12201513 Page 126 of 208 Compilation of Technical Notes on less studied elements

Deff values obtained for HTO could be used conservatively to model Am migration. Given the close chemical similarity between Am and Ac, and the fact that Am has been used by many workers as an analogue for Ac when suggesting parameters for PA calculations, it is proposed here that diffusion accessible porosity for HTO should also be used for Ac.

For bentonite-bearing backfill at 25 °C Wersin et al. (2014) proposed a best estimate value of -11 2 -10 the effective diffusion coefficient, Deff, of 9×10 m /s and a lower value of Deff of 2×10 m2/s. A corresponding effective porosity of 0.38 was specified. They pointed out that diffusion coefficients would increase with increasing temperature and proposed the following relationship to relate these parameters:

. (± . ) 2 ( ) = (0 ) , where Deff, is in m /s 𝑜𝑜 𝑜𝑜 0 026 0 03 𝑇𝑇 𝑒𝑒𝑒𝑒𝑒𝑒 𝑒𝑒𝑒𝑒𝑒𝑒 More directly relevant𝐷𝐷 𝑇𝑇to 𝐶𝐶the BC𝐷𝐷 conditions𝐶𝐶 𝑥𝑥 𝑒𝑒are measurements of HTO diffusion coefficients in mudrocks. Wenk et al. (2008) summarize data from the Callovo-Oxfordian at Bure in France, the Opalinus Clay at Mont Terri in Switzerland, and the Opalinus Clay at Benken in Switzerland. These results are given in Table 23.

Table 23: Measured diffusion coefficients for HTO in mudrocks at three different localities (m2/s). From Wenk et al. (2008).

Callovo- Opalinus Clay Opalinus Clay Oxfordian at Benken at Mont Terri at Bure Effective diffusion coefficient for 2.1 x 10-11 5.4 x 10-12 1.4 x 10-11 HTO normal to bedding (m2/s) Effective diffusion coefficient for 3.3 x 10-11 3.2 x 10-11 5.4 x 10-11 HTO parallel to bedding (m2/s)

4.2.1.5 Justification Neither in-house sorption nor migration data are available for Ac(III). Therefore, the analogy approach was applied for the grouping of Ac(III). As in other safety assessments, we also consider Am(III) as being the best suited analogue for Ac(III). With respect to speciation, it is 3- thus considered that Ac would occur as Ac(CO3)3 species, which is consistent with calculations performed by QUINTESSA using the JAEA database. Due to the predicted anionic nature of the major species, sorption should be limited/rather low. Sorption coefficients determined in-house for Am(III) are however generally high, revealing strong sorption. In presence of DOM, the Kd-values decreased however by around 1 order of magnitude, which has been referred to the complexation of Am with DOM, keeping Am in solution. Also solubility experiments using synthesized Eu(OH)3(s) showed enhanced dissolution in presence of organics leading to apparent solubilitites being few orders of magnitude higher than the thermodynamic ones. The high sorption, mainly occurring via surface complexation, as well as the interaction with organic matter was also revealed by the literature review of QUINTESSA. Based on in-house performed sequential migration experiments with Cm(III) (for details see Maes et al., 2014), DOM linked transport has been also put forward for Ac(III).

SCK•CEN/12201513 Page 127 of 208 Compilation of Technical Notes on less studied elements

4.2.1.6 References Berner U. (2002) Project Opalinus Clay: Radionuclide concentration limits in the near-field of a repository for spent fuel and vitrified high-level waste. NAGRA Technical Report NTB 02-10.

Bethke C.M. (1996) Geochemical Reaction Modeling, Concepts and Applications. Oxford University Press, 397 pp.

Bethke C.M. (2008) Geochemical and Biogeochemical Reaction Modeling. Cambridge University Press, 547 pp.

Bradbury M.H., Baeyens B. and Thoenen T. (2010) Sorption data bases for generic Swiss argillaceous rock systems. Nagra Technical Report NTB 09-03.

Bruggeman, C., Salah, S. and Maes, N. 2014. Americium retention and migration behaviour in Boom Clay. External Report, SCK•CEN-ER-201.

CRC (2011) Handbook of Chemistry and Physics, CRC Press, 92nd Edition.

Crawford J. (2010) Bedrock Kd data and uncertainty assessment for application in SR-Site geosphere transport calculations. SKB Report R-10-48.

Duro L., Cera E., Grivé M., Domènech C., Gaona X. and Bruno J. (2006) Development of the ThermoChimie thermodynamic database. Janvier 2006, Prepared by Enviros Spain S. L.

National Radiactive Waste Management Agency (ANDRA) report C.RP.0ENQ.06.0001, Châtenay- Malabry cedex, France.

Fuger J. and Oetting F.L. (1976). The chemical thermodynamics of actinoid elements and compounds Part 2: The actinoid aqueous ions. International Atomic Energy Agency.

Fukasawa T., Kawasuji I., Mitsugashira T., Sato A. and Suzuki S. (1982) Investigation of the complex formation of some lanthanoids(III) and actinoids(III) with chloride and bromide. Bulletin of the Chemical Society of Japan, 55, 726-729.

Grivé M., García D., Campos I. and Colàs E. (2014) Release of ThermoChimie Version 7c: Track changes document Project ANDRA-TDB8. ANDRA Report CCRPFSTRI400I0.

IAEA (2010). Handbook of parameter values for the prediction of radionuclide transfer in terrestrial and freshwater environments. IAEA Technical Report Series 472.

JNC (1999). Confidence in technology for the disposal of high-level radioactive waste in Japan: Development of Geological Disposal 2nd Overview Report. JNC Report TN1400 99-020. (in Japanese).

JNC (2000). H12 Project to Establish the Scientific and Technical Basis for HLW Disposal in Japan. Supporting Report 3 – Safety Assessment of the Geological Disposal System. JNC Technical Report TN1410 2000-001, Japan Nuclear Cycle Development Institute, Tokai-mura Japan.

SCK•CEN/12201513 Page 128 of 208 Compilation of Technical Notes on less studied elements

Kitamura A., Fujiwara K. and Yui M. (2010a). JAEA Thermodynamic Database for Performance Assessment of Geological Disposal of High-level and TRU Wastes: Refinement of Thermodynamic Data for Trivalent Actinoids and Samarium. JAEA-Review 2009-047.

Kitamura A., Fujiwara K., Doi R., Yoshida Y., Mihara M., Terashima M. and Yui M. (2010a). JAEA Thermodynamic database for Performance Assessment of geological diposal of high-level radioactive and TRU wastes. JAEA-Data/Code 2009-024.

Maes, N., Bruggeman, C., Govaerts, J., Martens, E., Salah, S. and Van Gompel, M. 2006. The role of natural organic matter in the migration behaviour of americium in the Boom Clay - Part I: Migration experiments. Physics and Chemistry of the Earth, 31(10-14): 541-547.

NDA (2010) Geological Disposal: Radionuclide behaviour status report. NDA Report NDA/RWMD/034.

Östhols E. and Wanner H. (2000). TDB-0: The NEA Thermochemical Database Project. OECD/NEA, Paris.

Nirex (2003) Generic repository studies: Generic post-closure Performance Assessment. Nirex Report no. N/080.

United States Department of Energy (USDOE) (2007) In-Drift Precipitates/Salts Model. ANL-EBS-MD- 000045 REV 03.

Wersin P and Schwyn B (2004) Project Opalinus Clay: Integrated approach for the development of geochemical databases used for safety assessment. Nagra Technical Report NTB-03-06.

Ziv D.M. and Shestakova I.A. (1965) Investigation of the solubility of certain actinium compounds II. Determination of the solubility and estimation of the relative basicity of actinium hydroxide. Radiokhimiya, 7, 176-186.

SCK•CEN/12201513 Page 129 of 208 Compilation of Technical Notes on less studied elements

4.2.2 Technical note for Curium (Cm)

4.2.2.1 General Curium belongs to the series of actinides, more specifically the transuranic elements and has an atomic number of 96. Fourteen isotopes of Cm are known and they are all highly radioactive. 247Cm represents the most stable isotope with a half-life of 15.6 million years. Other long-lived isotopes are 248Cm (half-life: 3.40 × 105 years), 245Cm (half-life 8.51 × 103 years), 250Cm (8.01 ×103 years) and 246Cm (4.73×103 years). Another isotope comprised in the radioactive waste inventory is 244Cm, which is an alpha-emitter with a half-life of 18 years. Curium does not occur naturally, most of the curium isotopes are produced in nuclear reactors via neutron bombardment of uranium or plutonium. In solid compounds, curium usually exhibits valence +III and sometimes +IV, while in aqueous solutions the +III valence is dominant. Due to its high fluorescence spectroscopic sensitivity (enabling detection down to trace concentration range), Cm(III) frequently features in literature studies elucidating the speciation of trivalent actinides, both in aqueous solution and adsorbed on surfaces (e.g., Wimmer et al., 1992; Fanghänel et al., 1994; Stumpf et al., 2001).

4.2.2.2 Speciation and solubility Recently Kaplan et al. (2016) performed a detailed TRLFS study in order to investigate the Cm(III) complexation with Boom Clay DOM and binding properties of trivalent radionuclides. The investigated samples were EG/BS pore water and porewater sampled in the MORPHEUS piezometer (i.e. F20: the lowest sub-surface level investigated filter). In these natural porewaters, besides the organic ligand(s) also carbonate is present. The EG/BS groundwater - (pH 8.7) contained 121 ± 2 mg/L DOC and 15 mM HCO3 as the most relevant complexing ligands and the F20 Boom Clay groundwater comprised a DOC of 88 mg/L (pH 8.3 ± 0.1) and the same inorganic carbon content as for EG/BS. Cm(III) addition was done either (1) to different DOM size fractions or (2) to the bulk solution with subsequent ultrafiltration to different size fractions. Five sets of experiments were performed, one in the absence of DOM and two sets with EG/BS and F20, respectively. For more experimentals details, it is referred to the final report of Kaplan et al. (2016). In the following only the major results are summarized.

Cm(III) carbonate system (without DOM): Based on thermodynamic calculations Cm(III) speciation in 15 mM bicarbonate solution at pH 8.7 and in the absence of DOC should be dominated by carbonate complexes. Besides the dominating 1:2 carbonate complex also minor contributions of the 1:3 and 1:1 complex should occur (Figure 52 left). The Cm(III) species distribution was confirmed in the corresponding emission spectra (Figure 52 right). The three colored lines represent the emission maxima of the three Cm(III) carbonate pure component spectra.

SCK•CEN/12201513 Page 130 of 208 Compilation of Technical Notes on less studied elements

Figure 52: Left: Calculated Cm(III) speciation in carbonate containing systems. The red line symbolizes the conditions in the investigated system. Right: Cm emission spectra in 15 mM bicarbonate solution at pH 8.7

Cm(III) in EG/BS groundwater in the presence and absence of carbonate: TRLFS spectra were also recorded for Cm(III) in the presence and absence of bicarbonate in the EG/BS groundwater at two different excitation wavelengths. Very similar results for all different size fractions (bulk, < 100 kDa, < 10 kDa and < 1 kDa) as for the pure carbonate experiments (i.e. in the absence of DOM, see above) were obtained. Figure 53 shows this exemplarily for the <100kDa size fraction. The spectra had a peak position similar to the pure bicarbonate system (i.e. in the absence of DOM, see Figure 52 right). Nevertheless, the high fluorescence intensities for the 355 nm excitation, which are even slightly higher compared to the 396.6 nm excitation, clearly point to a strong energy transfer from the DOM to Cm(III). From these findings it was concluded that Cm(III) seemed to be quantitatively (> 95%) coordinated to the organic components and that the presence of 15 mM bicarbonate had no significant effect on the emission spectra.

Figure 53: Cm(III) fluorescence spectra in the EG/BS < 100kDa size fraction in presence (left) and absence (right) of 15 mM bicarbonate at two excitation wavelengths at pH 8.7.

SCK•CEN/12201513 Page 131 of 208 Compilation of Technical Notes on less studied elements

Based on the very similar TRLFS results obtained in the presence and absence of carbonate, it was also concluded that there were also no indications for the formation of mixed ternary organic-carbonate-Cm(III) complexes as described in literature (e.g. Panak, 1996).

In Figure 54 the normalized Cm emission spectra (“indirect excitation” mode) for EG/BS groundwater, i.e. bulk and different size fractions are compiled. As can be seen, no difference is evident for the investigated size fractions. This lead to the conclusion that the different DOM size fractions do not show a distinct behavior with respect to Cm(III) complexation. As mentioned already before, the presence or absence of carbonate had no influence on the Cm emission spectra in presence of DOM, indicating that the chemical environment of Cm(III) is unaffected by carbonate if DOM is present - at least under the studied experimental conditions.

Figure 54: Compilation of the Cm(III) fluorescence spectra recorded by the “indirect excitation” method. The spectra are scaled to the same peak height.

Cm(III) in F20 groundwater in the presence and absence of carbonate: Very similar experiments and also very similar results as for EG/BS were obtained by using the F20 Boom Clay groundwater. For more details, it is also here referred to the report of Kaplan et al. (2016). All the obtained results confirmed the findings and interpretation of the EG/BS system, namely the formation of Cm(III)- DOM complexes and no significant impact of the natural bi-carbonate/carbonate content on Cm(III) complexation and speciation under the existing experimental conditions. From all spectral features no significant difference between EG/BS and F20 DOM in respect to Cm(III) complexation could be identified. This means that at least the part of DOM which is able to bind Cm(III) exhibits similar/identical behavior in both groundwater samples.

All samples were analyzed by LSC for the Cm(III) concentrations in the different DOM size fractions, both in absence and presence of carbonate. In order to compare these data with the respective DOM content in the different samples analogue experiments with inactive Eu under identical experimental conditions have been performed. The LSC analyses revealed that the main part of DOM (~ 50%) was present in the <1 kDa size fraction. Contrary to that, only ~ 2.5% of Eu(III) and ~ 4.5 % Cm(III) was detected in this smallest size fraction with the highest potential mobility. This clearly showed that the <1 kDa DOM fraction does not exhibit

SCK•CEN/12201513 Page 132 of 208 Compilation of Technical Notes on less studied elements

the strongest interaction/complexation with Ac(III) and might not completely consist of FA and HA entities.

Comparable results were obtained for the EG/BS system. Nevertheless, Cm(III) added directly to different size fractions of EG/BS groundwater was found to be completely bound to DOM even for the separated <1 kDa fraction.

Based on absolute data for Ac(III) binding, the 100-10 kDa fraction possesses the strongest Cm(III) interaction with ~ 60 % Cm(III) complexation at only 12 % contribution to the total DOC. For Eu(III) the 10-1 kDa fractions dominates with ~ 50% Eu(III) complexation at 30% DOC. The deviation between Eu(III) and Cm(III) in the medium size fractions presently could not be explained as both cations are expected to behave identical. However, the congruent behavior of Cm(III) in presence and absence of carbonate indicates the reproducibility of the experiments. Normalized to the accessible DOC content, the larger size fractions (bulk-10 kDa) show a stronger interaction with An(III) compared to the lower DOC size fractions (10-1 kDa and < 1 KDa): only 20 % of DOC (bulk-10 kDa) is binding 50 % of Eu(III) and ~85 % of Cm(III).

Curium data are not available in the present MOLDATA TDB. Although curium data are sparse, there is a consensus that curium is a chemical analogue of americium (Lumetta et al., 2006). In Stumpf et al. (2001), Cm(III) species distribution in absence of CO2(g) is depicted from pH 3 up to pH 9 (Figure 55). Kim et al. (1994) found that the stability constant for the + first carbonato complex of Cm(III) (i.e., Cm(CO3) ) compares well with values for Am(III). Vercouter et al. (2005) also found virtually the same values for the equilibrium constants of 3- 3- the aqueous complexes Cm(CO3)3 and Am(CO3)3 .

Figure 55: Species distribution of Cm(III) as a function of pH (Stumpf et al., 2001)

The solubility of Cm is taken to be identical to Am. The commonly cited solubility controlling phase in carbonate rich waters is CmOHCO3(cr), an analogue of AmOHCO3(cr) (Berner, 1995; Bruno et al., 2001).

4.2.2.3 Sorption and retardation Sorption modelling of trivalent Am(III) on illite and montmorillonite clay minerals is generally performed using the 2 site protolysis non-electrostatic surface complexation and cation

SCK•CEN/12201513 Page 133 of 208 Compilation of Technical Notes on less studied elements

exchange (2 SPNE SC/CE) model, developed by Bradbury and Baeyens (Bradbury and Baeyens, 1997; Bradbury and Baeyens, 2009a). Using these models, it is suggested that sorption of Am(III) on Boom Clay solid phase can be reasonably predicted (Bruggeman et al., 2012). Since hydrolysis constants of Am(III) and Cm(III) are assumed to be quite comparable, also the site binding constants on montmorillonite (Bradbury and Baeyens, 2005) and on illite (Bradbury and Baeyens, 2009b) will be very similar, given the correlations (linear free energy relationships, LFER) between the two constants.

Sorption of Am(III) on conditioned Silver Hill Illite (sorption edge) and Boom Clay (sorption isotherms) was extensively studied and modelled by Salah et al. (2007, 2009) and the results reported in detail in Maes et al. (2009), Van Laer et al. (in prep.), and Bruggeman et al. (2012). The logKd-values for Am(III) on Boom Clay in absence (SBCW) and presence of OM (RBCW) were generally high, i.e. 5.2 and 3.7, respectively. The latter value is significantly lower, which is assumed to be due to the complexation of Am(III) with the dissolved organic matter present in RBCW reducing the sorption by more than one log unit. By analogy, it can be assumed that the sorption of Cm(III) should be similar as for Am(III). For further details concerning the Am experiments (edge and isotherms), it is referred to the above mentioned reports.

Stumpf et al. (2001) performed a time-resolved laser fluorescence spectroscopy (TRLFS) study -7 of the sorption of 3×10 M Cm(III) onto smectite and kaolinite in 0.025 M NaClO4. Results showed that at low pH, Cm is sorbed as an outer-sphere complex retaining its complete primary hydration sphere. With increasing pH, inner-sphere adsorption occurs via the aluminol edge sites. Different surface complexes as function of pH were observed. These spectroscopic data are in line with the sorption complexes predicted by the 2 SPNE SC/CE model. Marques Fernandes et al. (2010) investigated influence of dissolved CO2 on the sorption of Cm(III) on alumina (γ-Al2O3) and kaolinite also by means of TRLFS. The spectra were observed to show features which were judged to be fully consistent with the formation of Cm(III) surface species involving carbonate complexes (so-called ternary complexes). In addition, at least two different Cm(III)-carbonate species were found to exist at the mineral- water interface.

Regarding dissolved organic matter complexation, a similar line of reasoning can be applied and differences between Am(III) and Cm(III) are not assumed to be significant. Moulin and Moulin (1995) experimentally determined interaction constants of Am and Cm with humic acids in the laboratory. Conditional interaction constants of Am(III) and Cm(III) with Aldrich humic acids under comparable pH and ionic strength conditions were very comparable. Figure 56 shows the effect of metal concentration on the interaction constant of Cm(III), Dy(III) and Am(III) with Aldrich humic acids. These constants were used to predict speciation under different conditions, one of which reflected Boom Clay conditions (pH ~ 8.5, pCO2 (atm) ~ 10-3, [HA] ~ 150 mg/L). In case of Am, between 60 and 100% of the aqueous speciation would occur as actinide-humic acid complex (Moulin and Moulin, 1995).

Kim et al. (1991) studied Cm(III)-humic acid complexation in 0.1 M NaClO4 at pH 6.0 by means of TRLFS. Humic acids were extracted from Gorleben groundwaters and Cm(III) concentration range varied from 2×10-8 to 10×10-8 M. Results showed a tridentate complexation and an average complexation constant could be determined. This constant was found to be in good agreement with values determined for Am(III). Under similar geochemical conditions, Monsallier et al. (2003) investigated the dissociation kinetics of

SCK•CEN/12201513 Page 134 of 208 Compilation of Technical Notes on less studied elements

Eu(III)/Tb(III)/Cm(III) humate complexes using both batch experiments (scavenging of dissociated metal ions and isotope exchange studies) and spectroscopic techniques (TRLFS and extended X-ray absorption fine structure, EXAFS).

Figure 56: Effect of metal concentration on the interaction constant of trivalent actinides with Aldrich humic acids (at a ionic strength of 0.1 M). TRLIF: Time-Resolved Laser-Induced Fluoresence; SP: spectrophotometry (Moulin and Moulin, 1995)

It was verified that slow and fast dissociation modes showed differences in molecular environment (Figure 57), and that a kinetic exchange equilibrium exists between the two modes.

Figure 57: Separation of curium humate TRLFS spectrum after four weeks of contact time (Cm-humate) into the spectrum of the slow dissociation mode and a rest representing curium humate that has dissociated during the five hours contact with the cation exchanger Chelex 100. The latter spectrum is a mixture between fast dissociation mode and a shoulder for the slow mode that also shows partial dissociation within the five hours contact time with Chelex (Monsallier et al., 2003)

Finally, interaction of Cm(III) with humic acid (10 mg/L) was studied as function of metal ion -8 -5 concentration (1×10 to 5×10 M) and pH (2.5 to 7.0) in 0.01 M NaClO4 by Rabung and Geckeis (2009). Spectroscopic results showed a certain heterogeneity in binding sites, depending on metal ion loading and pH (Figure 58 and Figure 59), which was in line with

SCK•CEN/12201513 Page 135 of 208 Compilation of Technical Notes on less studied elements

batch experiments performed with Eu(III)-humic acid revealing a variation in complexation constant with metal loading. Differences in related constants were, however, smaller than one order of magnitude. Variations in fluorescence lifetimes were interpreted as a consequence of increasing humic acid agglomeration at increasing metal loading and decreasing pH, thus leading to an enhanced local density of chromophoric groups close to the Cm(III) ion.

The results show that Cm(III) solid-liquid distribution in the Boom Clay is dependent on the sorption reactions with clay minerals and complexation reactions with dissolved organic matter (apart from inorganic aqueous speciation). At the pH and pCO2 of Boom Clay at the Mol site, surface complexes on clay minerals would dominate sorption on the solid clay fraction. These surface complexes are likely ternary Cm(III) carbonate species. Complexation with dissolved organic matter may be simulated using appropriate models taking into account variations in type of complexes formed as function of geochemical conditions, such as the humic ion-binding model VI (Tipping, 1998). Thus, the same phenomenological model as given in Bruggeman et al. (2012) would apply. Moreover, due to its high fluorescence intensity, Cm(III) can be used as a good proxy for spectroscopic identification and verification of speciation of trivalent lanthanides and actinides under Boom Clay conditions.

Figure 58: Trivalent metal ion humic acid complexation in dependence of the metal ion loading for 1×10-8 and -7 1×10 M Cm(III), at a constant pH (6.0) and humic acid concentration of 10 mg/L in 0.01 M NaClO4. The spectrum 3+ of the free Cm aq in 0.1 M HClO4 in the absence of humic acid is added to (a) (Rabung and Geckeis, 2009)

SCK•CEN/12201513 Page 136 of 208 Compilation of Technical Notes on less studied elements

4.2.2.4 Migration and diffusion properties Two percolation experiments (type C4) and one sequential migration experiment have been performed with Cm in confined Boom Clay cores. Details of these experiments can be found below:

1. Percolation C4 experiment Cm244/1/7a (NRM015A) • Initiated 19/01/1994; initially under Ar and since 28/10/1996 under Ar + 0.04%CO2 • Clay core code R88 3.5 – 3.8, MIG 06 coring 11/05/1992 • Initial activity 84 kBq, in chemical form Cm(NO3)3 in 0.1 N HNO3 • Cm concentration in source solution 1.16×10-6 M • Total clay core length: 72 mm (36 mm "inlet" + 36 mm "outlet"), diameter 38 mm • Used as input for sequential migration experiment 18/01/2006

2. Percolation C4 experiment Cm244/1/8 (NRM015B) • Initiated 19/01/1994; initially under Ar and since 28/10/1996 under Ar + 0.04%CO2 • Clay core code R88 3.5 – 3.8, MIG 06 coring 11/05/1992 • Initial activity 84 kBq, in chemical form Cm(NO3)3 in 0.1 N HNO3 • Cm concentration in source solution 1.16×10-6 M • Total clay core length: 72 mm (43 mm "inlet" + 29 mm "outlet"), diameter 38 mm • Percolated solutions followed until now

3. Sequential migration experiment Cm244/1/7b • Initiated 18/01/2006; under Ar + 0.04%CO2

SCK•CEN/12201513 Page 137 of 208 Compilation of Technical Notes on less studied elements

• Mounted after clay core Cm244/1/7a • Total clay core length: 104 mm (72 mm first core + 32 mm second core), diameter 38 mm

The hydraulic conductivity, K (m/s), for the three experiments is given in Figure 60. The value for K is in line with those normally observed in Putte Member of Boom Clay (Yu et al., 2013). The two clay cores exhibit virtually the same hydraulic conductivity. In all cores it is observed that the hydraulic conductivity is slowly decreasing in time, which is presently unexplained. Possibly, dissolved organic matter obstruction in the first filter before entering the clay core may be a possible source for this observation.

Both percolation experiments show a very rapid breakthrough of a very small fraction of Cm (observed in the first 200 days after start of the experiment), followed by a slow release from the clay core (Figure 61and Figure 62). The observed concentration range is orders of magnitude below the solubility of CmOHCO3(cr) and tends towards a more or less constant value (although maybe still slowly increasing with time). The very rapid breakthrough is also in contrast with high sorption (retardation) commonly associated with Cm in clay-rich environments.

In the sequential migration experiment, the Cm elution curve out of the second clay core exhibits similar features as the elution out of the first core, but without the "peak" feature observed in the percolation experiments. Rather, Cm is slowly increasing in the percolate solutions and a constant outlet value tends to be reached, but the observed concentrations are below (factor 5 – 7) the ones measured in the percolation experiments.

Hydraulic Conductivity 1.80 E-12

1.60 E-12

1.40 E-12

1.20 E-12

1.00 E-12

8.00 E-13

6.00 E-13 Cm244m1c7 Hydraulic conductivity K conductivity (m/s) Hydraulic

4.00 E-13 Cm244m1c8

2.00 E-13 Cm244m1c7 SM

0.00 E+00 0 1000 2000 3000 4000 5000 6000 7000 8000 Days since start experiment

Figure 60: Calculated hydraulic conductivity, K (in m/s), as function of time in 2 percolation experiments (Cm244m1c7 and Cm244m1c8) and sequential migration experiment (Cm244m1c7 SM)

This indicates that the transported species is not a conservative tracer (as otherwise the outlet concentration should equal the inlet concentration) and confirms that the observed concentration indeed does not correspond to the solubility of a certain solid phase.

SCK•CEN/12201513 Page 138 of 208 Compilation of Technical Notes on less studied elements

Cm concentration outlet 70

Cm244m1c7 60 Cm244m1c8

50 Cm244m1c7 SM

30 244 (Bq/L) -

Cm 20

0 1000 2000 3000 4000 5000 6000 7000

-10

Days since start experiment

Figure 61: Cm concentration in outlet (244Cm, as Bq/L) as function of time

The results of these experiments can be interpreted as resulting from DOM linked/facilitated transport of Cm(III). The Cm species which are eluted from the (first) clay core in the percolation experiments are Cm-DOM species, while the majority of the introduced Cm is retained at the source position (either as precipitate or adsorbed onto the solid phases). The concentration decrease after elution through the second clay core implies that the Cm-DOM species eluting from the first core are slowly dissociating. The dissociated species are subsequently retained in the second clay core.

Cm concentration outlet

9.00 E-14 Cm244m1c7 8.00 E-14 Cm244m1c8 Cm244m1c7 SM 7.00 E-14

6.00 E-14

5.00 E-14

4.00 E-14

3.00 E-14 Cm concentration (mol/L) Cm concentration 2.00 E-14

1.00 E-14

0.00 E+00 .0 200.0 400.0 600.0 800.0 1000.0 1200.0 Percolated volume (mL) since start experiment

Figure 62: Cm concentration in outlet (mol/L) as function of percolated volume

In Maes et al. (2011), a transport model was developed based on the conceptual model described in Figure 63. Radionuclides in solution will either be present as a mobile RN-DOM complex or "free inorganic" radionuclide species in solution ([RNinorg]liquid). The transfer between [RNinorg]liquid and the RN-DOM complex is described by a complexation constant and dissociation kinetics. Both species can interact with the solid phase. It is assumed that this SCK•CEN/12201513 Page 139 of 208 Compilation of Technical Notes on less studied elements

interaction in case of [RNinorg]liquid is mainly due to sorption processes and can be described by a retardation factor (RRN) which can be linked to batch sorption data. In case of RN-DOM, the retardation factor (RRN-DOM) is considered as lumped factor accounting for both sorption and colloid filtration processes. Overall, dissolved OM is only poorly retarded within Boom Clay and RRN-DOM is therefore expected to be only of secondary importance to describe RN- coupled transport. Within this transport model, the amount of parameters remains limited and most of them can be obtained from independent measurements (batch complexation/solubility experiments, batch sorption experiments, DOM transport experiments).

mobile RN-DOM complex

kdecomp [ RNinorg ]liquid RN-DOM

kcomp

Linear RRN R Sorption RN-DOM

RN RN-DOM Boom Clay solid phase

Figure 63: Conceptual model used for the interpretation of organic matter linked radionuclide migration in Boom clay (Maes et al., 2011)

The transport of the RN-DOM mobile complex is described by: ∂()c ηρ+ m −∇⋅η ∇ + =−ληρ + + η ()bK d,, m Dpore m cV m Darcy c m ()b K d, m c m Q sol− OM ∂t

The transport of the RN-species in solution is described by: ∂()c ηρ+ s −∇⋅η ∇ + =−ληρ + − η ()bK d,, s Dpore s cV s Darcy c s ()b K d, s c s Q sol− OM ∂t

Where cs, cm, cOM are the concentrations of free inorganic radionuclide species in solution, concentration of the mobile RN-DOM complex and the concentration of the mobile DOM. Dpore,m and Dpore,s are respectively the pore diffusion coefficients (Dpore) of the RN complexed to the mobile DOM and of the free inorganic RN species in solution. Kd,m, Kd,s are respectively the sorption distribution coefficients (Kd) for the RN complexed to the mobile DOM and for the free inorganic RN species in solution.

The mass transfer of the RN between the DOM complexed form and the "free" RN in solution is given by: Q= k cc − k c sol− OM comp s OM decomp m

SCK•CEN/12201513 Page 140 of 208 Compilation of Technical Notes on less studied elements

The symbols kcomp and kdecomp are respectively the kinetic rate constants for the RN-DOM complexation and decomplexation reactions and are linked to the equilibrium constant KRN- DOM according to following general reaction: kcomp c K RN −DOM = = m k c ⋅c decomp s DOM

To constrain the degrees of freedom present in the model, the RN-DOM interaction constant was fixed during fitting to a value of logKRN-DOM = 4.7. The model provides excellent fits (Figure 64) to the experimental data and the fitted parameters are presented in Table 24 (Maes et al., 2011).

Table 24: Summary of the fitted parameters kdecomp, RRN-DOM and RRN (Maes et al., 2011)

-1 logKRN-DOM kdecomp [s ] RRN-DOM [-] logRRN [-] Cm(III) 4.7 1.3±0.3×10-6 28±6 3.57±0.22

The fitted parameters (decomplexation constants, RN-DOM and RN retardation factors) are within a narrow range. Sensitivity analysis showed that RRN and KRN-DOM are the most influential parameters in the model and that they are correlated. If KRN-DOM is increased in the model, the fitted value for RRN will be higher (or vice versa) to obtain an equally good fit. In contrast, the model fit is not very sensitive to other fitting parameters.

Curium

Figure 64: Results of the fitting of the Cm elution curve in a sequential migration experiment using the proposed conceptual model (Maes et al., 2011)

4.2.2.5 Justification The reasoning for Cm(III) to associate it to the group of DOM linked RN transport is based in the first place on the available experimental data, either in-house or in literature, and in the second place on the analogy with Am(III). The strong affinity of Cm(III) for dissolved organic matter was clearly revealed by the different types of experiments reported in the previous paragraphs. Complexation experiments using RBCW revealed that Cm(III) was almost quantitatively associated to DOM (95%), and despite the fact of present bicarbonate (0.015 M NaHCO3), there were no indications for the formation of mixed ternary organic-carbonate- SCK•CEN/12201513 Page 141 of 208 Compilation of Technical Notes on less studied elements

Cm(III) complexes. These recently obtained results are quite surprising, as under BC conditions carbonte complexes have been generally considered to be the major species, with - Ac(III)(CO3)2 being the dominant one. As anionic carbonate complex, sorption is not expected to be high, and as the exact nature of the Cm-DOM complex is not known (yet), it is also difficult to judge how this complex would “play” on sorption. Ac(III) uptake by clays/BC was however determined to be high under BC conditions. In contrast to that, observations made during percolation experiments, i.e. fast breakthrough of 244Cm, indicate rather a low retardation of Cm(III). The seemingly contradictory results are however completely in line with the phenomenological model proposed for Am(III) (Bruggeman et al., 2012), in which the fast DOM linked transport on the one hand and strong retardation on the other hand is rationalized by incorporating kinetic complexation/decomplexation reactions into the RN- DOM interaction. In this context, it should be reminded here that some of the data for Cm were even used to construct and parameterise this model.

4.2.2.6 References Berner U. (1995) Estimate of solubility limits for safety relevant radionuclides. PSI-Bericht-95-07.

Bradbury M.H., Baeyens B. (1997) A mechanistic description of Ni and Zn sorption on Na- montmorillonite. Part II: Modelling, Journal of Contaminant Hydrology 27, 233-248.

Bradbury M.H., Baeyens B. (2005) Modelling the sorption of Mn(II), Co(II), Ni(II), Zn(II), Cd(II), Eu(III), Am(III), Sn(IV), Th(IV), Np(V) and U(VI) on montmorillonite: Linear free energy relationships and estimates of surface binding constants for some selected heavy metals and actinides, Geochimica et Cosmochimica Acta 69, 875-892.

Bradbury M.H., Baeyens B. (2009a) Sorpion modelling on illite. Part I: Titration measurements and the sorption of Ni, Co, Eu and Sn, Geochimica et Cosmochimica Acta 73, 990-1003.

Bradbury M.H., Baeyens B. (2009b) Sorption modelling on illite. Part II: Actinide sorption and linear free energy relationships, Geochimica et Cosmochimica Acta 73, 1004-1013.

Bruggeman C., Salah S., Maes N. (2012) Americium retention and migration behaviour in Boom Clay. Topical Report. First Full Draft, External report SCK•CEN-ER-20.

Bruno J., Duro L., Grivé M. and Merino J. (2001) Review of the radionuclide solubility limits for their use in PA exercises under clayey and cementitious conditions. Technical Report ANDRA C.RP.0ENQ.01.001.

Fanghänel T., Kim J.I., Paviet P., Klenze R., Hauser W. (1994) Thermodynamics of radioactive trace- elements in concentrated electrolyte solutions – Hydrolysis of Cm3+ in NaCl-solutions, Radiochimica Acta 66/67, 81-87.

Ugras, K., Bouby, M., Rabung, T., Kaden, P., Schild, D., Heck, S., and Schäfer, T. (2016) Characterization of Natural organic matter (NOM) derived from different layers within the Boom Clay Formation and their radionuclide interaction. KIT-INE internal report 04/2016.

Kim J.I., Wimmer H., Klenze R. (1991) A study of curium(III) humate complexation by time resolved laser fluorescence spectroscopy (TRLFS), Radiochimica Acta 54, 35-41.

Kim J.I., Klenze R., Wimmer H., Runde W., Hauser W. (1994) A study of the carbonate complexation of CmIII and EuIII by time-resolved laser fluorescence spectroscopy, Journal of Alloys and Compounds 213/214, 333-340.

SCK•CEN/12201513 Page 142 of 208 Compilation of Technical Notes on less studied elements

Lumetta G.J., Thompson M.C., Penneman R.A., Eller P.G. (2006) Chapter 9. Curium. In: The chemistry of the Actinide and Transactinide Elements. Third Edition. Morss L.R., Edelstein N.M., Fuger J., Katz J.J. (Eds.), Springer, Dordrecht, The Netherlands, 1397-1443.

Maes, N., Aertsens, M., Salah, S., Jacques, D., Van Gompel, M. (2009). Cs, Sr and Am retention on argillaceous host rocks: comparison of data from batch sorption tests and diffusion experiments. Scientific Report SCK•CEN-ER-98, SCK•CEN, Mol, Belgium.

Maes, N., Bruggeman, C., Govaerts, J., Martens, E., Salah, S., and van Gompel, M. (2011) A consistent phenomenological model for natural organic matter linked migration of Tc(IV), Cm(III), Np(IV), Pu(III/IV) and Pa(V) in the Boom Clay. Physics and Chemistry of the Earth, 36, 1590-1599.

Marques Fernandes M., Stumpf T., Baeyens B., Walther C., Bradbury M.H. (2010) Spectroscopic identification of ternary Cm-carbonate surface complexes, Environmental Science & Technology 44, 921-927

Monsallier J.-M., Artinger R., Denecke M.A., Scherbaum F.J., Buckau G., Kim J.I. (2003) Spectroscopic study (TRLFS and EXAFS) of the kinetics of An(III)/Ln(III) humate interaction, Radiochimica Acta 91, 567- 574

Moulin V., Moulin C. (1995) Fate of actinides in the presence of humic substances under conditions relevant to nuclear waste disposal, Applied Geochemistry 10, 573-580

Panak, P., R. Klenze, J. I. Kim (1996) A study of ternary complexes of Cm(III) with humic acid and hydroxide or carbonate in neutral pH range by time-resolved laser fluorescence spectroscopy. Radiochimica. Acta, 74, 141-146.

Rabung Th., Geckeis H. (2009) Influence of pH and metal ion loading on the Cm(III) humate complexation: a time resolved laser fluorescence spectroscopy study, Radiochimica Acta 97, 265-271

Salah, S., Maes, N. and Wang, L. (2007): Sorption behaviour of Am(III) and Th(IV) on Boom Clay. Poster presentation at the Migration Conference, 26/08-31/08/2007, München Germany.

Salah, S., Wang, L., Maes, N., Van Gompel, M. (2009). Sorption experiments and modeling of Am(III) onto Boom Clay. Poster presentation at the Migration Conference, 20/09-25/09/2009, Kennewick, Washington (USA).

Stumpf Th., Bauer A., Coppin F., Kim J.I. (2001) Time-resolved laser fluorescence spectroscopy study of the sorption of Cm(III) onto smectite and kaolinite, Environmental Science & Technology 35, 3691-3694

Vercouter T., Vitorge P., Amekraz B., Giffaut E., Hubert S., Moulin C. (2005) Stabilities of the aqueous 3- 3- complexes Cm(CO3)3 and Am(CO3)3 in the temperature range 10-70°C, Inorganic Chemistry 44, 5833-5843

Wimmer H., Kim J.I., Klenze R. (1992) A direct speciation of Cm(III) in natural aquatic systems by time- resolved laser-induced fluorescence spectroscopy (TRLFS), Radiochimica Acta 58/59, 165-171

Van Laer, L., Durce, D. , Salah, S., and Maes, N. (in prep.) Sorption studies on Boom Clay and clay minerals - status 2015. External Report SCK•CEN-ER-xx, SCK•CEN, Mol, Belgium.

Yu L., Rogiers B., Gedeon M., Marivoet J., De Craen M., Mallants D. (2013) A critical review of laboratory and in-situ hydraulic conductivity measurements for the Boom Clay in Belgium, Applied Clay Science 75-76, 1-12

SCK•CEN/12201513 Page 143 of 208 Compilation of Technical Notes on less studied elements

4.2.3 Technical note for Plutonium (Pu)

4.2.3.1 General Plutonium (atomic number = 94) is a transuranic radioactive element. Of great importance is the isotope 239Pu, with a half-life of 24,100 years, produced in extensive quantities in nuclear reactors from natural uranium:

238U → 239U → 239Np → 239Pu

Neutrons of the fission of 235U are captured by 238U to form 239U. Through two beta decays, and 239Np as intermediate, 239Pu is synthesized. Twenty isotopes of plutonium have been 238 239 characterized. The most stable plutonium isotopes are: Pu (t1/2 = 87.7 years), Pu (2.41 × 4 240 3 241 242 5 10 years), Pu (t1/2 = 6.56 × 10 years), Pu (t1/2 = 14.3 years), Pu (t1/2 = 3.74 × 10 years) and 244Pu (8.00 × 107 years).

4.2.3.2 Speciation and solubility In aqueous solutions, plutonium can exist in four oxidation states, i.e. the tri-, tetra-, penta- and hexavalent state. It has been shown that Pu(III), Pu(IV), Pu(V) and Pu(VI) can coexist in unequal quantities in the same solution. This can be explained by the similar redox potentials between the Pu(IV), Pu(V) and Pu(VI) states (Choppin, 2003). Disproportionation reactions may also influence the oxidation states of Pu. The tendency of plutonium to hydrolyze follows the effective charge of the ion in the order of (Kim, 1986):

4+ 2+ 3+ + Pu > PuO2 > Pu > PuO2

Pu(III) (when not complexed by stabilising ligands) is considered to be easily oxidized and becomes less stable as pH increases (Lemire et al., 2001). Therefore, under neutral or alkaline conditions, the trivalent state is much less likely to be present than under strong acid and anoxic conditions. Pu(III) has however been observed under anoxic conditions and in the presence of Fe(II) (Kirsch et al., 2011; Felmy et al., 2011; Felmy et al., 2013). Pu(VI) is only prevalent under highly oxidising conditions and reduces quickly to Pu(V). By contrast, Pu(IV) and Pu(V) are considered to be the most common species under (mildly oxic) environmental conditions. Pu(V) reduction to P(IV) has been frequently observed on a variety of minerals containing Mn(II) or Fe(III), as well as silica and montmorillonite (Powell et al., 2006; Zavarin et al., 2012; Begg et al., 2013; Hixon and Powell, 2014). The speciation of pentavalent Pu is relatively simple since it is only a very weak complexant (Choppin, 2003) and begins to hydrolyse between pH 7-8. Penta- as well as hexavalent plutonium form dioxo-cations, such + 2+ as Pu(V)O2 and Pu(VI)O2 . Pu(IV) is characterised by a high tendency to hydrolyse and easily forms intrinsic colloids or nanoparticles. These species play a specific role in the redox housekeeping of plutonium, as described in the special section below.

According to the speciation calculations using MOLDATA, mono-nuclear Pu(III)-carbonate 3- - complexes, i.e. Pu(CO3)3 and Pu(CO3)2 are predicted to represent the dominant aqueous species under BC conditions (Figure 65). These species (and their respective reaction constants) are taken from the ANDRA TDB Thermochimie v5. If other databases are used (LLNL TDB, NAGRA/PSI TDB, NEA TDB), rather Pu(OH)4(aq) species are predicted to dominate

SCK•CEN/12201513 Page 144 of 208 Compilation of Technical Notes on less studied elements

the aqueous speciation (Salah and Wang, 2014). At the moment we do not have any direct (e.g., spectroscopic) proof of Pu(III) versus Pu(IV) equilibrium redox state in Boom Clay.

++ 1 PuO2

++ Pu(OH)2 PuO CO (aq) PuO+ 2 3 2 -- PuO2(CO3)2

---- .5 + PuO2(CO- 3)3 Pu(OH)3 PuO2CO3 PuO2(OH)2(aq) PuO OH(aq) Pu+++ 2 - Pu(CO3)2(OH)2 2

Eh (volts) 0 Pu(CO )+ 3 Pu(OH)4(aq) - Pu(COµ3)2 --- –.5 Pu(CO3)3 25°C Pu(OH)3(aq) 0 2 4 6 8 10 12 14 pH

Figure 65: Eh-pH diagram of plutonium (Pu-C-S-O-H) for the BC reference porewater system. Assumed dissolved activity of [Pu] = 10-8. Database MOLDATA_R2. Code: The Geochemist's Workbench - 10.0

Results of the solubility calculations are summarized in Table 25. Using a Pu activity of 10-8, the crystalline Pu-oxide represents the most stable solid phase. However, taking into account the Ostwald principle, rather the amorphous phase, i.e. PuO2(am,hyd) would precipitate first and with time/ageing convert to the crystalline form. At higher Pu activities (e.g. 10-4), the mixed Pu-hydroxy-carbonate solid becomes stable and would be the solubility controlling mineral. Pu(OH)3(cr) occupies the stability field under more alkaline conditions.

Table 25: Solubility of Pu in the BC reference porewater system at 25°C, pH 8.355 and Eh -281 mV. Database: MOLDATA. Code: The Geochemist's Workbench - 10.0

Solubility controlling phases Solubility, [Pu], mol/L

1 -4 Pu(OH)3(cr) 1.43 × 10 2 -5 PuCO3OH(s) 1.58 × 10 1 -8 PuO2 (am,hyd) 1.26 × 10 1 -14 PuO2 (cr) 2.52 × 10

Source data: 1NEA TDB, 2ANDRA TDB

The reaction constants of the Pu solids comprised in Table 25 and MOLDATA are the following:

+ 4+ Pu(OH)3(cr) + 4 H + 0.25 O2(aq) ↔ Pu + 3.5 H2O log K = 19.6 + 4+ - PuCO3OH(s) + 0.25 O2(aq) +3 H ↔ Pu + HCO3 + 1.5 H2O log K = 8.39 + 4+ PuO2(am,hyd) + 4 H ↔ Pu + 2 H2O log K = -2.33

SCK•CEN/12201513 Page 145 of 208 Compilation of Technical Notes on less studied elements

+ 4+ PuO2(cr) + 4 H ↔ Pu + 2 H2O log K = -8.03

In Table 26, the calculated species distribution in equilibrium with PuO2(am,hyd) is illustrated. It can be seen, that 98.8 % are Pu(III) species and only 1.2 % of the species occur in the tetravalent oxidation state. It is remarked that, similar to uranium, a Pu precipitate with a specific valence state (+IV) is calculated to be in equilibrium with aqueous species which are predominantly present in another valence state (+III) (Salah et al., 2014).

Table 26: Species distribution of Pu in equilibrium with PuO2(am,hyd). Database: MOLDATA. Code: The Geochemist's Workbench - 10.0

Aqueous species Pu [mol/L] Fraction [%]

III 3- -9 Pu (CO3)3 8.81 × 10 69.8

III - -9 Pu (CO3)2 3.35 × 10 26.5

III + -10 Pu CO3 3.18 × 10 2.5

(IV) 2- -10 Pu (CO3)2(OH)2 1.51 × 10 1.2 Total 1.26 × 10-8 100

Pu solubility studies were performed during the TRANCOM-II project (Maes et al., 2004). Purified Pu(IV) solutions were titrated to pH 13 forming a hardly visible grey Pu(IV) precipitate/gel (presumably PuO2). Aliquots of this suspension were mixed with pure water and Synthetic Boom Clay Water (SBCW) containing different concentrations of TROM-type Boom Clay humic substances, and with Real Boom Clay Water (RBCW) in an Ar/0.4%CO2 atmosphere. Supernatant samples were taken after 15 and 84 days equilibration, and analysed for Pu after ultrafiltration at 30000 MWCO (Figure 66).

Figure 66: Evolution in Pu(IV) concentrations (after filtration at 30000 MWCO) in H2O, SBCW (with increasing concentrations of TROM humic substances) and in RBCW. Measurements after 15 and 84 days (Maes et al., 2004)

SCK•CEN/12201513 Page 146 of 208 Compilation of Technical Notes on less studied elements

Relatively high Pu concentrations (~ 1.5×10-6 M) were noted in the blank samples (no organic matter), possibly due to oxidation of Pu(IV) to Pu(VI) (Eh up to ~300 mV). Upon addition of dissolved organic matter (DOM), Pu(VI) concentrations dropped suggesting that humic substances are capable of keeping Pu in reduced form, i.e. Pu(IV) (Marquardt et al., 2004). Upon increasing DOM from 20 to 500 mgC/L, also Pu solubility increased by about an order of magnitude from ~10-7 to 10-6 M. The solubility in RBCW (~ 100 mgC/L) was 2.6×10-7 M. The results indicate that Pu forms complexes with Boom Clay humic substances, but the exact nature of the interaction and the valence state of Pu is unclear. Marquardt et al. (2004) witnessed both Pu(III) and Pu(IV) in Gorleben-type groundwater under reducing conditions. Marsac et al. (2014) investigated Pu(IV) interaction with humic acid at pH < 3 and found evidence of stabilisation of polynuclear plutonium(IV) species by humic acid at [Pu] > ~10- 8 M. Humic Ion-Binding Model VII introduced into PhreeqC could be used to simulate Pu(IV) monomer binding to humic acids.

Note on Pu redox and role of Pu(IV) colloids

It is well known that tetravalent actinides have a high tendency towards polynucleation and colloid formation (Neck et al., 2007). In contact with solid AnO2(am,hyd), these oxyhydroxide colloids (AnO2(coll, hyd)) remain stable in solution. On the one hand they have properties of small solid particles, on the other hand they must be considered as large polynuclear aqueous species in equilibrium with both solid AnO2(am, hyd) and aqueous species 4m-n Anm(OH)n . Neck et al. (2007) described and quantified Pu(IV) colloid formation by the reaction:

PuO2(am, hyd) ⇔ PuO2(coll, hyd) with log K = -8.3±1.0 for Pu(IV). This means that the level of [Pu(IV)]coll is about two orders of magnitude higher than the Pu(IV) concentration after 1.5 nm ultrafiltration or ultracentrifugation (log K = -10.4±0.5) (Figure 67). These Pu(IV) colloids are not a single species but have a certain size distribution, estimated to be in the range 1.5 – 5 nm (Neck et al., 2007; Walther et al., 2009). Colloidal Pu(IV) oxyhydroxide particles formed at low pH are larger (size > 5 nm) than those in neutral and alkaline solutions (Neck et al., 2007). Small Pu(IV) colloids or polymers present in neutral to alkaline solutions are also observed to play an important role in the redox behaviour of Pu (Neck et al., 2007; Walther et al., 2009). A schematic drawing is given in Figure 68 which highlights their importance in the redox reaction between Pu(IV) and Pu(V). Contrary to the reversible redox couples Pu3+/Pu4+ and + 2+ 4+ + PuO2 /PuO2 , the Pu and PuO2 ions and their hydroxide complexes are not directly in equilibrium with each other, but only indirectly via their reactions with solid, colloidal or polymeric Pu(IV) (Neck et al., 2007).

Several studies have highlighted the role of Pu colloid formation and the interactions of Pu colloids with environmentally relevant solid phases to elucidate subsurface migration of Pu. Powell et al. (2011) characterized Pu nanocolloids (2-5 nm in diameter) using transmission electron microscopy (TEM) and studied their interactions with goethite and quartz. On goethite, epitaxial growth of Pu colloid was observed together with lattice distortion relative to the ideal fluorite-type structure, leading to stronger binding of Pu to goethite.

SCK•CEN/12201513 Page 147 of 208 Compilation of Technical Notes on less studied elements

Delegard (2011) investigated Pu in Hanford waste tanks and found through thermodynamic considerations and laboratory studies that freshly precipitated Pu is likely present as 2-3-nm PuO2×H2O crystallite particles. From that point they would grow at exceedingly slow rates (particle size values of about 5 nm are put forward). In hydrothermal conditions, growth would be extended to tens of nanometers after decades of ripening.

Figure 67: Concentration of colloidal/polymeric Pu(IV) (squares with crosses inside) and Pu concentration in the supernatant after ultrafiltration (crosses) determined in a PuO2 solubility study under Ar atmosphere (Neck et al., 2007)

Figure 68: Solid-liquid and redox equilibria of plutonium in the presence of oxygen (Neck et al., 2007)

Abdel-Fattah et al. (2013) investigated the electrokinetic properties, critical coagulation concentration and effective Hamaker constant (important for calculating colloid-colloid and colloid-interface van der Waals interactions) of intrinsic Pu(IV) colloids. The colloids were prepared using the sol-gel method and after several conditioning steps, a stable colloidal suspension was obtained with dominant particle size between 30 and 40 nm. The dependence of ζ-potential of the intrinsic Pu(IV) colloids on pH was quantified and an

SCK•CEN/12201513 Page 148 of 208 Compilation of Technical Notes on less studied elements

isoelectric point (IEP) of pH 8.6 was obtained (Figure 69). The ζ-potential/pH dependence of the Pu(IV) colloids is similar to that of goethite and hematite colloids. Aggregation experiments were conducted at pH 11.4 (ζ-potential of -40 mV) and indicate a critical coagulation concentration of 0.1 M of 1:1 electrolyte. The results indicate that intrinsic Pu colloids have very strong colloid−colloid interactions, which makes the Pu(IV) colloids favour aggregation among themselves or with other colloids in the system (heterocoagulation).

Figure 69: Measured ζ-potential as a function of pH of intrinsic Pu colloids and natural smectite colloids (Abdel-Fattah et al., 2013)

4.2.3.3 Sorption and retardation Henrion et al. (1985) examined several factors likely to influence Pu sorption on Boom Clay, like the solid-to-liquid ratio (40 to 500 g/L), the contact time (1, 6 and 60 days) and the degree of phase separation. Pu was added from an acid stock solution of Pu(IV) and the initial Pu concentration in the systems was 2.3×10-9 M. Phase separation was performed by centrifugation (21000 rpm, 2 hrs) and further by ultrafiltration (Millipore, 1000 and 40000 MWCO). The Pu sorption as function of the reciprocal of the DOC content of the supernatant (measured as absorbance at 280 nm) is shown in Figure 70.

Kd values tend to increase with time and with the degree of phase separation (data not shown). However, a considerable dispersion is observed in Figure 70, which is in contrast with corresponding plots with Np and Eu (Henrion et al., 1985). The increase of Kd with the reciprocal of the absorbance is interpreted as an indication of Pu complexation by humic substances. The data scatter could be due to the presence of multiple Pu oxidation states, or due to Pu(IV) intrinsic colloid formation and possible precipitation.

No other (more recent) experiments have been performed on Pu sorption under geochemical conditions relevant to Boom Clay.

It is generally recognised that Pu(III) and Pu(IV) would sorb strongly on various minerals, including clay minerals, because of their strong hydrolysis. In the neutral to alkaline pH range surface complexation will dominate the sorption process. This process can be adequately described and predicted using the 2-site protolysis surface complexation/cation exchange

SCK•CEN/12201513 Page 149 of 208 Compilation of Technical Notes on less studied elements

model developed by Bradbury and Baeyens (Bradbury and Baeyens, 1997; Bradbury and Baeyens, 2009a). Although no formal studies have been presented on Pu uptake by illite or montmorillonite within this framework, Pu sorption may be predicted based on linear free energy relationships (Bradbury and Baeyens, 2005; Bradbury and Baeyens, 2009b).

Figure 70: Plutonium sorption (Kd) as a function of the reciprocal of the absorbance at 280 nm (Henrion et al., 1985)

Begg et al. (2013) studied Pu(IV) sorption to SWy-1 montmorillonite (1 g/L) at initial -11 -6 concentrations of 10 – 10 M under air in 0.7 mM NaHCO3, 5 mM NaCl buffer solution (pH 8). Pu(IV) sorption occurred quickly (< 1 h) but sorption continued to increase at a very slow rate until 30 days (attributed to slow sorption of Pu(V) impurities). The sorption isotherm of Pu(IV) is linear over the whole range of initial Pu concentrations (Figure 71).

Figure 71: 30 day Pu(V) (circles) and 30 day Pu(IV) (squares) sorption isotherms for SWy-1 Na-montmorillonite (1 g/L) in 0.7 mM NaHCO3, 5 mM NaCl buffer solution at pH 8 (Begg et al., 2013)

SCK•CEN/12201513 Page 150 of 208 Compilation of Technical Notes on less studied elements

In numerous studies on Pu(V) sorption, a surface-mediated reduction of Pu(V) to Pu(IV) is observed (Zavarin et al., 2005; Zavarin et al., 2012; Begg et al., 2013). After extended equilibration times, Pu(V) adsorption on montmorillonite becomes very similar to that observed for Pu(IV) (Begg et al., 2013). Minerals containing significant Fe and Mn (thus, minerals containing redox couples) could remove Pu(V) from solution in increasing rate compared to minerals which did not (Beggs et al., 2013).

Recently, some authors also studied sorption of plutonium oxide nanoparticles to goethite (Zavarin et al., 2014) and muscovite (Schmidt et al., 2012). Uptake of 1 nm edge size (cubic) Pu nanoparticles (NPs) to the muscovite basal plane was observed in 100 mM NaCl background electrolyte at pH 2.6 by Schmidt et al. (2012). The authors found evidences of both "pure" adsorption of Pu NPs directly on the surface as well as aggregation of the NPs driven by sorption and accumulation of Pu NPs at the interface. Zavarin et al. (2014) studied 3-5 nm PuO2 NPs sorption to goethite and found a weaker association compared to aqueous Pu(IV). Similar to Schmidt et al. (2012) aggregation of the NPs at the solution-surface interface was observed.

Dissolved organic carbon of colloidal nature was shown to profoundly inhibit the adsorption of reduced Pu (oxidation states III or IV) to suspended sediment particles (Nelson et al., 1985). A strong inverse relationship between Kd and DOC was observed in a variety of environmental systems (Figure 72). At 100 mg/L DOC, Kd values decreased over about 3 orders of magnitude compared to organic-free systems. It was also found that the systems showed high reversibility when equilibrium was approached from the directions of both adsorption and desorption.

Figure 72: Relationship of Kd of reduced Pu to dissolved organic carbon concentration for 15 lakes and four rivers. 237 The results of four laboratory measurements of Kd using added Pu tracer are shown for comparison (Nelson et al., 1985)

Buda et al. (2008) studied the pH dependency as well as the influence of contact time, sequence of addition and DOM concentration in the ternary systems Pu(III)-kaolinite-DOM and Pu(IV)-kaolinite-DOM (0.1 M NaClO4 background electrolyte, Aldrich humic acid 10-100

SCK•CEN/12201513 Page 151 of 208 Compilation of Technical Notes on less studied elements

mg/L, [Pu(III)] ~ 10-6 M, [Pu(IV)] ~ 10-8 M). DOM tended to increase the sorption of Pu below pH 6 and decrease sorption at higher pH. Sequence of addition also seemed to have an influence on measured Kd values.

4.2.3.4 Diffusion and transport Three percolation experiments (type C4) and one sequential migration experiment have been performed with Pu in confined Boom Clay cores. Details of these experiments can be found below:

1. Percolation C4 experiment Pu238/3/2 • Initiated 13/07/1988; under Ar • Clay core code Test Drift R124 21/10/1987 (vertical) • Initial activity 47 kBq, in chemical form Pu in 4 N HNO3 • Pu concentration in source solution 3.13×10-6 M • Total clay core length: 70 mm (35.5 mm "inlet" + 34.5 mm "outlet"), diameter 32 mm • Stopped 07/04/2009

2. Percolation C4 experiment Pu238/2/4 (NRM013A) • Initiated 06/07/1995; under Ar • Clay core code R17 4.5 – 4.83, number 7, coring 07/09/1994 (vertical) • Initial activity 139 kBq, in chemical form Pu in 3 N HNO3 • Pu concentration in source solution 4.86×10-6 M • Total clay core length: 72 mm (40 mm "inlet" + 32 mm "outlet"), diameter 38 mm • Percolated solutions followed until now

3. Percolation C4 experiment Pu238/2/6a (NRM013B) • Initiated 06/07/1995; under Ar • Clay core code R17 4.5 – 4.83, number 7, coring 07/09/1994 (vertical) • Initial activity 138 kBq, in chemical form Pu in 3 N HNO3 • Pu concentration in source solution 4.68×10-6 M • Total clay core length: 72 mm (40 mm "inlet" + 32 mm "outlet"), diameter 38 mm • Used as input for sequential migration experiment 19/01/2006

4. Sequential migration experiment Pu238/2/6b

• Initiated 19/01/2006; under Ar • Mounted after clay core Pu238/2/6b • Total clay core length: 104 mm (72 mm first core + 32 mm second core), diameter 38 mm

The hydraulic conductivity, K (m/s), for the four experiments is given in Figure 73. The value for K is in line with those normally observed in Putte Member of Boom Clay (Yu et al., 2013). The two clay cores Pu238/2/4 and Pu238/2/6 exhibit virtually the same hydraulic conductivity, while the value for core Pu238/3/2 is slightly higher. In all cores it is observed that the hydraulic conductivity is slowly decreasing in time, which is presently unexplained. Possibly, dissolved organic matter obstruction in the first filter before entering the clay core may be a possible source for this observation.

SCK•CEN/12201513 Page 152 of 208 Compilation of Technical Notes on less studied elements

All three percolation experiments show a similar breakthrough pattern Figure 74 and Figure 75. After an initial breakthrough of a very small fraction of Pu (observed in the first 50 days after start of the experiment and most prominently visible in Pu238/3/2), a slow release of Pu from the clay core is observed. The observed concentration range is orders of magnitude below the solubility of PuO2(am,hyd) and tends towards a more or less constant value (although maybe still slowly increasing with time, as visible in Pu238/2/4). The very rapid breakthrough is in contrast with high sorption (retardation) commonly associated with Pu in clay-rich environments. The breakthrough curves show some striking similarities with the ones observed for Cm in similar percolation experiments (although the initial peak tends to be a bit weaker) and may point to Pu(III) as most prominent valence state.

Hydraulic Conductivity 4.00 E-12 Pu238m3c2 3.50 E-12 Pu238m2c4 Pu238m2c6 3.00 E-12 Pu238m2c6 SM

2.50 E-12

2.00 E-12

1.50 E-12 Hydraulic conductivity K conductivity (m/s) Hydraulic 1.00 E-12

5.00 E-13

0.00 E+00 0 1000 2000 3000 4000 5000 6000 7000 8000 Days since start experiment

Figure 73: Calculated hydraulic conductivity, K (in m/s), as function of time in 3 percolation experiments (Pu238m3c2, Pu238m2c4, Pu238m2c6) and 1 sequential migration experiment (Pu238m2c6 SM)

In the sequential migration experiment, the Pu elution curve out of the second clay core exhibits similar features as the elution out of the first core, but without the "peak" feature observed in the percolation experiments. Rather, Pu is slowly increasing in the percolate solutions, but the observed concentrations are below (factor 3 - 4) the ones recorded in the percolation experiments. This indicates that the transported species is not a conservative tracer (as otherwise the outlet concentration should equal the inlet concentration) and confirms that the observed concentration indeed does not correspond to the solubility of a certain solid phase.

The results of these experiments can be interpreted as resulting from Pu-dissolved organic matter (DOM) colloid facilitated transport. The Pu species which are eluted from the (first) clay core in the percolation experiments are Pu-DOM species, while the majority of applied Pu is retained in the source position (either as precipitate or adsorbed onto the solid phases). The concentration decrease after elution through the second clay core implies that the Pu- DOM species eluting from the first core are slowly dissociating. The dissociated species are subsequently retained in the second clay core.

SCK•CEN/12201513 Page 153 of 208 Compilation of Technical Notes on less studied elements

Pu concentration outlet 60.00 Pu238m3c2

50.00 Pu238m2c4 Pu238m2c6 Pu238m2c6 SM 40.00

30.00 238 (Bq/L) - 20.00 Pu

10.00

0.00 0 100 200 300 400 500 600 700 800 900 1000

-10.00

Days since start experiment

Figure 74: Pu concentration in outlet (238Pu, in Bq/L) as function of time

Pu concentration outlet 1.80 E-13

1.60 E-13

1.40 E-13

1.20 E-13

1.00 E-13

8.00 E-14 Pu238m3c2 Pu238m2c4 6.00 E-14 Pu238m2c6 Pu concentration (mol/L) Pu concentration 4.00 E-14 Pu238m2c6 SM

2.00 E-14

0.00 E+00 0.0 200.0 400.0 600.0 800.0 1000.0 1200.0 1400.0 1600.0 Percolated volume (mL) since start experiment

Figure 75: Pu concentration in outlet (mol/L) as function of percolated volume (mL)

In Maes et al. (2011) a transport model was constructed based on the conceptual model described in Figure 76. Radionuclides in solution will either be present as a mobile RN-DOM complex or "free inorganic" radionuclide species in solution ([RNinorg]liquid). The transfer between [RNinorg]liquid and the RN-DOM complex is described by a complexation constant and dissociation kinetics. Both species can interact with the solid phase. It is assumed that this interaction in case of [RNinorg]liquid is mainly due to sorption processes and can be described by a retardation factor (RRN) that can be linked to batch sorption data. In case of RN-DOM the retardation factor (RRN-DOM) is considered as lumped factor accounting for

SCK•CEN/12201513 Page 154 of 208 Compilation of Technical Notes on less studied elements

both sorption and colloid filtration processes. Overall, dissolved OM is only poorly retarded within Boom Clay and RRN-DOM is therefore expected to be only of secondary importance to describe RN-coupled transport. Within this transport model, the amount of parameters remains limited and most of them can be obtained from independent measurements (batch complexation/solubility experiments, batch sorption experiments, DOM transport experiments).

mobile RN-DOM complex

kdecomp [ RNinorg ]liquid RN-DOM

kcomp

Linear RRN R Sorption RN-DOM

RN RN-DOM Boom Clay solid phase

Figure 76: Conceptual model used for the interpretation of organic matter linked radionuclide migration in Boom clay (Maes et al., 2011)

The transport of the RN-DOM mobile complex is described by: ∂()c ηρ+ m −∇⋅η ∇ + =−ληρ + + η ()bK d,, m Dpore m cV m Darcy c m ()b K d, m c m Q sol− OM ∂t

The transport of the RN-species in solution is described by: ∂()c ηρ+ s −∇⋅η ∇ + =−ληρ + − η ()bK d,, s Dpore s cV s Darcy c s ()b K d, s c s Q sol− OM ∂t

The mass transfer of the RN between the DOM complexed form and the "free" RN in solution is given by: Q= k cc − k c sol− OM comp s OM decomp m

SCK•CEN/12201513 Page 155 of 208 Compilation of Technical Notes on less studied elements

kcomp c K RN −DOM = = m k c ⋅c decomp s DOM

To constrain the degrees of freedom present in the model, the RN-DOM interaction constant was fixed during fitting to a value of logKRN-DOM = 4.7 (the same value as used for Cm). The model provides excellent fits (Figure 77) to the experimental data and the fitted parameters are presented in Table 27 (Maes et al., 2011).

Table 27: Summary of the fitted parameters kdecomp, RRN-DOM and RRN (Maes et al., 2011)

-1 Element logKRN-DOM kdecomp [s ] RRN-DOM [-] logRRN [-] Pu 4.7 1.0±0.1×10-6 22±2 4.00±0.23

The fitted parameters (decomplexation constants, RN-DOM and RN retardation factors) are within a narrow range. Sensitivity analysis showed that RRN and KRN-DOM are the most influential parameters in the model and that they are correlated. If KRN-DOM is increased in the model, the fitted value for RRN will be higher (or vice versa) to obtain an equally good fit. By contrast, the model fit is not very sensitive to other fit parameters.

Plutonium

Figure 77: Results of the fitting of the Pu elution curve in a sequential migration experiment using the proposed conceptual model (Maes et al., 2011)

Concluding, the results are completely in line with the phenomenological model proposed for Am(III) (Bruggeman et al., 2012).

Several authors reported colloid-facilitated transport of traces of Pu over distances more than 1 km near nuclear legacy sites (Penrose et al., 1990; Kersting et al., 1999; Novikov et al., 2006; Kersting, 2013). Although many studies experimentally demonstrated adsorption of Pu onto a variety of minerals and mineral assemblage, little is known of the colloid-mediated transport in the far field where there are many competing processes, such as desorption from the colloids and resorption onto minerals. It has been shown that colloids rich in Al-silicates, organic carbon, or iron are capable of facilitating transport of Pu. Under most surface soil

SCK•CEN/12201513 Page 156 of 208 Compilation of Technical Notes on less studied elements

conditions, +4 is the dominant Pu oxidation state (Santschi et al., 2002; Novikov et al., 2006). This state has the ability to form complexes with a range of environmental ligands (e.g., metal oxides, organic macromolecules, cell walls).

Pu colloids, also sometimes referred to as Pu nanoparticles or Pu nanocolloids, can be intrinsic Pu oxide/hydroxide particles as small as 2−5 nm in size or particle aggregates ranging between 30 and 120 nm, depending on the solution chemistry (Abdel-Fattah et al., 2013). Another form of Pu colloids, pseudo Pu colloids, arises when dissolved Pu or intrinsic Pu particles associate with colloidal particles that coexist in the system (e.g., natural groundwater colloids in the case of subsurface systems). Unlike dissolved Pu species, which can readily diffuse into small and dead-end pore spaces or sorb onto the immobile rock and soil grains, Pu colloids may experience size and volume exclusion effects during their transport in porous media, rendering them “facilitators” of Pu transport. During their transport in the subsurface media, the intrinsic Pu(IV) colloids may (a) interact with other Pu(IV) or native colloidal particles (colloid−colloid interactions), (b) interact with the surrounding interfaces (colloid-interface interactions), or (c) slowly dissolve and subsequently adsorb onto native colloidal particles and/or surrounding interfaces.

Kersting et al. (1999) reported migration of Pu over distances > 1 km in 30 years in aquifers (estimated flow velocity ranging from 1 to 80 m/year) at the Nevada Test Site, USA. Water samples were filtered in series using 1000-nm (particulate fraction), 50-nm and ~7-nm (colloidal fractions) filter sizes. X-ray diffraction (XRD) and scanning electron microscopy (SEM) analyses showed that the particulate and colloidal size fractions were composed of clays (illite and smectite), zeolites and cristobalite. Since Pu in groundwater samples was associated with the colloidal fraction, they argued that colloidal groundwater migration played an important role in transporting the plutonium. From this study, it was impossible to distinguish whether Pu was transported adsorbed or associated with the clays and zeolites, or as intrinsic colloid composed of Pu oxide.

Similarly, Novikov et al. (2006) reported Pu activities in the groundwater at a distance of 3 kilometers within ~ 55 years from the Mayak Production Association, Urals, Russia. The redox potential of the groundwater was +50 to ~+480 mV, and the pH was ~6 to 8. Water samples were filtered at 200 nm, 50 nm, 15 nm, 10 kDa (~ 1.5 nm) and 3 kDa (~ 1 nm). Between 70 and 90% of Pu was found to be sorbed onto colloids (mostly between ~1 and 15 nm), confirming the role that colloids play in the long-distance transport of Pu. Moreover, the further the distance from the source, the more Pu was found in the colloidal fraction, which was ascribed to a disequilibrium derived from the slow desorption of Pu from the colloids or to the irreversible incorporation of trace Pu in aquatic colloids. By contrast, U (in the +VI form) and Np (in the +V form) were mostly present as dissolved species (< 1 nm). Electron microscopy analysis showed a variety of colloid phases, with amorphous Fe hydroxide (HFO) as the most abundant phase. Nano-secondary ion mass spectrometry (SIMS) elemental maps revealed that Pu(IV) was associated with the HFO colloids.

Although these studies focused on inorganic colloids as vector in the "enhancement" of Pu mobility, also organic colloids can play this role (Santschi et al., 2002). Xu et al. (2008) reported Pu being mobilized from a contaminated soil (dating from the 1960s) near the Rocky Flats Nuclear Weapons Plant (USA). Pu was found to be transported, at sub-pM concentrations, by a cutin-like natural substance containing siderophore-like moieties.

SCK•CEN/12201513 Page 157 of 208 Compilation of Technical Notes on less studied elements

Because of its amphiphilicity and small size (~ 6 kDa), these macromolecules were perceived as an ideal vector for the dispersal of actinides and other metals.

The results obtained by Abdel-Fattah et al. (2013) described earlier in this report, lead the authors to the hypothesis that intrinsic Pu(IV) colloids could transport associated with other colloids that coexist in sufficient concentrations in the groundwater as a stable phase. This hypothesis was tested and verified through a packed column transport experiment (30 cm long column, 2.5 cm diameter, alluvium packing material, 41% porosity, flow rate 6 mL/h, residence time ~10h). The column was injected first with natural groundwater containing smectite colloids and afterwards with an aliquot of Pu(IV) intrinsic colloid stock suspension (Pu concentration 1.5×10-6M). The normalized breakthrough curves of Pu activity, smectite colloids and conservative tritium tracer are shown in Figure 78. The BTCs reveal that about 100% of the smectite colloids and 30% of Pu transported completely unretarded through the column, despite favourable deposition conditions. The results suggest that intrinsic Pu(IV) colloid may migrate in the subsurface associated with stable groundwater colloids and point out the importance of heterocoagulation processes. Similar results were obtained by Xie et al. (2013).

Figure 78: Normalized column breakthrough curves of Pu, smectite colloids and conservative tritium tracer. The dotted line curve is obtained by dividing the measured effluent Pu activity at any time by the maximum measured effluent activity (Abdel-Fattah et al., 2013)

4.2.3.5 Justification The reasoning for Pu to associate it to the trivalent actinide group is not very straightforward, as Pu occurs in different oxidation states. Under BC conditions, Pu(III) is considered to be the prevalent oxidation state, which is based on the one hand on the speciation calculations, III 3- predicting Pu (CO3)3 to be the dominant species, and on the other hand on the in-house performed percolation experiments, reflecting for 238Pu a similar transport behaviour as for 244Cm. Not only that a strong affinity of Pu for organics was described previously, the latter

SCK•CEN/12201513 Page 158 of 208 Compilation of Technical Notes on less studied elements

are even assumed to be able to keep or stabilize Pu in the reduced oxidation state. Despite the fact that the interaction mechanisms between Pu and DOM might be diverse (colloid- colloid interaction, sorption, heterocoagulation) and not well characterized, the transport behaviour could be well described by the conceptual model put forward by Maes et al. (2011).

4.2.3.6 References Abdel-Fattah, A.I., Zhou, D., Boukhalfa, H., Tarimala, S., Ware, S.D., Keller, A.A. (2013) Dispersion stability and electrokinetic properties of intrinsic plutonium colloids: Implications for subsurface transport, Environmental Science & Technology, 47, 5626-5634

Begg, J.D., Zavarin, M., Zhao, P., Tumey, S.J. Powell, B., Kersting, A.B. (2013) Pu(V) and Pu(IV) sorption to montmorillonite, Environmental Science & Technology, 47, 5146-5153

Bradbury, M.H., Baeyens, B. (1997) A mechanistic description of Ni and Zn sorptionon Na- montmorillonite. Part II: modelling, Journal of Contaminant Hydrology, 27, 223-248.

Bradbury, M. H. and Baeyens, B. (2005) Modelling the sorption of Mn(II), Co(II), Ni(II), Zn(II), Cd(II), Eu(III), Am(III), Sn(IV), Th(IV), Np(V) and U(VI) on montmorillonite: Linear free energy relationships and estimates of surface binding constants for some selected heavy metals and actinides, Geochimica et Cosmochimica Acta, 69, 5391-5392.

Bradbury, M. H. and Baeyens, B. (2009a) Sorption modelling on illite. Part I: Titration measurements and the sorption of Ni, Co, Eu and Sn, Geochimica et Cosmochimica Acta, 73, 990-1003

Bradbury, M.H. and Baeyens, B. (2009b) Sorption modelling on illite. Part II: Actinide sorption and linear free energy relationships, Geochimica et Cosmochimica Acta, 73, 1004-1013

Buda, R.A., Banik, N.L., Kratz, J.V., Trautmann, N. (2008) Studies of the ternary systems humic substances – kaolinite – Pu(III) and Pu(IV), Radiochimica Acta, 96, 657-665

Choppin G. R. (2003) Actinide speciation in the environment, Radiochimica Acta, 91, 645-649.

Delegard, C.H. (2011) Ostwald ripening and its effect on PuO2 particle size in Hanford tank waste, PNNL-20747, http://www.pnnl.gov/main/publications/external/technical_reports/PNNL-20747.pdf, 40 pp.

Felmy, A.R., Moore, D.A., Rosso, K.M., Qafoku, O., Rai, D., Buck, E.C., Ilton, E.S. (2011) Heterogeneous reduction of PuO2 with Fe(II): Importance of the Fe(III) reaction product, Environmental Science & Techology, 45, 3952-3958

Felmy, A.R., Moore, D.A., Qafoku, O., Buck, E., Conradson, S.D., Ilton, E.S. (2013) Heterogeneous 239 reduction of PuO2 by aqueous Fe(II) in the presence of hematite, Radiochimica Acta, 101, 701-710

Henrion, P.N., Monsecour, M., Fonteyne, A., Put, M., De Regge, P. (1985) Migration of radionuclides in Boom Clay. In: Radioactive Waste Management and the Nuclear Fuel Cycle, Volume 6 (3-4), Harwood Academic Publishers GmbH, pp. 313-359

Hixon, A.E., Powell, B.A. (2014) Observed changes in the mechanism and rates of Pu(V) reduction on hematite as a function of total plutonium concentration, Environmental Science & Technology, 48, 9255-9262

SCK•CEN/12201513 Page 159 of 208 Compilation of Technical Notes on less studied elements

Kersting, A.B., Efurd, D.W., Finnegan, D.M., Rokop, D.J., Smith, D.K., Thompson, J.L. (1999) Migration of plutonium in groundwater at the Nevada Test Site, Nature, 397, 56-59

Kersting, A.B. (2013) Plutonium transport in the environment, Inorganic Chemistry, 52, 3533-3546

Kim J. I. (1986): Chemical behaviour of Transuranic Elements in Natural Aquatic Systems. In: Handbook on the Physics and Chemistry of the Actinides, Eds. A. J. Freeman and c. Keller, Elsevier Science Publishers B. V., 1986, Chapter 8.

Kirsch, R., Fellhauer, D., Altmaier, M., Neck, V., Rossberg, A., Fanghanel, T., Charlet, L., Scheinost, A.C. (2011) Oxidation state and local structure of plutonium reacted with magnetite, mackinawite and chukanovite, Environmental Science & Technology, 45, 7267-7274

Lemire, R.J., Fuger, J., Nitsche, H., Potter, P., Rand, M.H., Rydberg, J., Spahiu, K., Sullivan, J.C., Ullman, W.J., Vitorge, P., Wanner, H. (2001) Chemical Thermodynamics of Neptunium and Plutonium, OECD- NEA, Issy-les-Moulineaux, France

Maes, N., Wang, L., Delécaut, G., Beauwens, T., Van Geet, M., Put, M., Weetjens, E., Marivoet, J., van der Lee, J., Warwick, P., Hall, A., Walker, G., Maes, A., Bruggeman, C., Bennett, D., Hicks, T., Higgo, J., Galson, D. (2004) Migration Case Study: Transport of radionuclides in a reducing Clay sediment (TRANCOM-II), Final Report. SCK•CEN-R-3790, 91 pp.

Marquardt, C.M., Seibert, A., Artinger, R., Denecke, M.A., Kuczewski, B., Schild, D., Fanghänel, Th. (2004) The redox behaviour of plutonium in humic rich groundwater, Radiochimica Acta, 92, 617-623

Marsac, R., Banik, N.L., Marquardt, C.M., Kratz, J.V. (2014) Stabilization of polynuclear plutonium(IV) species by humic acid, Geochimica et Cosmochimica Acta, 131, 290-300

Neck, V., Altmaier, M., Seibert, A., Yun, J.I., Marquardt, C.M., Fanghänel, Th. (2007) Solubility and redox reactions of Pu(IV) hydrous oxide: Evidence for the formation of PuO2+x(s, hyd), Radiochimica Acta, 95, 193-207

Novikov, A.P., Kalmykov, S.N., Utsunomiya, S., Ewing, R.C., Horreard, F., Merkulov, A., Clark, S.B., Tkachev, V.V., Myasoedov, B.F. (2006) Colloid transport of plutonium in the far-field of the Mayak production association, Russia, Science, 314, 638-641

Penrose, W.R., Polzer, W.L., Essington, E.H., Nelson, D.M., Orlandini, K.A. (1990) Mobility of plutonium and americium through a shallow aquifer in a semiarid region, Environmental Science & Technology, 24, 228-234

Powell, B.A., Duff, M.C., Kaplan, D.I., Field, R.A., Newville, M., Hunter, D.B., Bertsch, P.M., Coates, J.T., Eng, P., Rivers, M.L., Serkiz, S.M., Sutton, S.R., Triay, I.R., Vaniman, D.T. (2006) Plutonium oxidation and subsequent reduction by Mn(IV) minerals in Yucca Mountain Tuff, Environmental Science & Technology, 40, 3508-3514

Powell, B.A., Dai, Z., Zavarin, M., Zhao, P., Kersting, A.B. (2011) Stabilization of plutonium nano-colloids by epitaxial distortion on mineral surfaces, Environmental Science & Technology, 45, 2698-2703

Salah, S., Maes, N., Bruggeman, C. (2014) Uranium retention and migration in Boom Clay. Topical Report. First Full Draft, SCK•CEN-ER-305.

SCK•CEN/12201513 Page 160 of 208 Compilation of Technical Notes on less studied elements

Salah, S., Wang, L. (2014) Speciation and solubility calculations for waste relevant radionuclides in Boom Clay. First Full Draft, SCK•CEN-ER-198, 154 pp.

Santschi, P.H., Roberts, K.A., Guo, L. (2002) Organic nature of colloidal actinides transported in surface water environments, Environmental Science & Technology, 36, 3711-3719

Schmidt, M., Wilson, R.E., Lee, S.S., Soderholm, L., Fenter, P. (2012) Adsorption of plutonium oxide nanoparticles, Langmuir, 28, 2620-2627

Walther, C., Rothe, J., Brendebach, B., Fuss, M., Altmaier, M., Marquardt, C.M., Büchner, S., Cho, H.-R., Yun, J.-I., Seibert, A. (2009) New insights in the formation processes of Pu(IV) colloids, Radiochimica Acta, 97, 199-207

Xie, J., Lu, J., Lin, J., Zhou, X., Li, M., Zhou, G., Zhang, J. (2013) The dynamic role of natural colloids in enhancing plutonium transport through porous media, Chemical Geology, 360-361, 134-141

Xu, C., Santschi, P.H., Zhong, J.Y., Hatcher, P.G., Francis, A.J., Dodge, C.J., Roberts, K.A., Hung, C.-C., Honeyman, B.D. (2008) Colloidal cutin-like substances cross-linked to siderophore decomposition products mobilizing plutonium from contaminated soils, Environmental Science & Technology, 42, 8211-8217

Zavarin, M., Roberts, S.K., Hakem, N., Sawvel, A.M., Kersting, A.B. (2005) Eu(III), Sm(III), Np(V), Pu(V) and Pu(IV) sorption to calcite, Radiochimica Acta 93, 93-102

Zavarin, M., Powell, B.A., Bourbin, M., Zhao, P., Kersting, A.B. (2012) Np(V) and Pu(V) ion exchange and surface-mediated reduction mechanisms on montmorillonite, Environmental Science & Technology, 46, 2692-2698

Zavarin, M., Zhao, P., Dai, Z., Kersting, A.B. (2014) Plutonium sorption and precipitation in the presence of goethite at 25 and 80°C, Radiochimica Acta, 102, 983-997

SCK•CEN/12201513 Page 161 of 208 Compilation of Technical Notes on less studied elements

4.2.4 Technical note for Samarium (Sm)

4.2.4.1 General Samarium (Sm) is a metal, a lanthanide element (rare earth element, REE) with an atomic weight of 150.36 and an atomic number of 62 (CRC, 2011). There are 33 isotopes and isomers of Sm, with natural Sm being comprised of 7 isotopes. Three of these isotopes are unstable, albeit with very long half-lives. Sm has the common oxidation states +3 and +2.

There is relatively little data for Sm in natural waters, although REE as a group have been studied extensively. At 25°C in Cl-bearing solutions, the dominant Sm species is uncomplexed Sm3+ (Migdisov et al., 2008). However, in natural waters, the formation of negatively charged − dicarbonato complex (REE(CO3)2 ) accounts for significant fractions of each REE (e.g Johannesson et al., 1997). Wood (1990) reviewed the thermodynamic data available for the REE and Y and noted that trivalent REE exhibit strong, predominantly electrostatic complexing with "hard" ligands such as fluoride, sulfate, phosphate, carbonate and hydroxide. These metals form complexes only weakly with chloride and nitrate and extremely weakly or not at all with ammonia and bisulfide. At acidic pH simple ions and the sulphate complexes are most important, but complexes with carbonate become dominant at near- neutral to alkaline pH. Since these latter pH conditions tend to typify most groundwater, carbonate complexes are the most important inorganic REE-transporting species in groundwater. Samarium is found along with other lanthanide metals in several minerals, the principal ones being the Sm-bearing variety of monazite-(Sm) ((Sm,Gd,Ce,Th)(PO4)), samarskite (from which it was first isolated) and bastnaesite.

4.2.4.2 Speciation and solubility The predicted dominant species of Sm (as for Am) under BC conditions and using MOLDATA - is the second carbonate complex, i.e. Sm(CO3)2 (Figure 79 c). Speciation over the pH range 0 - - ~11.5 using the LLNL, ANDRA and MOLDATA TDB is the same, but at pH > 11.5, Sm(OH)4 represents the prevalent species when calculating with MOLDATA and ThermoChimie v.5, - while SmO2 is the dominant species using the LLNL TDB.

a) b) 1 1

.5 .5 +++ Sm+++ Sm

+ + SmCO3 SmCO3

- (volts) Eh - Eh (volts) Eh Sm(CO ) Sm(CO ) 0 3 2 0 3 2 - - SmO2 Sm(OH)4 µ µ

–.5 –.5

25°C 25°C 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 pH pH

SCK•CEN/12201513 Page 162 of 208 Compilation of Technical Notes on less studied elements

c) 1

.5 +++ Sm + SmCO3

Eh (volts) Eh Sm(CO )- - no graph - 0 3 2 - Sm(OH)4 µ

–.5

25°C 0 2 4 6 8 10 12 14 pH

Figure 79: Eh-pH diagram of samarium (Sm-C-S-O-H) for the BC reference porewater system. Assumed activity of dissolved [Sm] = 10-8. Diagram a) LLNL TDB, b) ANDRA, c) MOLDATA TDB, Code: The Geochemist's Workbench - 8.08/8.10

At [Sm] = 10-8, no solid phase is predicted to be stable. Increasing the Sm activity to [Sm] = -7 10 , SmOHCO3(s) becomes stable under BC conditions (Salah and Wang, 2014).

The solubilities calculated for different Sm-phases are summarized in the following table.

Table 28: Solubility of Sm in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV. Database: MOLDATA. Code: The Geochemist's Workbench- 8.08/8.10.

Solubility controlling phases Solubility, [Sm], mol/L -2 Sm(OH)3 (am) 4.9 × 10 -4 Sm(OH)3 (s) 3.7 × 10 -7 Sm2(CO3)3 (s) 5.0 × 10 -8 SmOHCO3 (s) 8.9 × 10 Source data: ANDRA TDB

The reaction constants for the Sm-solids comprised in (Table 18) and MOLDATA are the following:

+ 3+ Sm(OH)3(am) + 3 H ↔ Sm + 3 H2O log K = 18.6 + 3+ Sm(OH)3(s) + 3 H ↔ Sm + 3 H2O log K = 16.5 + 3+ - Sm2(CO3)3(s) + 3 H ↔ 2 Sm + 3 HCO3 log K = -3.52 + 3+ - SmOHCO3(s) + 2 H ↔ Sm + HCO3 + H2O log K = 2.63

In the Turva-2012 safety case, Posiva specified the solubility of Sm to be controlled by amorphous SmOHCO3:0.5H2O (Wersin et al., 2014). This solubility constraint was considered to be conservative, calculations using the Thermochimie thermodynamic of ANDRA having predicted that the less soluble phase SmOHCO3(cr) would be more likely to be solubility limiting in the waters considered by Posiva.

SCK•CEN/12201513 Page 163 of 208 Compilation of Technical Notes on less studied elements

Using SmOHCO3:0.5H2O as a solubility-limiting phase, a reference Sm concentration of 4.3×10-7 M was specified for saline water and 7.2×10-7 M for brackish water. An upper limit, taking into account estimated uncertainties was specified to be 1.1×10-5 M.

4.2.4.3 Sorption and retardation There are few sorption data for Sm and generally data for Eu(III), which is considered analogous to Sm, have been used (Bradbury and Baeyens, 2003; Mrabet et al., 2014; Wersin et al., 2014). Bradbury and Baeyens (2003) recommended a value of 4.7 m3/kg for near-field MX-80 bentonite. In contrast, Wersin et al. (2014) recommended values between 11 and 54 m3/kg for sorption on bentonite-bearing backfill in the presence of different reference waters for use in the Turva-2012 safety assessment. The smallest lower limit recommended by these workers is 1 m3/kg (for sorption in the presence of brine), whereas the largest upper limit is 316 m3/kg (for sorption in the presence of dilute, carbonate-rich water).

4.2.4.4 Diffusion and transport For use in the Finnish Turva-2012 safety assessment, Wersin et al. (2014) recommended that values of diffusion accessible porosity and an effective diffusion coefficient obtained from measurements of HTO transport should be used to assess Sm migration through a bentonite- -11 bearing backfill. This approach was considered to be conservative. A Deff value of 9×10 m2/s and an η of 0.38 were accordingly specified.

4.2.4.5 Justification Neither in-house sorption nor migration data are available for Sm. As for the trivalent actinides Pu(III), Am(III), and Cm(III), the dominant species under BC conditions is predicted to - correspond to the dicarbonate complex Sm(CO3)2 . Due to the anionic character, sorption and retardation should be rather low. But log Kd-values for Sm sorption on bentonite are reported to be high, i.e. ranging between 3 and 5.5 in the presence of different reference waters. Although the association of Sm with DOM was not experimentally evidenced, organic colloid mediated transport is also – by analogy with Am(III) and Eu(III) – put forward for Sm. Gaining more insight into the behaviour of trivalent lanthanides could be useful to support the grouping/reasoning.

4.2.4.6 References Bradbury M.H. and Baeyens B. (2003) Near-field sorption data bases for compacted MX-80 bentonite for performance assessment of a high-level radioactive waste repository in Opalinus Clay host rock. Nagra Technical Report NTB02-18.

Bruno, J. Cera, E., De Pablo, J. Duro, L., Jordana, S., and Savage, D. (1997) Determination of radionuclide solubility limits to be used in SR 97. Uncertainties associated to calculated solubilities. SKB Technical Report 97-33.

CRC (2011). Handbook of Chemistry and Physics, CRC Press, 92nd Edition.

SCK•CEN/12201513 Page 164 of 208 Compilation of Technical Notes on less studied elements

De Craen M., Wang L., Van Geet M. and Moors H. (2004). The geochemistry of Boom Clay pore water at the Mol site, status 2004. SCK•CEN-BLG-990.

Johannesson K.H., Stezenbach K.J. and Hodge V.F. (1997). Rare earth elements as geochemical tracers of regional groundwater mixing. Geochimica et Cosmochimica Acta, 61, 3605–3618.

Mrabet El S., Castro M.A., Hurtado S., Orta M.M., Pazos M.C., Villa-Alfageme M. and Alba M.D. (2014). Competitive effect of the metallic canister and clay barrier on the sorption of Eu3+ under subcritical conditions. Applied Geochemistry, 40, 25–31.

Migdisov A.A., Williams-Jones A.E., Normand C. and, Wood S.A. (2008). A spectrophotometric study of samarium (III) speciation in chloride solutions at elevated temperatures. Geochimica et Cosmochimica Acta, 72, 1611–1625.

Salah, S. and Wang, L. (2014) Speciation and solubility calculations for waste relevant radionuclides in Boom Clay. Scientific Report SCK•CEN-ER-198, SCK•CEN, Mol, Belgium.

Wersin P., Kiczka M., Rosch D., Ochs M. and Trudel D. (2014). Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto Radionuclide Solubility Limits and Migration Parameters for the Backfill. Posiva Report POSIVA 2012-40.

Wood S.A. (1990). The aqueous geochemistry of the rare-earth elements and yttrium: 1. Review of available low-temperature data for inorganic complexes and the inorganic REE speciation of natural waters. Chemical Geology, 82, 159-186.

SCK•CEN/12201513 Page 165 of 208 Compilation of Technical Notes on less studied elements

4.3 Group IVc: Tetravalent lanthanides and actinides (+ pentavalent Pa)

4.3.1 Technical note for Neptunium (Np)

4.3.1.1 General Neptunium belongs to the group of actinides and with its atomic number of 93 it represents the first transuranic element. The isotope 237Np (half-life of 2.14×106 years) is obtained as a by-product from nuclear reactors in the production of plutonium. Trace quantities of the element are actually found in nature due to transmutation reactions in uranium ores produced by the neutrons which are present. In total, 19 neptunium radioisotopes have been characterized, with the most stable being 237Np with a half-life of 2.14 million years, 236Np with a half-life of 154,000 years, and 235Np with a half-life of 396.1 days. All of the remaining radioactive isotopes have half-lives less than 4.5 days. Np exists in five oxidation states, i.e. + III, +IV, +V, +VI and +VII, but only the tetravalent and pentavalent states are important in natural waters (Langmuir, 1997).

4.3.1.2 Speciation and solubility According to MOLDATA_R2, the Np(IV) hydrolysis species Np(OH)4(aq) is predicted to be the dominant aqueous species under undisturbed Boom Clay conditions (Figure 80). Pirlet et al. (1998) and Pirlet and Van Iseghem (2003) studied the speciation of Np leached from doped glass in Boom Clay pore water and the redox stability of Np(IV) and Np(V) added directly to Boom Clay pore water. Neptunium redox speciation was quantified using UV-Vis spectroscopy. Unless redox conditions were not controlled (and more oxidising with respect to reference conditions), Np was mostly present in the +IV oxidation state in Boom Clay pore water (Pirlet et al., 1998; Pirlet and Van Iseghem, 2003). It is remarked that in these experiments, the initial oxidation state of Np in the glass was probably pentavalant. Therefore, Boom Clay pore water contains enough reducing capacity to ensure reduction of Np(V) to Np(IV), even when no reducing solid phases were present. The reduction is likely mediated by organic matter (Zeh et al., 1999).

As for other tetravalent actinides, colloid formation of Np(IV) is an important process which should be taken into account when evaluating the speciation of Np (Fanghänel and Neck, 2002; Zänker and Hennig, 2014), but only few experimental studies are available (Neck et al., 2001). Tetravalent neptunium is also observed to form complexes with humic and fulvic acids (Pirlet et al., 1998; Pirlet and Van Iseghem, 2003; Schmeide et al., 2005; Zeh et al., 1999). The actinides are bound to the carboxylic and possibly phenolic groups of the humics as inner- sphere complexes (Schmeide et al., 2005; Zänker and Hennig, 2014).

In Table 29, the solubilities of NpO2(cr) and (NpO2,am,hyd) are summarized. The former represents the most stable phase under BC conditions. However, considering kinetic aspects, the amorphous, hydrated neptunium oxide (NpO2,am,hyd) would most probably precipitate first and with time/ageing transform into the more stable and crystalline NpO2. Np(V)O2OH phases are characterized by very high solubilites under the reference conditions. They are reported to become more stable under more alkaline and oxidizing conditions (Berner, 2002). The Np(V) solid Np2O5(cr) is only stable under more oxidizing, i.e. atmospheric pO2(g) conditions (Duro et al., 2006) and is thus not considered to be relevant under BC conditions.

SCK•CEN/12201513 Page 166 of 208 Compilation of Technical Notes on less studied elements

++ NpO2 1

NpO (CO )-- + 2 3 2 NpO2 ---- NpO2(CO3)3 .5 +++ NpOH NpO (OH)- NpO CO- 2 3 ++ 2 3NpO (OH)-- Np(OH)2 2 4 ---- Eh (volts) +++ NpO (CO ) OH 0 Np + 2 3 2 Np(OH)3 µ Np(OH)4(aq) –.5 25°C 0 2 4 6 8 10 12 14 pH

Figure 80: Eh-pH diagram of neptunium (Np-C-S-O-H) for the BC reference porewater system. Assumed activity of dissolved Np = 10-8. Database: MOLDATA_R2. Code: The Geochemist's Workbench - 10.0.

Table 29: Solubility of Np in the BC reference porewater system at 25°C, pH 8.355 and Eh -281 mV. Database: MOLDATA_R2. Code: The Geochemist's Workbench - 10.0.

Solubility controlling phases Solubility, [Np], mol/L

-18 NpO2 (cr) 1.8 × 10 -9 NpO2 (am,hyd) 2.0 × 10

NpO2OH (am, fresh) very soluble

NpO2OH(am, aged) very soluble Source data: all solid phase data were taken from the NEA TDB.

Remark: NpO2 × 2 H2O is thermodynamically equivalent to NpO2(am,hyd) and was thus not considered in the calculations.

The reaction constants comprised in MOLDATA of the minerals mentioned in Table 29 are:

+ 4+ NpO2(cr) + 4 H ↔ Np + 2 H2O log K = -9.75 + 4+ NpO2 (am,hyd) + 4 H ↔ Np + 2 H2O log K = -0.70 + 4+ NpO2OH (am, fresh) + 4 H ↔ Np + 2.5 H2O + 0.25 O2(aq) log K = -5.98 + 4+ NpO2OH (am, aged) + 4 H ↔ Np + 2.5 H2O + 0.25 O2(aq) log K = -6.58

Pirlet et al. (2004) determined the solubility of Np in static leach tests with R7T7 reference glass doped with 237Np (0.34 wt%). The glass was contacted with FoCa7-clay together with pyrite, metallic iron (to ensure reducing conditions) and Boom Clay water. Np concentrations were measured in the leachates after ultrafiltration through 100 kDa membranes. After ~ 540 days duration, the Np concentration in the leachate reached 1.5×10-10 M.

SCK•CEN/12201513 Page 167 of 208 Compilation of Technical Notes on less studied elements

4.3.1.3 Sorption and retardation It is generally recognised that Np(IV) would sorb strongly on various minerals, including clay minerals, because of its strong hydrolysis. In the neutral to alkaline pH range surface complexation will dominate the sorption process. This process can be adequately described and predicted using the 2-site protolysis surface complexation/cation exchange model developed by Bradbury and Baeyens (Bradbury and Baeyens, 1997; Bradbury and Baeyens, 2009a). Although no studies have been performed on Np(IV) uptake by illite or montmorillonite at SCK•CEN, Np sorption may be predicted based on linear free energy relationships (Bradbury and Baeyens, 2005; Bradbury and Baeyens, 2009b).

Hart et al. (1994) studied the time-dependent uptake of Np on Boom Clay. Np was added as Np(V) in two initial concentrations, 8×10-7 M and 3×10-4 M, to a Boom Clay suspension of 20 g/L. The supernatant was deionized MilliQ water and the experiments were carried out in a glove box under N2 atmosphere, resulting in a final pH of ~10. Samples were taken until 28 days contact time of Np solution with the clay. Phase separation was performed by filtration at different cut-off: 0.45 µm, 0.1 µm, 100 kDa and 10 kDa. The results show a rapid sorption of Np for the experiment with initial 8×10-7 M and a clear influence of filtration cut-off on the measured Np concentrations (Figure 81). The calculated Rd values ranged from logRd ~ 3 (after 0.45 µm) until logRd ~5-6 (after 10 and 100 kDa). The effect of filtration cut-off was suggested to originate from an association of Np with colloids or organic matter in the experiments. In the experiments with higher initial Np, the influence of filtration was not clearly visible, perhaps as a result of residual Np(V) (data not shown).

Figure 81: Neptunium activity in filtrates and ultrafiltrates. Initial Np 8×10-7 M (or 5 Bq/mL) (Hart et al., 1994)

Nagasaki et al. (1999) studied sorption of 1×10-7 M Np(IV) on bentonite between pH 6 and -3 10 at 0.01 M NaClO4 and total carbonate 1×10 M under anoxic conditions. A maximum Kd of 2.7×104 L/kg was observed around pH 8.5. The strong competition between dissolved humic substances and clay minerals for Np(IV) was evidenced by Schmeide and Bernhard (2010), who studied sorption of Np onto kaolinite as a function of pH, ionic strength, carbonate and humic acid.

SCK•CEN/12201513 Page 168 of 208 Compilation of Technical Notes on less studied elements

4.3.1.4 Transport and diffusion Two percolation experiments (type C4) and one sequential migration experiment have been performed with Np in confined Boom Clay cores. Details of these experiments can be found below:

1. Percolation C4 experiment Np237/1/5a (NRM006A) • Initiated 01/07/1993; under Ar; from 28/10/1996 under Ar + 0.04% CO2 • Clay core code MIG 09+11 coring 11/05/1992 R88 5.0-5.3 + 6.0-6.3 (vertical) • Initial activity 45 kBq, in chemical form Np in hydrochloric acid • Np concentration in source solution 1.45×10-2 M • Total clay core length: 72 mm (45.5 mm "inlet" + 26.5 mm "outlet"), diameter 38 mm • Used as input for sequential migration experiment 18/01/2006

2. Percolation C4 experiment Np237/1/6 (NRM006B) • Initiated 01/07/1993; under Ar; from 28/10/1996 under Ar + 0.04% CO2 • Clay core code MIG 05 coring 11/05/1992 R88 3.0-3.3 (vertical) • Initial activity 66 kBq, in chemical form Np in 2.2 N HNO3 • Np concentration in source solution 1.06×10-1 M • Total clay core length: 72 mm (40 mm "inlet" + 32 mm "outlet"), diameter 38 mm • Percolated solutions followed until now

3. Sequential migration experiment Np237/1/5b

• Initiated 19/01/2006; under Ar + 0.04% CO2 • Mounted after clay core Np237/1/5a • Total clay core length: 104 mm (72 mm first core + 32 mm second core), diameter 38 mm

The hydraulic conductivity, K (m/s), for the three experiments is given in Figure 82. The value for K is in line with those normally observed in Putte Member of Boom Clay (Yu et al., 2013). In all cores it is observed that the hydraulic conductivity is slowly decreasing in time, which is presently unexplained. Possibly, dissolved organic matter obstruction in the first filter before entering the clay core may be a possible source for this observation.

The two percolation experiments show a similar breakthrough pattern (Figure 83 and Figure 84). Breakthrough of Np occurs fairly rapidly in both experiments but concentrations in the percolate keep increasing steadily, even after approximately 20 years of percolation (> 1 liter of percolated water). The observed concentration range is in the order of the solubility of NpO2(am,hyd). The very rapid breakthrough is in contrast with high sorption (retardation) commonly associated with Np(IV) in clay-rich environments. The breakthrough curves show some striking similarities with the ones observed for uranium in similar percolation experiments and may confirm the tetravalent neptunium form as most prominent valence state.

In the sequential migration experiment, the Np elution curve out of the second clay core exhibits similar features as the elution out of the first core. Breakthrough occurs quite rapidly and then a steady increase in percolated Np concentration is observed. The increase itself seems to be a bit faster compared to the percolation from the first core. Probably, the high initial Np load in the first core, in a valence state that is not in equilibrium with the Boom Clay, results in a different kinetic behaviour compared to the second core, where the input SCK•CEN/12201513 Page 169 of 208 Compilation of Technical Notes on less studied elements

Np is assumed to be in equilibrium with the Boom Clay. At the moment, it is also unclear whether the observed Np concentration percolated out of the second core will level off to a constant value, or keeps increasing towards the same concentration as percolated from the first core.

Hydraulic Conductivity 2.00 E-12

1.80 E-12

1.60 E-12

1.40 E-12

1.20 E-12

1.00 E-12

8.00 E-13

6.00 E-13 Np237m1c5 Hydraulic conductivity K conductivity (m/s) Hydraulic

4.00 E-13 Np237m1c6

Np237m1c5 SM 2.00 E-13

0.00 E+00 0 1000 2000 3000 4000 5000 6000 7000 8000 Days since start experiment

Figure 82: Calculated hydraulic conductivity, K (in m/s), as function of time in 2 percolation experiments (Np237m1c5 and Np237m1c6) and 1 sequential migration experiment (Np237m1c5 SM)

Np concentration outlet 16

8 237 (Bq/L)

- 6

Np Np237m1c5 4 Np237m1c6 2 Np237m1c5 SM

0 1000 2000 3000 4000 5000 6000 7000 8000 -2

Days since start experiment

Figure 83: Np concentration in outlet (237Np, in Bq/L) as function of time

SCK•CEN/12201513 Page 170 of 208 Compilation of Technical Notes on less studied elements

Np concentration outlet

2.50 E-09 Np237m1c5

Np237m1c6 2.00 E-09 Np237m1c5 SM

1.50 E-09

1.00 E-09 Np concentration (mol/L) Np concentration

5.00 E-10

0.00 E+00 .0 200.0 400.0 600.0 800.0 1000.0 1200.0 Volume percolated since start experiment (mL)

Figure 84: Np concentration in outlet (mol/L) as function of percolated volume (mL)

The results of these experiments can be interpreted as resulting from Np-dissolved organic matter (DOM) colloid facilitated transport. The Np species which are eluted from the (first) clay core in the percolation experiments are Np-DOM species, while the majority of applied Np is retained in the source position (either as precipitate or adsorbed onto the solid phases). The elution behaviour through the second clay core implies that the Np-DOM species are only very slowly dissociating. The dissociated species are subsequently retained in the second clay core. The results are in line with published literature on humic colloid-mediated transport of Np(IV) species in sedimentary environments (Artinger et al., 2000; Artinger et al.; 2003).

In Maes et al. (2011) a transport model was constructed based on the conceptual model described in Figure 85. Radionuclides in solution will either be present as a mobile RN-DOM complex or "free inorganic" radionuclide species in solution ([RNinorg]liquid). The transfer between [RNinorg]liquid and the RN-DOM complex is described by a complexation constant and dissociation kinetics. Both species can interact with the solid phase. It is assumed that this interaction in case of [RNinorg]liquid is mainly due to sorption processes and can be described by a retardation factor (RRN) that can be linked to batch sorption data. In case of RN-DOM the retardation factor (RRN-DOM) is considered as lumped factor accounting for both sorption and colloid filtration processes. Overall, dissolved OM is only poorly retarded within Boom Clay and RRN-DOM is therefore expected to be only of secondary importance to describe RN- coupled transport. Within this transport model, the amount of parameters remains limited and most of them can be obtained from independent measurements (batch complexation/solubility experiments, batch sorption experiments, DOM transport experiments).

SCK•CEN/12201513 Page 171 of 208 Compilation of Technical Notes on less studied elements

mobile RN-DOM complex

kdecomp [ RNinorg ]liquid RN-DOM

kcomp

Linear RRN R Sorption RN-DOM

RN RN-DOM Boom Clay solid phase

Figure 85: Conceptual model used for the interpretation of organic matter linked radionuclide migration in Boom clay (Maes et al., 2011)

The transport of the RN-DOM mobile complex is described by: ∂()c ηρ+ m −∇⋅η ∇ + =−ληρ + + η ()bK d,, m Dpore m cV m Darcy c m ()b K d, m c m Q sol− OM ∂t

The transport of the RN-species in solution is described by: ∂()c ηρ+ s −∇⋅η ∇ + =−ληρ + − η ()bK d,, s Dpore s cV s Darcy c s ()b K d, s c s Q sol− OM ∂t

The mass transfer of the RN between the DOM complexed form and the "free" RN in solution is given by: Q= k cc − k c sol− OM comp s OM decomp m

To constrain the degrees of freedom present in the model, the RN-DOM interaction constant (logKRN-DOM) was fixed during fitting. Two values were tested, one which would resemble the interaction constant of Tc(IV) with DOM (logKRN-DOM = 5.3), and the other which would be

SCK•CEN/12201513 Page 172 of 208 Compilation of Technical Notes on less studied elements

more representative of Th(IV) interaction with DOM (logKRN-DOM = 6.1). The experiments were therefore fitted for 3 variables: RRN, RRN-DOM, and kdecomp. The model provides excellent fits (Figure 86) to the experimental data and the fitted parameters are presented in Table 30 (Maes et al., 2011).

Table 30: Summary of the fitted parameters kdecomp, RRN-DOM and RRN (Maes et al., 2011)

RN-DOM -1 logK kdecomp [s ] RRN-DOM [-] logRRN [-]

Np 5.3a 0.7±0.2×10-6 27±4 3.61±0.14

6.1b 0.7±0.2×10-6 27±4 4.40±0.14

aUsing the derived logK value for Tc(IV) as analogue b Using the derived logK value for Th(IV) as analogue

The fitted parameters (decomplexation constants, RN-DOM and RN retardation factors) are RN-DOM within a narrow range. Sensitivity analysis showed that RRN and K are the most influential parameters in the model and that they are correlated. If KRN-DOM is increased in the model, the fitted value for RRN will be higher (or vice versa) to obtain an equally good fit. By contrast, the model fit is not very sensitive to other fit parameters.

Figure 86: Results of the fitting of the Np elution curve in a sequential migration xperiment using the proposed conceptual model (Maes et al., 2011)

The results from the percolation and sequential migration experiments with Np are therefore completely in line with the proposed phenomenological models for Tc(IV) (Bruggeman et al., 2010) and U(IV) (Salah et al., 2014). The good agreement between the different tetravalent actinides and Tc(IV) regarding their affinity for humic colloid-mediated transport was previously also noted by Artinger et al. (2003).

SCK•CEN/12201513 Page 173 of 208 Compilation of Technical Notes on less studied elements

4.3.1.5 Justification The justification for Np is quite straightforward, as there are sorption data as well as migration data available, that revealed the strong affinity of Np(IV) for DOM, mediating its transport through BC. At the same time, sorption on BC was shown to be strong, as expected from the speciation calculations that predict Np to occur as neutral hydrolysis species (Np(OH)4,aq). Due to the former, Np represents one of the elements that have been used to develop and constrain the conceptual model put forward by Maes et al. (2011).

4.3.1.6 References Artinger, R., Marquardt, C.M., Kim, J.I., Seibert, A., Trautmann, N., Kratz, J.V. (2000) Humic colloid-borne Np migration: influence of the oxidation state, Radiochimica Acta, 88, 609-612

Artinger, R., Buckau, G., Zeh, P., Geraedts, K., Vancluysen, J., Maes, A., Kim, J.I. (2003) Humic colloid mediated transport of tetravalent actinides and technetium, Radiochimica Acta, 91, 743-750

Berner, U. (2002) Project Opalinus Clay: Radionuclide concentration limits in the cementitious near- field of an ILW repository, PSI Bericht Nr. 02-26, Nagra NTB 02-22.

Bradbury, M.H., Baeyens, B. (1997) A mechanistic description of Ni and Zn sorption on Na- montmorillonite. Part II: modelling, Journal of Contaminant Hydrology, 27, 223-248.

Bradbury, M. H. and Baeyens, B. (2009a) Sorption modelling on illite. Part I: Titration measurements and the sorption of Ni, Co, Eu and Sn, Geochimica et Cosmochimica Acta, 73, 990-1003

Bradbury, M.H. and Baeyens, B. (2009b) Sorption modelling on illite. Part II: Actinide sorption and linear free energy relationships, Geochimica et Cosmochimica Acta, 73, 1004-1013

Bruggeman, C., Maes, N., Aertsens, M., Govaerts, J., Martens, E., Jacops, E., Van Gompel, M., Van Ravestyn, L. (2010) Technetium retention and migration behaviour in Boom Clay. Topical Report, First full draft, SCK•CEN-ER-101, SCK•CEN, Mol, Belgium, 120 pp.

Duro, L., Grivé, M., Cera, E., Doènech, C., Bruno, J. (2006) Update of a thermodynamic database for radionuclides to assist solubility limits calculation for performenace assessment. SKB Technical Report, TR-06-17, 128 pp.

Fanghänel, Th., Neck, V. (2002) Aquatic chemistry and solubility phenomena of actinide oxides/hydroxides, Pure and Applied Chemistry, 74, 1895-1907

Hart, K.P., Payne, T.E., Robinson, B.J., Van Iseghem, P. (1994) Neptunium uptake on Boom Clay – Time dependence and association of Np with fine particles, Radiochimica Acta, 66/67, 19-22

Langmuir (1997) Chapter 13. In: Aqueous Environmental Geochemistry. Upper Saddle River, NJ: Prentice Hall.

SCK•CEN/12201513 Page 174 of 208 Compilation of Technical Notes on less studied elements

Nagasaki, S., Tanaka, S., Suzuki, A. (1999) Sorption of neptunium on bentonite and its migration in geosphere, Colloids and Surfaces A: Physicochemical and Engineering Aspects, 155, 137-143

Neck, V., Kim, J.I., Seidel, B.S., Marquardt, C.M., Dardenne, K., Jensen, M.P., Hauser, W. (2001) A spectroscopic study of the hydrolysis, colloid formation and solubility of Np(IV), Radiochimica Acta, 89, 439-446

Pirlet, V., Van Iseghem, P., Dierckx, A., Desreux, J.-F. (1998) The investigation of the neptunium complexes formed upon interaction of high level waste glass and Boom Clay media, Journal of Alloys and Compounds, 271-273, 267-271

Pirlet, V., Van Iseghem, P. (2003) Neptunium speciation in humic acid-rich clay water upon interaction with radioactive waste glass samples, Scientific Basis for Nuclear Waste Management XXVI, In: Materials Research Society Symposium Proceedings, Finch, R.J., Bullen, D.B., Eds., 757, 465-476

Pirlet, V., Lemmens, K., Van Iseghem, P. (2004) Influence of the near-field conditions on the mobile concentrations of Np and Tc leached from vitrified HLW, Scientific Basis for Nuclear Waste Management XXVIII, In: Materials Research Society Symposium Proceedings, Hanchar, J.M., StroesGascoyne, S., Browning, L., Eds., 824, 385-390

Salah, S., Bruggeman, C., Maes, N. (2014) Uranium retention and migration behaviour in Boom Clay. Topical Report, First full draft (Status 2014), in preparation (https://nirond-km.be/gm/folder- 1.11.482111)

Schmeide, K., Reich, T., Sachs, S., Brendler, V., Heise, K.H., Bernhard, G. (2005) Neptunium(IV) complexation by humic substances studied by X-ray absorption fine structure spectroscopy, Radiochimica Acta, 93, 187-196

Schmeide, K., Bernhard, G. (2010) Sorption of Np(V) and Np(IV) onto kaolinite: Effects of pH, ionic strength, carbonate and humic acid, Applied Geochemistry, 25, 1238-1247

Zänker, H., Hennig, C. (2014) Colloid-borne forms of tetravalent actinides: A brief review, Journal of Contaminant Hydrology, 157, 87-105

Zeh, P., Kim, J.I., Marquardt, C.M., Artinger, R. (1999) The reduction of Np(V) in groundwater rich in humic substances, Radiochimica Acta, 87, 23-28

SCK•CEN/12201513 Page 175 of 208 Compilation of Technical Notes on less studied elements

4.3.2 Technical note for Thorium (Th)

4.3.2.1 General Thorium belongs to the actinide series of the periodic table and its atomic number is 90. All natural thorium is found as 232Th, which decays to stable 208Pb with a half-life of 1.4 × 1010 years. The abundance of Th in nature is estimated to be 3-4 times as high as uranium. Thorium has around thirty radioisotopes with the most stable (after 232Th) being 230Th with a half-life of 75,380 years (daughter product of 238U), 229Th with a half-life of 7,340 years, and 228Th with a half-life of 1.92 years. All of the remaining radioactive isotopes have half-lives that are less than one month. Thorium metal is a source of nuclear power. Unlike natural uranium, which contains ~0.7% ‘fissile’ 235U isotope, natural thorium does not contain any 'fissile' material and is made up of the 'fertile' 232Th isotope only. During the pioneering years of nuclear energy, i.e. from the mid 1950's to mid 1970's, there was considerable interest worldwide to develop thorium fuels and fuel cycles in order to supplement uranium reserves. Due to the discovery of new U-deposits and their improved availability during the 80s and 90s, the interest declined despite the fact that the feasibility to use Th in different reactor types could be demonstrated. However, in recent times, the need for proliferation-resistance, longer fuel cycles, higher burnup improved waste form characteristics, reduction of plutonium inventories and in-situ use of bred-in fissile material has led again to renewed interest in thorium-based fuels and fuel cycles in several developed countries (IAEA, 2005). In Belgian power plants however, Th-based fuels were never used and are not foreseen to be used in future. Thorium is present in several silicates and oxides, such as for example thorite (ThSiO4) and thorianite (ThO2), and occurs also in traces in different heavy minerals, e.g. monazite and zircon.

4.3.2.2 Speciation and solubility In Figure 87, the Th-speciation as function of pH (in absence of carbonates) is illustrated. It can be seen that the onset of Th-hydrolysis occurs already at pH< 2. At low Th- concentrations (Figure 87 a), only the different hydrolysis species are predicted to be present in solution, while at higher Th-concentrations, e.g. 1×10-3 M, also polynuclear species, such as 4+ Th4(OH)12 may form.

The predicted dominant aqueous species under BC reference conditions are mixed hydroxo- - carbonate species, i.e. Th(OH)3(CO3) (131-complex; see Figure 88 a). It should be mentioned that the latter species was not included in the NEA TDB (Rand et al., 2009; Vol. 11). But based on the MOLDATA strategy, i.e. to strive for completeness, this species was copied from the ANDRA TDB and included in MOLDATA. Using the NEA TDB for the speciation calculations, 2- then Th(OH)2(CO3)2 (122-complex) represents the prevalent species (see Figure 88 b).

According to calculations with the LLNL TDB, Th(OH)4(aq) is predicted to be the predominant complex under BC representative conditions (see Figure 89).

SCK•CEN/12201513 Page 176 of 208 Compilation of Technical Notes on less studied elements

a) b) 1.1e–8 .0011

Th(OH) (aq) ++++ Th(OH)4(aq) 1e–8 Th++++ 4 .001 Th

9e–9 9e–4

8e–9 8e–4

+ (molal) (molal) Th(OH)3 7e–4 7e–9 +++ ++++

++++ Th(OH) ++ Th(OH)2 +++ 6e–9 6e–4 Th(OH)

5e–9 5e–4

4e–9 4e–4

3e–9 3e–4 ++++ Th4(OH)12 Some species w/ Th Some species w/ Th 2e–9 2e–4 ++ Th(OH)2 1e–9 1e–4 + Th(OH)3 0 0 2 3 4 5 6 7 8 9 10 11 12 2 3 4 5 6 7 8 9 10 11 12 pH pH

Figure 87: Thorium speciation in 0.01 M NaCl as function of pH in absence of carbonates. a) [Th] = 1×10-8 M, b) [Th] = 1×10-3 M. Database: Moldata_R2.Code: Geochemist's Workbench 10.0.

a) b) 1 1

++ ++ ThSO4 ThSO4 .5 .5 ++++ ++++ Th Th(OH)++ Th ++ 2 Th(OH)2

-- - Th(OH)2(CO3)2 Eh (volts) Eh (volts) Th(OH) (CO ) 0 3 3 0 Th(OH)+++ Th(OH)+++ Th(OH) (aq) Th(OH) (aq) 4 µ 4 µ

–.5 –.5 25°C 25°C 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 pH pH

Figure 88: Eh-pH diagram of thorium (Th-C-S-O-H) for the BC reference porewater system. Assumed activity of dissolved [Th] = 10-8. Database: a) MOLDATA_R2, b) NEA TDB. Code: The Geochemist's Workbench - 10.0.

ThSO++ .5 4 Th++++ ++ Th(OH)2

Eh (volts) Eh Th(OH) (aq) 0 4

–.5

25°C 0 2 4 6 8 10 12 14 pH Figure 89: Eh-pH diagram of thorium (Th-C-S-O-H) for the BC reference porewater system. Assumed activity of dissolved [Th] = 10-8. Database: LLNL TDB. Code: The Geochemist's Workbench - 10.0.

SCK•CEN/12201513 Page 177 of 208 Compilation of Technical Notes on less studied elements

Solubility calculations were performed for the phases comprised in Table 31. It can be seen that the solubilties vary over several orders of magnitude.

Table 31: Solubility of Th in the BC reference porewater system at 25 °C, pH 8.355 and Eh -281 mV. Code: The Geochemist's Workbench- 10.0. Database: MOLDATA_R2

Solubility controlling phases Solubility, [Th], mol/L

-5 ThO2 (am,hyd,fresh) 1.6 × 10 -6 ThO2 (am,hyd,aged) 2.6 × 10 -13 Thorianite (ThO2,cr) 4.7 × 10 Source data: all solid phase data were copied from the NEA TDB (Rand et al., 2008, Vol. 11).

The reaction constants for the thorium solids comprised in (Table 31) and MOLDATA_R2 are the following:

+ 4+ ThO2 (am,hyd,fresh) + 4 H ↔ Th + 2 H2O log K = 9.3 eq.1 + 4+ ThO2 (am,hyd,aged) + 4 H ↔ Th + 2 H2O log K = 8.5 eq.2 + 4+ Thorianite (ThO2,cr) + 4 H ↔ Th + 2 H2O log K = 1.8 eq.3

Table 32: Species distribution of Th in equilibrium with ThO2(am,hyd,aged). Database: MOLDATA_R2. Code: The Geochemist's Workbench- 10.0.

Aqueous species U [mol/l] Percentage [%]

- -6 Th(OH)3(CO3) 2.44 × 10 95.1

2- -7 Th(OH)2(CO3)2 1.12 × 10 4.4

4- -8 Th(CO3)4 1.02 × 10 0.4

4- -9 Th(OH)2(CO3)3 1.57 × 10 0.1 Total 2.6 × 10-6 100

It can be seen from Table 32, that the predominant species under BC representative conditions in equilibrium with ThO2(am,hyd,aged) corresponds to the 131-complex - Th(OH)3(CO3) .

The ThO2(cr) solubility in SBCW in absence and at various concentrations of HA was investigated by Delécaut et al. (2004). Experiments were performed in a glovebox with controlled Ar-5% H2-0.4%CO2 atmosphere (<1 ppm O2). The Th-concentrations measured after micro- (0.45µm) and ultrafiltration (30kDa) were below detection limit, i.e. 4.3×10-9 M (1 ppb).

In the absence of HA, the Th-concentrations were identical after micro- and ultrafiltration, revealing that there was no formation of real/intrinsic Th-colloids (Figure 90).

In the solubility experiments with SBCW and different HA (Boom Clay Humic Acid, BCHA; and Aldrich Humic Acid, AHA) concentrations (i.e. 0-900 mg C/L), the Th-concentrations after ultrafiltration (2 nm) were determined to be nearly constant over the studied HA range (see Figure 90) and not markedly increased compared to the solubility "measured" in absence of

SCK•CEN/12201513 Page 178 of 208 Compilation of Technical Notes on less studied elements

HA. Therefore, the solubility of ThO2(cr) was interpreted to be independent of the TOC concentrations after ultrafiltration and that dissolved HA did not form soluble organic complexes with Th(IV). No difference was observed between the tests using BC extracted HA (BCHA) and the ones with AHA. The colloidal Th-concentration was however about 3 orders of magnitude higher (i.e. ~7×10-6 M) than the solubility after ultrafiltration at 30 kDa (see Figure 90). The decrease in TOC concentration after UF, revealed the colloidal behavior of HA and the formation of organic Th(IV) pseudocolloids.

Figure 90: Th-concentrations in SBCW after 0.45 µm and 2 nm (ultra)filtration as function of the initial TOC concentrations Dashed line gives detection limit (i.e. 4.3×10-9 M; 1 ppb). Copied from Delécaut (2004; p. 132).

Main conclusions:

 Real Th-colloids do not seem to form and affect (i.e. increase) Th-solubility.  Solubility is not influenced by the formation of "dissolved" Th-HA complexes, but presence of organic pseudocolloids may increase the Th-solubility by 3 orders of magnitude.  No difference between BCOM and AHA "interaction/complexation" behaviour.

Liu et al. (unpublished data) studied the solubility ThO2(cr) in absence (SBCW) and presence of dissolved organic matter (RBCW, Aldrich HA) under BC representative conditions (logpCO2 = -2.4, pH = 8.3). The experiments were performed from undersaturation direction by dissolving 0.1 g ThO2(cr) in 25 ml RBCW/AHA with different TOC concentrations (~1.5-120 ppm) by adding appropriate amounts of SBCW for dilution. Results of the SBCW experiments (i.e. no OM) revealed solubilities of 5×10-11 M, 4×10-10 M and 8×10-8 M after 30 000 MWCO, 300 000 MWCO and 0.45 µm filtration, respectively.

The ThO2(cr) solubility given in the literature at near-neutral conditions (in absence of carbonates) range between ~10-17 and 10-14 M. The calculated solubility using MOLDATA_R2 -13 for ThO2(cr) under BC conditions corresponds to 4.7×10 M.

The experimentally determined solubilities in SBCW (i.e.absence of dissolved organic matter) are several orders of magnitude higher than the calculated solubility, which is most likely

SCK•CEN/12201513 Page 179 of 208 Compilation of Technical Notes on less studied elements

related to the presence of amorphous "intrinsic/real" Th-colloids. This interpretation is consistent with many observations, where experimental solubility data exceeding the thermodynamic solubility of ThO2(cr) are referred to the slow dissolution kinetics on the one hand and to the dissolution of small amounts of amorphous parts present within the crystal or at its surface on the other hand, leading to solubilities that approach those of the amorphous solid. In absence of carbonates, the latter is reported to range between ~10-8 – 10-9 M (1-10 nM), which is close to the experimentally determined solubility after ultrafiltration (30 000 MWCO/30 kDa). However, the experiments were performed in presence of carbonates, which may explain the slightly higher measured Th-concentration (8×10-8 M).

In MOLDATA_R2, two amorphous Th-phases are comprised, i.e. amorphous hydrated fresh -5 -6 ThO2 and amorphous aged ThO2, for which solubilties of 1.6×10 M and 2.6×10 M, respectively were calculated under BC representative conditions. These latter values are 2-3 orders of magnitude higher than the measured Th-concentration after ultrafiltration. Although the interpretation of the experiments has not been finalized up to now, it is - thought that from database point of view, the stability of the Th(OH)3(CO3) species, which are predicted to be the dominant ones (see above, Table 31) might be overestimated, resulting in too high calculated solubilities.

In presence of dissolved OM, the measured Th-concentrations increased with increasing TOC concentrations (see Figure 91:). Compared to the solubilities measured in absence of DOM, 3- 4 times higher Th-concentrations (up to 9.3×10-5 M, after 0.45 µm filtration) were determined in experiments with dissolved OM. The same dependence of solubility on the operational size cut-off in presence of DOM (as in absence of DOM), clearly points to Th-HA (30 000 MWCO) and Th-humic colloid interaction processes (besides intrinsic colloid formation).

1.E-04

1.E-05

1.E-06

1.E-07

1.E-08

concentration [mol/L] 30 000 MWCO - 1.E-09 Th 300 000 MWCO 1.E-10 0.45 µm

1.E-11 0 20 40 60 80 100 120 140

TOC [ppm]

Figure 91: ThO2(cr) solubilities as function of different TOC concentrations (Liu et al., unpublished data)

High-speed centrifugation and ultracentrifugation were used to study in more detail the effect of colloidal Th (i.e. organic and inorganic), as well as particle size effects on the ThO2(cr) solubility.

For more details it is referred to Salah and Durce (2014).

SCK•CEN/12201513 Page 180 of 208 Compilation of Technical Notes on less studied elements

Main conclusions:

 Real Th-colloids may increase ThO2(cr) solubility by several orders of magnitude compared to calculated thermodynamic solubility with a strong dependency on the operational size cut-off.

 ThO2(cr) solubility in presence of DOM is additionally increased by the formation of "dissolved" Th-HA complexes, as well as organic pseudocolloids. Same dependency on operational size cut-off as in absence of DOM was observed.

 Carbonates may also increase the ThO2(cr) solubility, but the presence of DOM has a much more pronounced effect.

Th-DOM interaction

The interaction of Th(IV) with DOM can be represented by the following equation (eq. 4):

inorgTh(IV)(aq) + DOM ⇔ Th-DOM eq. 4

In this equation, inorgTh(IV)(aq) refers to all inorganic species of Th(IV), and DOM refers to the dissolved natural organic matter concentration. In Maes et al. (2003), this concentration is expressed as eq/L, which is obtained by multiplying the dissolved organic carbon (DOC or TOC) concentration by 1.8 (the conversion factor from mg C/L to mg DOM/L, taking into account that the dissolved OM consists primarily of humic acids which have a mean C content of 55 w% (Stevenson, 1982) and by 2.9×10-3 (the charge of the DOM colloids, in eq/g, at pH 7 (Trancom-Clay, 2000).

DOM [eq/L] = DOC [g/L] × 1.8 × 2.9×10-3 [eq/g] eq. 5

Finally, Th-DOM refers to the resulting Th-containing organic pseudocolloid concentration.

Based on the ThO2(cr) solubility data of Liu et al. (unpublished, see section 3.2) a generalised Th-DOM interaction constant KTh-DOM can be obtained as described in the following. The aqueous Th equilibrium concentration at 0 ppm DOM concentration (S0, solubility for conditions without OM) can be described as follows:

= [ ] + [ ] , [ ] eq. 6 4+ 4+ 𝑛𝑛 0 𝑙𝑙 𝑛𝑛 Assuming that 𝑆𝑆the aqueous𝑇𝑇ℎ speciation𝑇𝑇ℎ ∑ 𝐾𝐾is dominated𝐿𝐿 by one species (i.e. Thinorg) this equation simplifies to:

[ ] eq. 7

𝑆𝑆0 ≈ 𝑇𝑇ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖

SCK•CEN/12201513 Page 181 of 208 Compilation of Technical Notes on less studied elements

The Th-concentration in presence of DOM (SOM, Solubility at a given OM concentration) can be described as follows, if it is considered that the main inorganic species interacts with the DOM:

= + [ ] [ ] eq. 8

𝐷𝐷𝐷𝐷𝐷𝐷 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑇𝑇ℎ−𝐷𝐷𝐷𝐷𝐷𝐷 Rearrangement gives:𝑆𝑆 �𝑇𝑇ℎ � 𝑇𝑇ℎ 𝐾𝐾 𝐷𝐷𝐷𝐷𝐷𝐷

= 𝑆𝑆 [𝐷𝐷𝐷𝐷𝐷𝐷 ] eq. 9 � 𝑆𝑆0 −1� 𝐾𝐾𝑇𝑇ℎ−𝐷𝐷𝐷𝐷𝐷𝐷 𝐷𝐷𝐷𝐷𝐷𝐷 This approach yielded an interaction constant, KTh-DOM with a value of logKTh-DOM equal to 6.1 ± 0.2 (Maes et al., 2011). In analogy to Tc, this type of interaction was labeled "hydrophobic sorption" or "colloid-colloid interaction" to distinguish it from complexation reactions between negatively-charged NOM ligands and metal or RN cations. The exact nature and driving force of the interaction mechanism (Van der Waals forces, H-bridging), remains however unknown.

Due to the nature of this mechanism, and to distinguish it from a complex formation mechanism with charged functional groups on the humic molecule, it is recommended to express the DOM concentration in g DOM/L instead of eq/L. Thus, the value of logKTh-DOM would change from 6.1 ± 0.2 (DOM in eq/L) to 3.6 ± 0.2 (DOM in g/L).

Concluding, it is assumed that the interaction between Th(IV) and DOM can be described – analogous to Tc(IV) - by the interaction mechanism given in equation 4 with a best estimate of logKTh-DOM equal to 3.6 ± 0.2 (DOM in g/L).

Literature data

PSI/NAGRA data evaluation

It is generally accepted that in absence of carbonates the predominant hydrolysis product of Th at pH > 6 is Th(OH)4(aq) and according to Hummel et al. (2002, p. 376), the most important solid that needs to be considered in systems relevant to waste management is ThO2(s). The equation (i.e. eq. 12) enabling to predict the solubility > pH 6 can be derived from the following equilibria:

+ 4+ * ThO2(s) + 4 H ⇔ Th + 2 H2O log10 Ks,0 = 9.9 ±0.8 eq. 10 4+ + * Th + 4 H2O ⇔ Th(OH)4(aq) + 4 H log10 β14 = -18.4 ± 0.6 eq. 11 with the total dissolved thorium concentration given by the sum of equations 10 and 11.

* ° * * log10 Ks,4 = log10 Ks,0 + log10 β14 eq. 12

* ° ThO2(s) + 2 H2O ⇔ Th(OH)4(aq) log10 Ks,4 = -8.5 ± 0.6 eq. 13

* ° The selected log10 Ks,4 (i.e. -8.5±0.6) corresponds to a value proposed by Neck and Kim (1999) which was confirmed by data of several other authors (Ryan & Rai, 1987; Moon, 1989; Felmy et al., 1991, Serne et al., 1996; Wierczinsky et al., 1998). The formation constant of the * mononuclear Th(IV) hydrolysis species, i.e. log10 β14 = -18.4±0.6 has been based on

SCK•CEN/12201513 Page 182 of 208 Compilation of Technical Notes on less studied elements

potentiometric studies of Ekberg et al. (2000), Grenthe and Lagermann (1991) and on a solubility study of Ryan and Rai (1987). Plugging the latter two values in eq. 12, results in * log10 Ks,0 of 9.9±0.8.

In Figure 92, the solubility data of the different studies mentioned above are visualized. It can be seen, that the measured solubilities are highly pH-dependent at pH < 6, but between pH 6 and 14, Th-concentrations are quite constant and pH independent. Furthermore, two groups of solubility data can be differentiated in the pH range < 6, indicating the presence of two different solubility limiting thorium dioxides (i.e. an amorphous and a crystalline), while at pH > 6 the Th-concentrations seem to be independent of the ThO2 crystallinity.

Figure 92: Solubility of ThO2 as function of pH measured by different authors. The solid line represents predictions made with the equilibria described above in equations 10-13 (Hummel et al., 2002).

The constants described above (eq. 10, 11, 13) were selected by Hummel et al. (2002), as they were able to reproduce the Th-solubilities under waste relevant conditions (i.e. solubility curve close to the measurements of more amorphous Th-oxides). It was explicitly mentioned that the solubility data of the more crystalline ThO2 could however not be described by the same dataset, and would require to decrease logKs,0 by around 8 orders of magnitude (!).

In more recent studies (Neck et al., 2002; Rothe et al., 2002, Bitea et al., 2003), it was shown by different techniques, such as laser induced breakdown detection (LIBD), X-ray absorption fine structure (EXAFS), ultrafiltration (UF) and coulometric titration, that besides crystallinity, particle size effects, kinetics, as well as colloid formation are strongly influencing the Th- solubility and may cause high discrepancies in solubility measurements.

Due to these findings, solubility data reviewed by NEA were generally evaluated separately, i.e. with respect to crystallinity (crystalline versus amorphous) and pH (acid versus neutral/alkaline).

SCK•CEN/12201513 Page 183 of 208 Compilation of Technical Notes on less studied elements

NEA data evaluation

Rard et al. (2008, NEA Vol. 11, p. 174) re-evaluated ThO2(cr), Th(OH)4(am) and ThO2(am,hyd) solubility data from different authors obtained in acidic, as well as neutral and alkaline solutions in order to re-calculate solubility constants from each of the studies using the hydrolysis constants and ion interaction coefficients (SIT) selected in their review.

Solubility of crystalline and microcrystalline ThO2(cr) in acidic solution

The dissolution of well crystallized ThO2(cr) was found to be characterized by very slow dissolution kinetics. This has been specifically observed in experiments performed from undersaturation direction with Th concentrations being significantly below the expected equilibrium value and almost independent of pH. If perfomed from oversaturation direction, 4+ equilibrium between crystalline ThO2(cr) and Th was however reached with Th- concentrations showing the expected pH dependence. This led to the conclusion that equilibrium in the solubility studies from undersaturation had not been reached. An interesting observation was made by Bundschuh et al. (2000) when performing coulometric titration of Th solutions at pH 1.5-2.5. The authors observed the formation of small ThO2 colloids with a mean diameter of 16-25 nm which agglomerated to microcrystalline ThO2 precipitates for which the solubility constant was determined to be around one log unit ° higher (logKsp = 3.0±0.4) than the thermochemical value of ThO2(cr), i.e. 1.77 ± 1.1.

Solubility of crystalline and microcrystalline ThO2(cr) in neutral and alkaline solution

The mean value of the experimental Th-concentrations determined by different authors corresponds to log[Th] = -9.3±1.4 and is close to the solubility of amorphous Th(OH)4(am,hyd) of -8.5±1.0. These concentrations were found to be orders of magnitude higher than the expected Th-concentrations of log[Th(OH)4(aq)] = -15.6±1.3. These increased solubilities were interpreted to be due to the presence of small amounts of more soluble amorphous material present in the crystalline solid or an amorphous hydrated layer on the ThO2(cr) surface.

Solubility of Th(OH)4(am)/ThO2(am,hyd) in acidic solutions (absence of carbonates)

The reported solubility constants derived from solubility data obtained in acidic solutions * range between log10 Ks,0 = 8.0 ± 0.7 and 9.8 ± 0.3 (p. 178) with the hydrolysis reactions given by:

4+ 4m-n + * mTh + nH2O ⇔ Thm(OH)n (aq)+ nH log10 βn,m (p. 140)

The following mean solubility constant was derived from the evaluated papers (p. 177):

+ 4+ ° Th(OH)4(am) or ThO2(am,hyd) + 4 H ⇔ Th + 2 H2O log10Ks,0 = 8.9 ± 1.1 which can also be written as:

SCK•CEN/12201513 Page 184 of 208 Compilation of Technical Notes on less studied elements

4+ - ° Th(OH)4(am) or ThO2(am,hyd) + 2 H2O ⇔ Th + 4 OH log10Ks,0 = -47.1 ± 1.1

Besides this, it was shown that depending on the (saturation) concentration of Th, either rather mononuclear species (i.e. low saturation concentration), or polynuclear hydroxide complexes (high saturation concentration) are formed.

Solubility in neutral and alkaline solutions (absence of carbonates)

The pH-independent solubility of Th(OH)4(am) or ThO2(am,hyd) in neutral to alkaline solutions is usually ascribed to the reaction:

Th(OH)4(am) or ThO2(am,hyd) ⇔ Th(OH)4(aq)

* Rard et al. (2008) determined a mean solubility of log10 Ks,4 of -8.5 ± 1.0 for the former reaction from solubility data measured in the pH range between 6-14 after ultrafiltration and/or ultracentrifugation (90 000 rpm/5×105 g) (see p. 179).

The selected value for the formation constant of the mononuclear Th(IV) hydrolysis species for the following reaction corresponds to log10*β4,1 = -17.4 ± 0.7 (see p. 180):

4+ + Th + 4 H2O ⇔ Th(OH)4(aq) + 4 H

*  From the data discussed above, Rard et al. (2008) recommend to use log10 Ks,4 of -8.5 ± 1.0 together with log10*β4,1 = -17.4 ± 0.7 as "operational values" for geochemical modelling.

Note: It is however explicitly mentioned that it cannot be ascertained that the Th-concentrations measured in solubility studies between pH 6-14 are only due to the mononuclear complex Th(OH)4(aq).

The thorium concentrations might (however) be higher in case of insufficient phase separation, i.e. due to the contribution of colloidal or large polynuclear/polymeric species to the total Th-concentrations. The respective equation corresponds to:

Th(OH)4(am) or ThO2(am,hyd) ⇔ 1/m Thm(OH)4m(aq)

* with log10 Ks,(4m,m) = -6.3 ± 0.8 (if Th(IV) polymers or colloids are not removed) (see p. 181).

Solubility in presence of carbonates

In presence of carbonates, the solubility of of ThO2(s) can be expressed as follows. At low carbonate concentrations, i.e. < 0.1 M, the predominant species corresponds to a mixed - hydroxide carbonate complex, namely Th(CO3)(OH)3 and the dissolution reaction can be expressed as:

2- + - * ° ThO2(s) + CO3 + H + H2O ⇔ Th(CO3)(OH)3 with log10 Ks,50 = 39.7 ±0.4

SCK•CEN/12201513 Page 185 of 208 Compilation of Technical Notes on less studied elements

At high carbonate concentrations, i.e. 0.15-2 M, the respective reaction can be written as:

2- + 6- * ThO2(s) + 5 CO3 + 4 H ⇔ Th(CO3)5 + 2 H2O log10 Ks,13 = 6.77 ± 0.21 with

4+ 2- - + * ° Th + CO3 + 3 H2O ⇔ Th(CO3)(OH)3 + 3 H log10 Ks,13 = -3.1 ± 1.0 and

4+ 2- 6- ° Th + 5 CO3 ⇔ Th(CO3)5 log10βCO3,5 = 29.8 ± 1.1

The solubility of ThO2×H2O(am) and formation of ternary Th(IV) hydroxide carbonate complexes was also investigated by Altmaier et al. (2005, 2006). An overview of the latter data and the selected complexation constants by NEA is given by Salah and Wang (2012; Annex IV), and it is referred to the discussion therein for further details concerning Th-complexation and solubility in presence of carbonates.

Summary

The dissolution behavior of ThO2(cr), Th(OH)4(am) and/or ThO2(am,hyd) in absence of carbonates has been studied since decades, but large discrepancies in the solubility values are observed in literature.

At basic to neutral pH, solubilitites of ThO2(cr) and Th(OH)4(am) vary between 1-10 nM, whereas under acid pH conditions, the solubilities differ by as much as 7 orders of magnitude. The similar solubilities at neutral pH, where Th(OH)4(aq) is the dominant aqueous species, were explained by the formation of an amorphous layer on the ThO2(cr) surface with logKsp° = -47.5±0.9 controlling the aqueous Th-concentrations. The solubility of amorphous Th-oxide/hydroxide was shown to be independent of pH (at pH > 6) and ionic strength with:

Th(OH)4(am) ⇔ Th(OH)4(aq)

However, the particle size was shown to have a "dramatic effect" on the Gibbs energy and solubility constant (independent of pH and ionic strength). In this context also ageing effects have to be mentioned, due to which NEA selected two solubility constants to be able to capture these effects, i.e. logKsp°[ThO2(am,hyd,aged)]=-47.5±0.9(-8.5±0.9) and logKsp°[ThO(am, hyd, fresh)] = -46.7±0.9 (-9.3±0.9).

At pH < 2.5, the ThO2(cr) dissolution results in the formation of small Th-colloids, with a mean diameter ranging between ~10-30 nm, as determined by LIBD. The solubility product of the freshly formed Th-colloids with logKsp°[colloids]= -52.9±0.5; ~20nm). Agglomeration and precipitation of these colloids was observed to result in the formation of microcrystalline thorium oxide, i.e. ThO2(mcr,hyd) with logKsp° determined to correspond to -53.2±0.4 (-3.0 ± ° 0.4), and being relatively close to solubility of crystalline Th (logKsp [ThO2,cr] = -54.2±1.3 (- 1.77±1.1) calculated from thermochemical data.

SCK•CEN/12201513 Page 186 of 208 Compilation of Technical Notes on less studied elements

As shown by Schindler, the solubility product of oxide and hydroxide particles depends on the particle size, because of the difference in Gibbs energy when either small solid particles or large solid particles with negligible molar surface are suspended in water, with:

logK°sp (particle size d) = -54.2±23/d.

The solubility products described above were found to be consistent with the determined particle size and the relation derived by Schindler.

At pH > 2.5-4.0 hydrolysis and polynucleation lead to the formation of amorphous thorium hydroxide colloids (eigencolloids), with the H+ and Th(IV) concentrations at the onset of colloid formation defining the solubility of Th(OH)4(am) excluding colloidal thorium species. These colloids formed by chemical polynucleation and condensation have been described to be hydrophilic in nature and long-time stable (i.e. they neither dissolve, nor agglomerate/precipitate). These colloids may be considered as small solid particles, but also as large aqueous species contributing to the total concentration in solution. As both tend to equilibrium with the aqueous species, consequently an equilibrium between the solid phase and eigencolloids should be approached as well.

a) ThO (cr) logK = -1.77 ± 1.1 0 a) b) c) d) e) 2 sp b) ThO2(mcr) logKsp = -3.0 ± 0.4 -2 c) Th(OH)4(coll) logKsp = -6.3 ± 0.8 d) Th(OH) (am,hyd,aged) logK = -8.5 ± 0.9 -4 4 sp e) Th(OH)4(am,hyd,fresh) logKsp = -9.3 ± 0.9 -6

-8

-10

og[Th] -12 l

-14

-16

-18 0 2 4 6 8 10 12 14 -log[H+] Figure 93: Th-solubility as function of pH calculated with values recommended by Rard et al. (2008)

4.3.2.3 Sorption and retardation Within the frame of the European Project FUNMIG (Fundamental Processes of Radionuclide Migration) and the Belgian radioactive waste management programme (ONDRAF/NIRAS), considerable effort was done to acquire high quality sorption data, as they represent key parameters used in safety assessment calculations, and to achieve a mechanistic understanding of the underlying processes. Sorption isotherms for Th were determined in suspensions of Real Boom Clay Water (RBCW), as well as Synthetic Boom Clay Water (SBCW; ~120 mg C/L) with a solid-to-liquid ratio of 6.7 g/L and at pH ~8.5. Based on gamma-activity measurements of the samples after phase separation (i.e. centrifugation at 21 000 x g/2h and ultrafiltration at 30 kDa), the sorption isotherms were determined and Kd values calculated for the centrifuged/non-filtered (NF) and ultrafiltrated (UF) samples.

SCK•CEN/12201513 Page 187 of 208 Compilation of Technical Notes on less studied elements

The results indicated that the sorption isotherms for both radionuclides and both media were linear within the investigated concentration range. In absence of DOM (i.e. SBCW experiments), the average logKd corresponds to ~4 after centrifugation (NF) and to ~5 after ultrafiltration (UF). This difference in logKd has been referred to the formation/presence of intrinsic/real Th-colloids (> 30 kDa) and/or pseudocolloids (i.e. sorption of thorium to inorganic Al-, Si-, Fe-, clay-colloids). The presence of DOM was shown to significantly influence the sorption behavior of Th, as log Kd values for the RBCW samples, i.e. ~3 (NF) – 4 (UF) were around one order of magnitude lower compared to those determined for the SBCW samples. This clearly revealed the formation of organic Th-colloids (pseudocolloids >30 kDa) and possibly Th-HA complexes (<30 kDa). The sorption and regression data for the non-filtered (NF) and ultrafiltered (UF) SBCW and RBCW samples are illustrated in Figure 94. The data points represent average values of duplicate experiments and analysis.

Th sorption on BC in RBCW (NF data) Th sorption on BC in SBCW (NF data) 1.0E-05 10000 1.0E-05 100000

y = 266.8x0.9794 R² = 0.9881

1.0E-06 10000 1.0E-06 1000

y = 2562x0.9558 R² = 0.9904 Kd [L/kg]Kd Kd [L/kg]Kd 1.0E-07 1000 1.0E-07 100 [Th]ads [mol/kg]

[Th] adsorbed [Th]ads [mol/kg] [Th] adsorbed y = 388.76x y = 7208.1x Kd R² = 0.9984 R² = 0.9918 Kd Linear ([Th] adsorbed) Linear ([Th] adsorbed)

Power ([Th] adsorbed) Power ([Th] adsorbed) 1.0E-08 100 1.0E-08 10 1E-12 1E-11 1E-10 1E-09 1E-08 1E-10 1E-09 1E-08 0.0000001 [Th]eq [mol/L] [Th]eq [mol/L] Th-sorption on BC in SBCW (UF data) Th-sorption on BC in RBCW (UF data)

1.0E-05 1000000 1.0E-05 100000

100000 8.0E-06 8.0E-06 10000

10000 6.0E-06 6.0E-06 1000 y = 99465x - 4E-07 y = 5336.5x - 3E-07 R² = 0.8081 1000 R² = 0.9825 4.0E-06 4.0E-06 100 [Th]ads [mol/kg] Kd [L/kg]Kd

100 [L/kg]Kd [Th]ads [mol/kg] [Th] adsorbed [Th] adsorbed 2.0E-06 Kd 2.0E-06 Kd 10 y = 2E+07x1.2258 10 Linear ([Th] adsorbed) y = 358.95x0.8894 Linear ([Th] adsorbed) R² = 0.9691 R² = 0.9521 Power ([Th] adsorbed) Power ([Th] adsorbed) 0.0E+00 1 0.0E+00 1 0.0E+00 1.0E-11 2.0E-11 3.0E-11 4.0E-11 5.0E-11 6.0E-11 7.0E-11 8.0E-11 0.0E+00 2.0E-10 4.0E-10 6.0E-10 8.0E-10 1.0E-09 1.2E-09 1.4E-09 1.6E-09

[Th]eq [mol/L] [Th]eq [mol/L]

Figure 94: Sorption isotherms and solid/liquid distribution coefficients (Kd-values) for Th-experiments performed in SBCW and RBCW after centrifugation (NF) and ultrafiltration (UF)

The data used to produce the graphs are summarized in Table 33.

SCK•CEN/12201513 Page 188 of 208 Compilation of Technical Notes on less studied elements

Table 33: Th-sorption data on Boom Clay in absence (SBCW) and presence of DOM (RBCW)

RBCW-NF Theq [mol/L] Th ads [mol/kg] Kd [L/kg] RBCW-UF Theq [mol/L] Th ads [mol/kg] Kd [L/kg] R0.0-NF 1.65E-08 6.31E-06 384 R0.0-UF 1.48E-09 7.88E-06 5880 R0.5-NF 7.95E-09 3.29E-06 418 R0.5-UF 8.78E-10 4.04E-06 4900 R1.0-NF 1.71E-09 6.40E-07 376 R1.0-UF 3.30E-10 7.86E-07 2389 R1.5-NF 8.67E-10 3.49E-07 403 R1.5-UF 1.67E-10 4.24E-07 2598 R2.0-NF 1.23E-10 7.58E-08 675 R2.0-UF 9.77E-12 8.79E-08 10197 R2.5-NF 1.11E-10 3.44E-08 378 R2.5-UF 7.97E-12 4.60E-08 6720

SBCW-NF Theq [mol/L] Th ads [mol/kg] Kd [L/kg] SBCW-UF Theq [mol/L] Th ads [mol/kg] Kd [L/kg] S0.0-NF 1.15E-09 8.59E-06 7535 S0.0-UF 6.67E-11 8.70E-06 136304 S0.5-NF 6.25E-10 3.99E-06 6390 S0.5-UF 6.70E-11 4.05E-06 60568 S1.0-NF 1.54E-10 7.96E-07 5211 S1.0-UF 1.78E-11 8.10E-07 46061 S1.5-NF 4.83E-11 4.29E-07 8973 S1.5-UF 8.67E-12 4.33E-07 50033 S2.0-NF 1.34E-11 8.66E-08 6490 S2.0-UF 1.64E-12 8.79E-08 55030 S2.5-NF 4.45E-12 4.18E-08 9554 S2.5-UF 1.65E-12 4.21E-08 27196

Literature data

Two key publications can be considered as main references when interpreting and modeling the sorption of actinides and heavy metals (i.e. Mn(II), Co(II), Ni(II), Zn(II), Cd(II), Eu(III), Am(III), Sn(IV), Np(V) and U(VI)) on montmorillonite and illite, i.e. Bradbury and Baeyens (2005) and Bradbury and Baeyens (2009). Therein, surface binding constants with the respective hydrolysis constants for most of the waste relevant actinides can be found. The latter were derived through modeling in-house and literature sorption data by using their well- established 2-site protolysis non-electrostatic surface complexation and cation exchange model (2SPNE SC/CE) and so-called linear free energy relationships (LFER), enabling also the estimation of surface complexation constants for which sorption data are either poorly known and/or completely lacking.

The following equations were found to describe well the correlations between the logarithms S W1 of strong (log Kx-1) and weak (log Kx-1) site binding constants on montmorillonite/illite and OH the logarithms of the aqueous hydrolysis constants (log Kx):

Montmorillonite

S OH strong (≡SOH) sites : log Kx-1 = 8.1 (±0.3) + 0.9 (±0.02) log Kx W1 OH weak (≡SOH) sites : log Kx-1 = 6.2 (±0.8) + 0.98 (±0.09) log Kx

Illite

S OH strong (≡SOH) sites : log Kx-1 = 7.9 (±0.4) + 0.83 (±0.02) log Kx

In the following tables, the main sorption parameters used in the 2 SPNE SC/CE, as well as the surface binding constants for Th on Na-montmorillonite (Table 34) and Na-illite (Table 35) are summarized. Table 36 comprises a compilation of the Th-hydrolysis constants from Bradbury and Baeyens (2005, 2009), the NEA TDB (Rand et al., 2008) and the NAGRA/PSI TDB (Hummel et al., 2002). Concerning the aqueous hydrolysis constants and SC constants of

SCK•CEN/12201513 Page 189 of 208 Compilation of Technical Notes on less studied elements

other actinides and heavy metals, it is referred to the papers of Bradbury and Baeyens mentioned above.

Table 34: Th(IV) sorption data on Na-montmorillonite (Bradbury and Baeyens, 2005)

Site types Site capacities ≡SSOH 2.0 × 10-3 mol/kg ≡SW1OH 4.0 × 10-2 mol/kg ≡SW2OH 4.0 × 10-2 mol/kg Total site density 8.2 × 10-2 mol/kg CEC 8.7 × 10-1 eq/kg SA (BET) 35 m2/g Protolysis constants strong sites logKprotolysis S + S + ≡S OH + H ⇔ ≡S OH2 4.5 ≡SSOH ⇔ ≡SSO- + H+ -7.9 1 Protolysis constants weak W sites logKprotolysis W1 + W1 + ≡S OH + H ⇔ ≡S OH2 4.5 ≡SW1OH ⇔ ≡SW1O- + H+ -7.9 2 Protolysis constants weak W sites logKprotolysis W2 + W2 + ≡S OH + H ⇔ ≡S OH2 6.0 ≡SW2OH ⇔ ≡SW2O- + H+ -10.5 Surface complexation reactions on strong sites logKSC of Na-montmorillonite ≡SSOH + Th4+ ⇔ ≡SSOTh3+ + H+ 7.2 S 4+ S 2+ + ≡S OH + Th + H2O ⇔ ≡S OThOH + 2 H 2.7 S 4+ S + + ≡S OH + Th + 2 H2O ⇔ ≡S OTh(OH)2 + 3 H -2.6 S 4+ S + ≡S OH + Th + 3 H2O ⇔ ≡S OTh(OH)3 + 4 H -9.1 S 4+ S + ≡S OH + Th + 4 H2O ⇔ ≡S OTh(OH)4 + 5 H -16.9 Surface complexation reactions on weak1 sites logKSC of Na-montmorillonie ≡SW1OH + Th4+ ⇔ ≡SW1OTh3+ + H+ no data W1 4+ W1 2+ + ≡S OH + Th + H2O ⇔ ≡S OThOH + 2 H no data Surface complexation reactions on weak2 sites logKSC of Na-montmorillonite ≡SW2OH + Ca2+ ⇔ ≡SW2OCa+ + H+ no data W2 2+ W2 + ≡S OH + Mg + H2O ⇔ ≡S OMg + H no data Cation exchange reactions on planar sites logKc 4 Na-momo + Th4+ ⇔ Th-momo + 4 Na+ no data 3 Na-momo + Al3+ ⇔ Al-momo + 3 Na+ no data 2 Na-momo + Ca2+ ⇔ Ca-momo + 2 Na+ no data 2 Na-momo + Mg2+ ⇔ Mg-momo + 2 Na+ no data

Table 35: Th(IV) sorption data on Na-illite (Bradbury and Baeyens, 2009)

Site types Site capacities ≡SSOH 2.0 × 10-3 mol/kg ≡SW1OH 4.5 × 10-2 mol/kg ≡SW2OH 4.5 × 10-2 mol/kg Total site density 9.2 × 10-2 mol/kg CEC 2.25 × 10-1 eq/kg SA (ethylene-glycol-monoethyl ether; EGME) 129 m2/g Protolysis constants strong sites logKprotolysis S + S + ≡S OH + H ⇔ ≡S OH2 5.5 ≡SSOH ⇔ ≡SSO- + H+ -6.2 1 Protolysis constants weak W sites logKprotolysis W1 + W1 + ≡S OH + H ⇔ ≡S OH2 5.5

SCK•CEN/12201513 Page 190 of 208 Compilation of Technical Notes on less studied elements

≡SW1OH ⇔ ≡SW1O- + H+ -6.2 2 Protolysis constants weak W sites logKprotolysis W2 + W2 + ≡S OH + H ⇔ ≡S OH2 9.0 ≡SW2OH ⇔ ≡SW2O- + H+ -10.5 Surface complexation reactions on strong sites of Na-illite logKSC ≡SSOH + Th4+ ⇔ ≡SSOTh3+ + H+ 7.4 S 4+ S 2+ + ≡S OH + Th + H2O ⇔ ≡S OThOH + 2 H 2.3 S 4+ S + + ≡S OH + Th + 2 H2O ⇔ ≡S OTh(OH)2 + 3 H -2.4 S 4+ S + ≡S OH + Th + 3 H2O ⇔ ≡S OTh(OH)3 + 4 H -8.8 S 4+ S + ≡S OH + Th + 4 H2O ⇔ ≡S OTh(OH)4 + 5 H -15.3 1 Surface complexation reactions on weak W sites of Na-illite logKSC ≡SW1OH + Th4+ ⇔ ≡SW1OTh3+ + H+ no data W1 4+ W1 2+ + ≡S OH + Th + H2O ⇔ ≡S OThOH + 2 H no data 2 Surface complexation reactions on weak W sites of Na-illite logKSC ≡SW2OH + Ca2+ ⇔ ≡SW2OCa+ + H+ no data W2 2+ W2 + ≡S OH + Mg + H2O ⇔ ≡S OMg + H no data Cation exchange reactions on planar sites logKc 4 Na-illite + Th4+ ⇔ Th-illite + 4 Na+ no data 3 Na-illite + Al3+ ⇔ Al-illite + 3 Na+ no data 2 Na-illite + Ca2+ ⇔ Ca-illite + 2 Na+ no data 2 Na-illite + Mg2+ ⇔ Mg-illite + 2 Na+ no data

Table 36: Hydrolysis constants of different databases and publications

log10β° log10β°

NAPSI TDB NAPSI TDB log10β° log10β° Hydrolysis Reaction Version 05/92 Version 01/01 (Bradbury and NEA (Rand species Pearson et al., (Hummel et al., Baeyens, 2005) et al., 2008) 1992) 2002)

3+ 4+ 3+ + ThOH Th + H2O ⇔ ThOH + H -3.2 -2.4 ± 0.5 -2.2 -2.5 ± 0.5 4+ ⇔ 2+ 2+ Th + 2 H2O Th(OH)2 + not included Th(OH)2 -6.95 -6.0 -6.2 ± 0.5 2 H+ anymore 4+ ⇔ + + Th + 3 H2O Th(OH)3 + not included Th(OH)3 -11.7 -11.0 not included 3 H+ anymore 4+ Th + 4 H2O ⇔ Th(OH)4(aq) Th(OH)4(aq) -15.9 -18.4 ± 0.6 -17.5 -17.4 ± 0.5 + 4 H+ 4+ – ⇔ 2+ 2+ *1 Th + HCO3 ThCO3 + not included ThCO3 0.671 n.a. not included H+ anymore 4+ – ⇔ 6– 6- *1 Th + 5 HCO3 Th(CO3)5 Th(CO3)5 not included -21.8 n.a. -20.6 + 5 H+ 4+ – ⇔ – *1 Th + HCO3 + 3 H2O Th(CO3)(OH)3 – + not included -13.4 n.a. not included Th(CO3)(OH)3 + 4 H 1 2- * Recalculated value using bicarbonate species instead of CO3

Sorption modeling

The experimentally obtained sorption data of Th on BC, as well on purified Na- montmorillonite and Na-illite were modelled by Salah et al. (2007, 2013) using the 2 SPNE SC/CE model and sorption data of Bradbury and Baeyens (2005, 2009), as well as other approaches.

4.3.2.4 Diffusion and transport No migration experiments with thorium were performed at SCKCEN. SCK•CEN/12201513 Page 191 of 208 Compilation of Technical Notes on less studied elements

4.3.2.5 Justification As described in the previous sections, the solubility of amorphous and crystalline ThO2 was extensively studied by many authors and SCK•CEN. While the former studied mainly the ThO2 solubility in absence of carbonate and organic ligands, the experiments at SCK•CEN were performed under BC relevant conditions. Results revealed a strong solubility increasing effect in presence of dissolved organic matter with “apparent solubilities” being few orders of magnitude higher than the calculated thermodynamic solubilities. The strong affinity of Th for DOM was also reflected by the sorption experiments and determined Kd-values in absence and presence of DOM. Also here, the sorption coefficients measured in presence of DOM were lower (~1 order of magnitude) than the Kd’s measured in absence of organics. With respect to transport unfortunately no in-house data are available, but we consider U(IV) to be a good analogue for Th(IV). Based on the experimental results and the U(IV)/Th(IV) analogy, transport of U through undisturbed BC is also considered to be organic matter (DOM) mediated.

4.3.2.6 References Altmaier M., Neck, V. Müller R. and Fanghänel Th. 2005. Solubility of ThO2 · xH2O(am) in carbonate solution and the formation of ternary Th(IV) hydroxide-carbonate complexes. Radiochim. Acta., Vol. 93, p. 83-92.

Altmaier M., Neck, V. Denecke M.A., Yin R. and Fanghänel Th. 2006. Solubility of ThO2 ·xH2O(am) and the formation of ternary Th(IV) hydroxide-carbonate complexes in NaHCO3-Na2CO3 solutions containing 0-4 M NaCl. Radiochimica Acta, Vol. 94, p. 495-500.

Bitea C., Müller R., Neck V., Walther C., and Kim J. I. 2003. Study of the generation and stability of Th(IV) colloids by LIBD combined with ultrafiltration. Colloids and Surfaces A:Physicochem. Eng. Aspects, 217, 63-70.

Bradbury M. H., and Baeyens B. 2005. Modelling the sorption of Mn(II), Co(II), Ni(II), Zn(II), Cd(II), Eu(III), Am(III), Sn(IV), Th(IV), Np(V) and U(VI) on montmorillonite: Linear free energy relationships and estimates of surface binding constants for some selected heavy metals and actinides. Geochimica et Cosmochimica Acta, 69(4), 875-892.

Bradbury M. H. and Baeyens B., 2009. Sorption modelling on illite. Part II: Actinide sorption and linear free energy relationships. Geochimica et Cosmochimica Acta, 73, 1004-1013.

Bruggeman C., Maes N., Aertsens M., Govaerts J., Martens E., Jacops, E., Van Gompel, M., and Van Ravesteyn, L. 2013. Technetium retention and migration behaviour in Boom Clay. Topical Report. Final Draft after Review. SCK•CEN-ER-101.

Delécaut G., 2004. The geochemical behaviour of uranium in the Boom Clay. PhD thesis, Université Catholique de Louvain – SCK•CEN, Louvain-La-Neuve, Belgium.

Ekberg C., Albinsson Y., Comarmond M. J., Brown P. L. 2000. Studies on the Complexation Behavior of Thorium(IV). 1. Hydrolysis Equilibria. J. Solution Chemistry, 29(1), p.63-86.

Felmy A. R., Rai D., Mason M. J. 1991. The Solubility of Thorium(IV) Oxide in Chloride Media: Development of an Aqueous Ion-Interaction Model. Radiochimica Acta, 55, 177-185.

Geraedts K., and Maes A. 2008. Determination of the conditional interaction constant between colloidal technetium(IV) and Gorleben humic substances. Applied Geochemistry, 23, 1127-1139.

SCK•CEN/12201513 Page 192 of 208 Compilation of Technical Notes on less studied elements

Grenthe I., Lagermann B. 1991. Studies on Metal Carbonate Equilibria. 23. Complex Formation in the Th(IV)-H2O-CO2(g) System. Acta Chemica Scandinavia, 45, 231-238.

Hummel W., Berner U., Curti E., Pearson F. J., and Thoenen T. 2002. NAGRA/PSI Chemical Thermodynamic DataBase 01/01. NAGRA Technical Report 02-16. Universal Publishers, Parkland, Florida.

IAEA (2005): Thorium fuel cycle — Potential benefits and challenges. AEA-TECDOC-1450. Vienna, Austria.

Langmuir D., Herman J. S. 1980. The Mobility of Thorium in Natural Waters at Low Temperatures. Geochimica et Cosmochimica Acta, 44, 1753-1766.

Liu D. J., Bruggeman C., and Maes N. (unpublished): Influence of organic matter on the solubility of Th in Boom Clay geochemical conditions.

Maes A., Bruggeman C., Breynaert E., Geraedts K., and Vancluysen J., 2003a. The interaction of technetium-99 with Boom Clay - Fifth progress report (09/2002 - 03/2003), Contract CCHO 2001- 845/00/00 - KULeuven report, Leuven, Belgium.

Maes A., Bruggeman C., Geraedts K., and Vancluysen J., 2003b. Quantification of the interaction of Tc with dissolved boom clay humic substances. Environmental Science & Technology 37, 747-753.

Maes, N., Bruggeman, C., Govaerts, J., Martens, E., Salah, S., Van Gompel, M., 2011. A consistent phenomenological model for natural organic matter linked migration of Tc(IV), Cm(III), Np(IV), Pu(III/IV) and Pa(V) in the Boom Clay. Physics and Chemistry of the Earth, 36, 1590-1599.

Martens E., Maes N., Weetjens E., van Gompel M., and Van Ravestyn, L., 2010. Modelling of a large- scale in-situ migration experiment with 14C-labelled natural organic matter in Boom Clay, Radiochimica Acta, 98, 695-701.

Moon H. C. 1989. Equilibrium Ultrafiltration of Hydrolysed Thorium(IV) Solutions. Bull. Korean Chem. Soc., 10, 270-272.

Neck V., and Kim J. I. 1999. Solubility and Hydrolysis of Tetravalent Actinides. Forschungszentrum Karlsruhe, Wissenschaftliche Berichte FZKA 6350. Additional information on Th are provided in the very recent paper: Neck V., and Kim J. I. 2001. Solubility and Hydrolysis of Tetravalent Actinides. Radiochimica Acta, 89,1-16.

Neck V., Müller R., Bouby M., Altmaier M. Rothe, J., Denecke M.A., and Kim J. I. 2002. Solubility of amorphous Th(IV) hydroxide – application of LIBD to determine the solubility product and EXAFS for aqueous speciation. Radiochim. Acta, 90, 485-494.

Östhols E., Bruno J., Grenthe I. 1994. On the influence of carbonate mineral dissolution: III. The solubility of microcrystalline ThO2 in CO2-H2O media. Geochimica et Cosmochimica Acta, 58, 613-623.

Pearson F. J., Berner U., Hummel W. 1992. Nagra Thermochemical Data Base II. Supplemental Data 05/92. Nagra Technical Report NTB 91-18, Nagra, Wettingen, Switzerland, 284 p.

Rand M., Fuger J., Grenthe I., Neck V., and Rai D. 2008. Chemical Thermodynamics of Thorium, OECD NEA, Issy-les-Moulineaux, France.

Rothe J., Denecke M. A., Neck V., Müller R., and Kim J. I. 2002. XAFS investigation of the structure of aqueous Th(IV) species, colloids and solid Th(IV) oxide/hydroxide. Inorg. Chem., 41, 249.

Ryan J. L., and Rai D. 1987. Thorium(IV) Hydrous Oxide Solubility. Inorganic Chemistry, 26, 4140-4142.

SCK•CEN/12201513 Page 193 of 208 Compilation of Technical Notes on less studied elements

Salah S., Maes N., and Wang L. 2007. Sorption behaviour of Am(III) and Th(IV) on Boom Clay. Poster presentation at the 7th International Conference on the Chemistry and Migration Behaviour of Actinides and Fission Products in the Geosphere. München (Germany) 26/08-31/08/2007.

Salah S., Wang L., Maes N., Van Gompel M. and Brassinnes S. 2009. Influence of natural organic matter on the Th sorption onto purified Na-montmorillonite - Experimental results and modelling -. Poster presentation at ANDRA conference (4th International Meeting) on "Clays in Natural and Engineered Barriers for Radioactive Waste Confinement. Nantes (France) 29/03-01/04/2010.

Salah S., Liu, D. and Wang, L. 2013. Influence of particle size, carbonates and organic matter on the th solubility of ThO2(cr). Poster presentation at the 13 International Conference on the Chemistry and Migration Behaviour of Actinides and Fission Products in the Geosphere. Brighton (U.K.) 08/09- 13/09/2013.

Salah, S. and Wang, L. 2012. Speciation and solubility calculations for waste relevant radionuclides in Boom Clay. External Report, SCKCEN-ER- 198. First Full Draft.

Serne R. J., Rai D., Martin P. F., Felmy A. R., Rao L., Ueta S. 1996. Leachability of Nd, U, Th, and Sr from Cements in a CO2 free Atmosphere. Mat. Res. Soc. Symp. Proc., 412, 459-467.

Wierczinski B., Helfer S., Ochs M., Skarnemark G. 1998. Solubility measurements and sorption studies of thorium in cement pore water. J. Alloys and Comp.

SCK•CEN/12201513 Page 194 of 208 Compilation of Technical Notes on less studied elements

4.3.3 Technical note for Protactinium (Pa)

4.3.3.1 General Protactinium (Pa) is a rare actinide with an atomic number of 91 (Winter, 2014). Experimental data on protactinium complexes and compounds are very scarce (Duro et al., 2006a). Aqueous protactinium can exist in two oxidation states, Pa(IV) and Pa(V) (Duro et al., 2006a). It is generally assumed that the Pa(V) is stabilised under repository-type environments (Berner, 2002; Hakanen et al., 2014). However, very little is known about the complexation chemistry of protactinium in groundwaters (Hakanen et al., 2014).

4.3.3.2 Speciation and solubility Thermodynamic data for Pa are still scarce, but were comprised in ThermoChimie v.5 from which they were copied to MOLDATA. More recent data have become available in the meantime (Trubert et al., 2002; Trubert et al., 2003) and were taken into account in the calculations presented here. Based on these data, the prevalent aqueous species under BC conditions is calculated to be Pa(OH)5(aq) (Figure 95).

PaO(OH)++ .5 + PaO(OH)2 Eh (volts) Eh 0 Pa(OH)5(aq) µ

–.5

25°C 0 2 4 6 8 10 12 14 pH Figure 95: Eh-pH diagram of protactinium (Pa-C-S-O-H) for the BC reference porewater system. Assumed activity of [Pa] = 10-8. Database: MOLDATA (2010_MOLDATA_nov_b O2.dat). Code: The Geochemist's Workbench - 8.12.

In the SAFIR 2 assessment, a best estimate solubility value for protactinium under Boom Clay conditions was given as 1×10-5 mol/dm3 (minimum 1×10-11 mol/dm3, maximum 1×10-5 mol/dm3, triangular log distribution; ONDRAF/NIRAS, 2001).

Three different solid phases have been proposed as likely to control protactinium concentrations under repository conditions: PaO2, Pa2O5 and Pa2O5·H2O (Duro et al., 2006a; Wersin et al., 2014a), with reported solubility limits generally being in the range ~10-6 3 -11 3 mol/dm (Pa2O5·H2O solubility) to ~10 mol/dm (PaO2 solubility) (Berner, 2002; Duro et al., 2006b; Wersin et al., 2014a).

In MOLDATA the following Pa-solids are comprised, i.e. Pa(cr), PaO2(s) and Pa2O5(s). The former two solids were calculated to be very soluble under BC conditions and the solubility -10 calculated for Pa2O5(s) corresponds to 9.8×10 (Table 37).

SCK•CEN/12201513 Page 195 of 208 Compilation of Technical Notes on less studied elements

Table 37: Solubility of Pa in the BC reference porewater system at 25°C, pH 8.355 and Eh -281 mV. Database: MOLDATA. Code: The Geochemist's Workbench - 8.08.

Solubility controlling phases Solubility Pa [mol/l]

-10 Pa2O5 (s) 9.8 × 10

PaO2 (s) very soluble Pa(cr) very soluble

Source data: ANDRA TDB

The reaction constant for the Pa-solids comprised in MOLDATA is the following:

+ 4+ Pa2O5(s) + 8 H ↔ 2 Pa + 4 H2O log K = -55.43 + 4+ PaO2(s) + 4 H ↔ Pa + 2 H2O log K = 166.78 + 4+ Pa(cr) + 4 H + O2(aq) ↔ Pa + 2 H2O log K = 184.73

For the SR-Can assessment, PaO2(OH)(aq) is identified as a key species under the conditions of interest (Duro et al., 2006b), but it is noted that relevant thermodynamic data for protactinium compounds are scarce.

In the review of protactinium chemistry undertaken for the TURVA-2012 assessment, Hakanen et al. (2014) note that little is known about its complexation behaviour under groundwater conditions. Hakanen et al. (2014) cite Tarapcik et al. (2005) who estimated the E0 to be -0.1 V for the reduction of Pa(V) to Pa(IV). Hakanen et al. (2014) state that speciation calculations undertaken using the data in Baes and Mesmer (1986) result in Pa(V) being dominated by PaO2(OH)(aq) from pH 6 to 10. Hakanen et al. (2014) note that under reducing conditions, the hydrolysis data selected by Duro et al. (2006a) gives the dominating species as PaO(OH)+ (PaO2+) for the pH range 6 to 10. Hakanen et al. (2014) cite Pa(V) hydrolysis data from Trubert et al. (2002), Le Naour et al. (2003) and Fourest et al. (2004). Fourest et al. (2004) measured the diffusion of Pa(V) in capillary experiments and suggested that in basic aerobic conditions, in addition to cationic and neutral hydroxo complexes, an anionic Pa(V) hydroxo complex is formed, probably PaO2(OH)2-. Hakanen et al. (2014) present speciation calculations for Pa(V) in the ‘OLSR’ saline reference water that were undertaken using the ThermoChimie database (version 7b), which incorporates data from Fourest et al. (2004), and found that key species under near-neutral pH are PaO2+ and PaO2(OH)(aq), with PaO2(OH)2- becoming more prevalent under relatively higher pH conditions (> pH 8).

According to the previous paragraph, more recent data (i.e. Fourest et al. 2004) than the ones included in the MOLDATA _novb.dat and MOLDATA_R2.dat TDB’s (i.e. Trubert et al., 2002) have become available. The more recent Pa-data were incorporated in the latest version of ThermoChimie, i.e. v.7b, while in ThermoChimie v.5 also the data from Trubert et al. (2002) are comprised. Based on the last sentence of the previous paragraph, it can be anticipated that the speciation under BC conditions might be different, if the updated Pa- data would be used for the calculations.

SCK•CEN/12201513 Page 196 of 208 Compilation of Technical Notes on less studied elements

4.3.3.3 Sorption and retardation In their review of sorption data, Linklater et al. (2003) found that surface complexation is the main retardation process associated with protactinium, and that it is often observed to be pH- dependant, showing sharp increases with increasing pH. Linklater et al. (2003) consider that un-complexed protactinium ions are unlikely, even under acidic pH conditions, and that ion exchange is not likely to be an important sorption mechanism. Linklater et al. (2003) note that there is the potential for strong sorption of protactinium to colloids. In addition, Pa-NOM association has been put forward in the newly developed phenomenological model by Bruggeman and Maes (2016).

In the TURVA-2012 assessment, Posiva base their in-situ bentonite buffer Kd values for protactinium on the sorption edge measurements made on Na-montmorillonite (SWy-1) that are described by Bradbury and Baeyens (2006). These measurements show a constant sorption value of 89 m3/kg over a pH range of 4 to 10.5 (Wersin et al., 2014a).

Hakanen et al. (2014) describe experiments of protactinium sorption to clays for different water compositions (‘ALLMR’ and ‘OLSR’) and report that for many samples, only the ‘detection limit’ value of Rd could be determined. The ‘best estimate’ Kd values for protactinium sorption to kaolinite and illite are given as 150 m3/kg and 60 m3/kg, respectively for a number of reference water compositions (Hakanen et al., 2014).

Bradbury and Baeyens (2003a, 2003b) review the data on protactinium sorption and consider that neither Th(IV) or Np(V) would be particularly good analogues for protactinium. They also note the experimental difficulties associated with working with protactinium, but state that the available sorption data suggests strong sorption on almost all rock types under neutral to slightly-alkaline conditions. Bradbury and Baeyens (2003b) cite Berry et al. (1988) who measured the sorption of Pa(V) under reducing conditions on six rock types in the pH range 6 to 9.5 and measured values in the range 1 to greater than 1000 m3/kg. The values for London clay, selected as being the most similar of the six materials to Opalinus Clay, are between -3 and >1000 m3/kg (pH = 8.8 to 9.1, initial protactinium concentration 5×10-11 mol/dm3). Bradbury and Baeyens (2003b) consider that the available data suggest a value >1 m3/kg for the Opalinus Clay and given the reported values for London Clay, the actual value is probably ≥ 3 m3/kg. Bradbury and Baeyens (2003b) therefore recommend a value of 5m3/kg for Opalinus Clay (with an uncertainty factor of 10). For ‘generic’ Swiss argillaceous 3 rock Bradbury et al. (2010) recommend Rd values of 13 m /kg for a range of clay porewater conditions.

Geibert and Usbeck (2004) describe experiments of protactinium sorption to different particles (smectite, biogenic opal from a cleaned diatom culture, manganese dioxide precipitate, and calcium carbonate) in the presence of seawater. The Kd values measured after establishment of an equilibrium ranged from 0.03 to 166×106 mL/g depending on particle type and on the type of seawater used.

4.3.3.4 Diffusion and transport -11 2 For the Opalinus Clay, protactinium was assigned ‘non anionic’ Deff values of: 1×10 m /s (reference value perpendicular to bedding); 1×10-10 m2/s (‘upper pessimistic’ value

SCK•CEN/12201513 Page 197 of 208 Compilation of Technical Notes on less studied elements

perpendicular to bedding); and 5×10-11 m2/s (reference value parallel to bedding). A value of 0.12 was assigned to η (Bradbury and Baeyens, 2003b).

In the TURVA-2012 assessment, protactinium (along with other ‘cationic/hydrolysable -11 2 radionuclides’ radionuclides) was assumed to have Deff and η values of 9×10 m /s and 0.38 for bentonite backfill, assuming behaviour similar to that of HTO (Wersin et al., 2014b).

Two percolation experiments (type C4) and one sequential migration experiment have been performed with 231Pa in confined Boom Clay cores. Details of these experiments can be found below:

1. Percolation C4 experiment Pa231m3c3(NRM021A) • Initiated 30/04/1998 (loading and start percolation); • Clay core code ANDRA7/8 core 19 code ANDRA7/8 9.65-10.0(date core sampling 21 aug 1997) • Initial activity 94 kBq injected, in chemical form Pa in 8M HNO3 + 0.5M HF • Pa concentration in original source solution 4.711×10-4 M (calculated) • Total clay core length: 72 mm (40 mm "inlet" + 32 mm "outlet"), diameter 38 mm • Percolated solutions followed until now

2. Percolation C4 experiment Pa231m3c4a (NRM021B) • Initiated 30/04/1998 (loading and start percolation); • Clay core code ANDRA7/8 core 19 code ANDRA7/8 9.65-10.0(date core sampling 21 aug 1997) • Initial activity 94 kBq injected, in chemical form Pa in 8M HNO3 + 0.5M HF • Pa concentration in original source solution 4.711×10-4 M (calculated) • Total clay core length: 72 mm (40 mm "inlet" + 32 mm "outlet"), diameter 38 mm • Used as input for sequential migration experiment 19/01/2006

3. Sequential migration experiment Pa231m3c4b

• Initiated 19/01/2006 • Mounted after clay core Pa231m3c4a • Total clay core length: 104 mm (72 mm first core + 32 mm second core), diameter 38 mm

The hydraulic conductivity, K (m/s), for the three experiments is given in Figure 96. The value for K is in line with those normally observed in Putte Member of Boom Clay (Yu et al., 2013).

SCK•CEN/12201513 Page 198 of 208 Compilation of Technical Notes on less studied elements

Figure 96: Calculated hydraulic conductivity, K (in m/s), as function of time in 2 percolation experiments (Pa231m3c3-top and Pa231m3c4a-middle) and sequential migration experiment (Pa231m3c4b-bottom)

SCK•CEN/12201513 Page 199 of 208 Compilation of Technical Notes on less studied elements

Both percolation experiments show a very rapid breakthrough of a very small fraction of Pa (observed in the first year after start of the experiment), followed by a slow release from the clay core (Figure 97) tending towards a more or less constant value (although maybe still slowly increasing with time). The very rapid breakthrough is also in contrast with high expected sorption (retardation) for Pa.

In the sequential migration experiment, the Pa elution curve out of the second clay core exhibits similar features as the elution out of the first core, but without the "peak" feature observed in the percolation experiments. Rather, Pa is slowly increasing in the percolate solutions and a constant outlet value tends to be reached, but the observed concentrations are still below the ones measured in the percolation experiments.

The results of these experiments can be interpreted as resulting from DOM linked/facilitated transport of Pa). The Pa species which are eluted from the (first) clay core in the percolation experiments are Pa-DOM species, while the majority of the introduced Pa is retained at the source position (either as precipitate or adsorbed onto the solid phases). The concentration decrease after elution through the second clay core implies that the Pa-DOM species eluting from the first core are slowly dissociating. The dissociated species are subsequently retained in the second clay core.

In Maes et al. (2011), a transport model was developed based on the conceptual model described in Figure 98. Radionuclides in solution will either be present as a mobile RN-DOM complex or "free inorganic" radionuclide species in solution ([RNinorg]liquid). The transfer between [RNinorg]liquid and the RN-DOM complex is described by a complexation constant and dissociation kinetics. Both species can interact with the solid phase. It is assumed that this interaction in case of [RNinorg]liquid is mainly due to sorption processes and can be described by a retardation factor (RRN) which can be linked to batch sorption data. In case of RN-DOM, the retardation factor (RRN-DOM) is considered as lumped factor accounting for both sorption and colloid filtration processes. Overall, dissolved OM is only poorly retarded within Boom Clay and RRN-DOM is therefore expected to be only of secondary importance to describe RN- coupled transport. Within this transport model, the amount of parameters remains limited and most of them can be obtained from independent measurements (batch complexation/solubility experiments, batch sorption experiments, DOM transport experiments).

The transport of the RN-DOM mobile complex is described by: ∂()c ηρ+ m −∇⋅η ∇ + =−ληρ + + η ()bK d,, m Dpore m cV m Darcy c m ()b K d, m c m Q sol− OM ∂t

SCK•CEN/12201513 Page 200 of 208 Compilation of Technical Notes on less studied elements

Figure 97: Pa concentration in outlet (231Pa, as Bq/L) as function of time in 2 percolation experiments (Pa231m3c3-top and Pa231m3c4a-middle) and sequential migration experiment (Pa231m3c4b-bottom)

SCK•CEN/12201513 Page 201 of 208 Compilation of Technical Notes on less studied elements

mobile RN-DOM complex

kdecomp [ RNinorg ]liquid RN-DOM

kcomp

Linear RRN R Sorption RN-DOM

RN RN-DOM Boom Clay solid phase

Figure 98: Conceptual model used for the interpretation of organic matter linked radionuclide migration in Boom clay (Maes et al., 2011)

The transport of the RN-species in solution is described by: ∂()c ηρ+ s −∇⋅η ∇ + =−ληρ + − η ()bK d,, s Dpore s cV s Darcy c s ()b K d, s c s Q sol− OM ∂t

The mass transfer of the RN between the DOM complexed form and the "free" RN in solution is given by: Q= k cc − k c sol− OM comp s OM decomp m

To constrain the degrees of freedom present in the model, the RN-DOM interaction constant was fixed during fitting to a value of logKRN-DOM = 4.7. The model provides excellent fits (Figure 99) to the experimental data and the fitted parameters are presented in Table 38 (Maes et al., 2011).

SCK•CEN/12201513 Page 202 of 208 Compilation of Technical Notes on less studied elements

Table 38: Summary of the fitted parameters kdecomp, RRN-DOM and RRN (Maes et al., 2011)

-1 logKRN-DOM kdecomp [s ] RRN-DOM [-] logRRN [-] Pa(V) 5.3 0.3±0.2×10-6 18±4 3.81±1.22

Figure 99: Results of the fitting of the Pa elution curve in a sequential migration experiment using the proposed conceptual model (Maes et al., 2011)

4.3.3.5 Justification Based on the speciation calculations with MOLDATA, protactinium occurs in the pentavalent oxidation state in BC porewater and represents a strongly hydrolyzing species (Pa(OH)5,aq). It should be mentioned however, that the Pa-speciation is quite uncertain due to quite big differences in the thermodynamic data. At the time the grouping approach and conceptual model were developed, Pa-data were still scarce and we relied on information of Bruno et al. (1997), according to which PaO2(OH)(aq) was considered to be the dominant aqueous species in the pH range > 8. As the latter showed resemblance to TcO(OH)2(aq), the transport of Pa was assumed to be goverened by the same processes as for Tc(IV). Besides this, sorption is also reported to be very strong for Pa on all rock types under near neutral and neutral conditions with surface complexation being the dominant mechanism even at lower pH. Unfortunately, no batch experiments have been performed at SCK•CEN with Pa to confirm the literature data. As sequential migration experiments were however well performed with 231Pa, it was possible to use these data to constrain the phenomenological model (Maes et al., 2011) also for pentavalent actinides.

SCK•CEN/12201513 Page 203 of 208 Compilation of Technical Notes on less studied elements

4.3.3.6 References Berner, U. (2002) Project Opalinus Clay: Radionuclide concentrations limits in the near-field of a repository for spent fuel and vitrified high-level waste. Nagra Technical Report NTB 02-10, Wettingen, Switzerland.

Berry, J.A, Hobley, J., Lane, S.A, Littleboy, A.K., Nash, M.J., Oliver, P. Smitt-Briggs, J.L. and Willams, S.J. (1988): The solubility and sorption of protactinium in the near-field and far-field environments of a radioactive waste repository. Safety Studies Nirex Radioactive Waste Disposal. NSS/R122.

Bradbury, M.H. and Baeyens, B. (2003b) Far-Field Sorption Data Bases for Performance Assessment of a High-Level Radioactive Waste Repository in Opalinus Clay Host Rock. Nagra Technical Report 02-19.

Bradbury, M.H. and Baeyens, B. (2006) Modelling sorption data for the actinides Am(III), Np (V) and Pa(V) on montmorillonite. Radiochimica Act, 94, 619-625.

Bradbury, M.H., Baeyens, B. and Thoenen, T. (2010) Sorption Data Bases for Generic Swiss Argillaceous Rock Systems. Nagra Technical Report 09-03.

Bruggeman, C. and Maes, N. (2016) Radionuclide migration and retention in Boom Clay. External Report, SCK•CEN-ER-0345, SCK•CEN, Mol, Belgium.

Duro, L., Grive, M., Cera, E., Domenech, C., Bruno, J. (2006a) Update of a thermodynamic database for radionuclides to assist solubility limits calculations for PA. Swedish Nuclear Fuel and Waste Management Co, Stockholm, Sweden. SKB Report TR-06-17. Swedish Nuclear Fuel and Waste Management Company. Stockholm, Sweden.

Duro, L., Grivé, M., Cera, E., Gaona, X., Domènech, C. and Bruno, J. (2006b) Determination and assessment of the concentration limits to be used in SR-Can. SKB Technical Report, TR-06-32. Swedish Nuclear Fuel and Waste Management Company. Stockholm, Sweden.

Fourest, B., Perrone, J., Tarapcik, P., Giffaut, E. (2004) The hydrolysis of protactinium (V) studied by capillary diffusion. Journal of Solution Chemistry, 33, 957-973.

Geibeck, W. and Usbeck, R. (2004) Adsorption of thorium and protactinium onto different particle types: experimental findings. Geochimica et Cosmochimica Acta, 69, 1489-1501.

Le Naour, C., Trubert, D., Jaussaud, C., 2003. Hydrolysis of protactinium (V).II. Equilibrium constants at + + - 40o and 60o: A solvent extraction study with TTA in the aqueous system Pa(V)/H2O/H /Na /ClO4 . Journal of Solution Chemistry, 33, 489-504.

Linklater, C.M., Moreton, A.D. and Tweed, C.J. (2003) Analysis and interpretation of geosphere sorption data for a Nirex performance assessment. UK Nirex Report N/083. Harwell, United Kingdom.

ONDRAF/NIRAS (2001) SAFIR 2 Safety Assessment and Feasibility Interim Report 2.

SCK•CEN/12201513 Page 204 of 208 Compilation of Technical Notes on less studied elements

Salah, S. and Wang, L. (2014) Speciation and solubility calculations for waste relevant radionuclides in Boom Clay. First Full Draft. EXTERNAL REPORT SCK•CEN-ER-19814/Ssa/P-16.

Tarapcik, P., Fourest, B., Giffaut, E. (2005) Comparative approach of the solubility of protactinium oxy/hydroxides. Radiochimica Acta, 93, 27-3.

Trubert, D., Le Naour, C., Jaussaud, C., (2002) Hydrolysis of protactinium (V) I. Equilibrium constants at + + - 25°C: a solvent extraction study with TTA in the aqueous system Pa(V)/H2O/H /Na /ClO4 . Journal of Solution Chemistry, 31, 261-277.

Trubert D., Le Naour C., Jaussaud C. and Mrad O. (2003) Hydrolysis of Protactiunium(V). III. Determination of Standard Thermodynamic Data. Journal of Solution Chemistry, 32, 6, 505-517.

Winter, M. (2014) Webelements. University of Sheffield and Webelements Ltd. http://www.webelements.com/protactinium/

SCK•CEN/12201513 Page 205 of 208 Compilation of Technical Notes on less studied elements

5 ANNEX I

This ANNEX was added, as the reader might have noticed that the Pourbaix diagrams and solubility diagrams enclosed in this report were not all calculated with the same geochemical code and thermodynamic database (TDB). As mentioned in the introduction, the compilation of the Technical Notes (TN’s) represented a common effort of QUINTESSA and SCK•CEN. This explains that the style of the TN’s is not uniform. QUINTESSA compiled the TN’s for which no experimental data are available at SCK•CEN and for which also only limited thermodynamic data exist. In order to provide however information on the general geochemistry, the speciation and transport behaviour for the elements belonging to this group (Ac, Ag, Mo, Nb, Ni, Pa, Zr), QUINTESSA performed a detailed literature review and summarized the most relevant outcome in the respective paragraphs. Besides this, different calculation approaches and several TDB’s were used, which explains that also the results may be (slightly) different from the ones obtained by SCK•CEN (Salah and Wang, 2014). It should be mentioned however that no major differences were recognized.

The speciation and solubility calculations performed by QUINTESSA were in general done using the porewater composition from De Craen et al. (2004) at 16°C. Depending on the system, QUINTESSA added other elements (e.g. F, Se) to the reference composition. In contrast, Salah and Wang (2014) performed their calculations using the BC porewater composition recalculated for 25°C, resulting mainly in different pH, Eh and pCO2 values. While calculations by SCK•CEN were all done with GWB v.8 and the in-house developed TDB MOLDATA (2010_MOLDATA nov b.dat), QUINTESSA used different versions of the GWB code and also different databases.

In the Table below, the specific conditions, the code(s) and database(s) used in the speciation and/or solubility calculations for the different elements are summarized.

Table 39: Thermodynamic data and geochemical conditions used by QUINTESSA for speciation and solubility calculations

Geochemical conditions, code, Result of calculation Remark and TDB

Ac Eh-pH diagram of Ac Ac data are currently not (system: Ac-C-S-Cl-F-O-H) available in MOLDATA. Speciation and solubility [Ac] = 10-5, 16°C taken to be identical to GWB v.7/JAEA TDB version 140331 Am.

+ Ag Solubility diagrams for Ag (log a Ag versus AgHS(aq) is predicted to be the Speciation is in pH) dominant aqueous species under agreement with Salah & (system: Ca-activity buffered by calcite, conditions associated with the Wang (2014). sulphate buffered by pyrite, Fe2+ buffered by Boom Clay and native silver is Solubility sequence is the siderite, predicted to be the most stable same when calculated - (least soluble) solid silver phase, log a Cl = -3.155, log fCO2(g) = -2.44, with MOLDATA, but followed by acanthite. Ag2CO3(s) and AgCl(s) log fO2(g) = -71.4) were also considered. Eh = -0.28 mV and pH 8.36, 16 °C, 1 bar Code: GWB v.8 ‘Act2’ module of GWB

SCK•CEN/12201513 Page 206 of 208 Compilation of Technical Notes on less studied elements

(Bethke, 2008) TDB: MINTEQ (thermo_minteq.dat) 2- Mo Eh-pH diagrams of Mo MoO4 is predicted to be the Speciation is in (system Mo-Ca-C-S-Cl-F-O-H) dominant aqueous species. The agreement with Salah & calculations using the MINTEQ Wang (2014). Mean Eh = -341 mV; mean pH = 8.31 and thermo.com.V8.R6 give -5 -8 [Mo] = 10 and [Se] = 10 MoS2 or MoSe2, respectively as MoSe2 was not evaluated GWB v.7/ Visual MINTEQ v.2.4 and possible stable phases under BC by Salah & Wang (2014), conditions. It should be noted LLNL v8.r6 ″combined dataset″ while MoS2 and that while the former is not (thermo.com.V8.R6) additionally Tugarinovite unreasonable, MoSe2 is not a (MoO2,s) were naturally-occurring mineral considered as potential phase. solubility limiting solids under BC conditions.

- Nb Speciation and solubilies calculated with Nb(OH)6 represents MOLDATA (Salah et al. 2014) were dominant aqueous

added species calculated with

MOLDATA

These calculations show that the Speciation of Nb(V) was calculated by A similar solubility at constraining concentrations by least soluble of the considered 25°C for Nb2O5(s) was phases is Nb2O5(s), which would equilibriumwith NaNbO3(s), Nb(cr) or calculated by Salah & constrain the solubility of Nb(V) Nb2O5(s) in the presence of BC Wang (2014), i.e. 2.4×10- to 2.54×10-6 molal. reference porewater at 16°C. 6 M. Code: PHREEQC Interactive v2.18.3514 TDB: Thermochimie v.9

The MINTEQ database predicts 2- Ni Solubility diagrams for Ni Ni(CO3)2 represents that NiCO3(aq) will be the (system: assuming Ca2+ activity buffered by dominant aqueous dominant nickel species under calcite, sulphate buffered by pyrite, Fe2+ species calculated with Boom Clay conditions, as the buffered by siderite, log fCO2(g) = -2.44, log MOLDATA - database does not include a Cl = -3.155, 2- Ni(CO3)2 . log fO2(g) = -71.4 More Ni-solids are Eh = -0.28 mV at pH 8.36, 16 °C, 1 bar The most stable solid in the comprised in Code: GWB v.8 ‘Act2’ module MINTEQ database is MOLDATA, and NiS(gamma), followed by NiS TDB: MINTEQ (thermo_minteq.dat) gaspéite (NiCO3,cr) is (beta) and then NiCO3. considered to represent (violarite and trevorite are not the most probable included in MINTEQ) solubility limiting solid under BC conditions.

Pa No calculations/diagrams provided by Pa(OH)5(aq) represents Quintessa, dominant aqueous But interesting general discussion species calculated with  Diagrams calculated by Salah et al. MOLDATA

(2014) were added Pa2O5(s) considered as solubility limiting solid under BC conditions

Zr Solubility diagrams for Zr In agreement with the Zr(OH)4(aq) 2+ (system: assuming Ca activity buffered by calculations presented by Salah 2+ calcite, sulphate buffered by pyrite, Fe and Wang (2014), Zr(OH)4(aq) is Zr(OH)4 represents buffered by siderite, log fCO2(g) = -2.44, log predicted to be the dominant potential solubility- fO2(g) = -71.4 aqueous species under limiting phase conditions associated with the Eh = -0.28 mV at pH 8.36, 16 °C, 1 bar, fresh Zr(OH)4 : 1.11×

SCK•CEN/12201513 Page 207 of 208 Compilation of Technical Notes on less studied elements

log a Cl- = -3.155. Solute activity buffers are Boom Clay 10-4 M and the same as those used for model Boom Clay  amorphous,aged -8 porewater described by Salah and Wang Zr(OH)4 :1.82×10 M (2014). under Boom Clay Code: GWB v.8 ‘Act2’ conditions TDB: MINTEQ (thermo_minteq.dat)

SCK•CEN/12201513 Page 208 of 208 Compilation of Technical Notes on less studied elements