Tin-113 and Selenium-75 radiotracer adsorption and desorption kinetics in contrasting estuarine salinity and turbidity conditions

Journal of Environmental Radioactivity 213 (2020) 106133

Contents lists available at ScienceDirect

Journal of Environmental Radioactivity journal homepage: http://www.elsevier.com/locate/jenvrad

Tin-113 and Selenium-75 radiotracer adsorption and desorption kinetics in contrasting estuarine salinity and turbidity conditions €ttle c, Teba Gil-Díaz a, b, Frank Heberling c, *, Virginia Keller d, Markus Fuss c, Melanie Bo d a €rg Scha €fer Elisabeth Eiche , Jo a

Universit�e de Bordeaux, UMR CNRS 5805 EPOC, All�ee Geoffroy Saint-Hilaire, 33615, Pessac, France Institute of Geosciences (IGW), Friedrich-Schiller-Universit€ at Jena, Burgweg 11, 07749, Jena, Germany Institute for Nuclear Waste Disposal (INE), Karlsruhe Institute of Technology (KIT), Hermann von Helmholtz Platz 1, 76344, Eggenstein-Leopoldshafen, Germany d Institute of Applied Geosciences (AGW), Karlsruhe Institute of Technology (KIT), Adenauerring 20b, 76131, Karlsruhe, Germany b c

A R T I C L E I N F O

A B S T R A C T

Keywords: Partition ratios (Rd) Gironde estuary Rh^ one river Radionuclide dispersion scenarios

Batch experiments were performed to study adsorption and desorption of 75Se and 113Sn radiotracers at envi­ ronmentally representative concentrations of ~0.3 ng L 1 and ~3 ng L 1, respectively. The radiotracers were incubated with wet bulk sediments from the Gironde Estuary and the Rh^ one River, combining freshwater and coastal seawater salinity (S ¼ 0, S ¼ 32) and three different Suspended Particulate Matter (SPM) concentrations (10 mg L 1, 100 mg L 1, 1000 mg L 1) to simulate six hydrologically contrasting situations for each particle type. Results showed no measurable adsorption for 75Se under the experimental conditions, whereas >90% of 113 Sn rapidly adsorbed onto the particles during the first hours of exposure. Adsorption efficiency increased with increasing SPM concentration and seemed to be slightly greater for the Rh^ one River sediments, potentially related to the intrinsic mineral composition. Desorption of spiked sediments exposed to filtered, unspiked freshwater and seawater only occurred for 113Sn (<15% of the previously adsorbed 113Sn) in the Garonne River sediments. This study provides insights to the potential environmental behaviour of hypothetical radionuclide releases of Se and Sn into highly dynamic and contrasting aquatic systems. Multiple accidental scenarios for the case of the Gironde Estuary and the Rh^ one River are discussed. These scenarios suggest that the environmental fate of soluble radionuclides like Se will be associated to water hydrodynamics and potentially more bioavailable whereas highly particle-active radionuclides like Sn will follow natural river/estuarine sedimentary regimes. Information on reactivity of radionuclides is important for improving the precision of current approaches aiming at modelling environmental radionuclide dispersion in continent-ocean transition systems.

1. Introduction Radioactive selenium (Se) and tin (Sn) isotopes are commonly pre­ sent in spent nuclear fuel forming high-level radioactive waste (HLW), generally stored in geological repositories (Asai et al., 2011, 2013). Environmental releases of Se and Sn can either occur from the leakage of radioactive wastes or directly from potential nuclear power plant acci­ dental events (Morewitz, 1981; Abe et al., 2014). Given the long-term persistence of some specific isotopes (e.g., half-lives of 3.27.105 y for 79 €rg et al., 2010; and 2.3.105 y for 126Sn), current concern lies on Se, Jo their potential releases into the environment and the related environ­ mental mobility, bioavailability and biological roles. Short-lived

radionuclides (e.g., half-lives of 9.64 d for 125Sn, 2.11 h for 127Sn) may also be released to the environment during nuclear power plant (NPP) accidents and may show non-negligible impacts, relevant at short timescales in aquatic systems (i.e., tidal cycles, seasonal phenomena, etc.). In fact, stable Se is a micronutrient and plays a fundamental role in major metabolic pathways including hormone metabolism, antioxidant defence, and Hg-triggered detoxifying mechanisms, especially in aquatic organisms (Cuvin-Aralar and Furness, 1991; Zeng, 2009). Stable Sn is required in very small quantities for organism welfare (Eichenberger, 1986). Accordingly, organisms exposed to radioactive Se and Sn iso­ topes can also take up these isotopes in significant amounts at different

* Corresponding author. E-mail addresses: [email protected] (T. Gil-Díaz), [email protected] (F. Heberling), [email protected] (V. Keller), [email protected] (M. Fuss), [email protected] (M. B€ ottle), [email protected] (E. Eiche), [email protected] (J. Sch€ afer). https://doi.org/10.1016/j.jenvrad.2019.106133 Received 24 August 2019; Received in revised form 28 November 2019; Accepted 7 December 2019 Available online 19 December 2019 0265-931X/© 2019 Elsevier Ltd. All rights reserved.

T. Gil-Díaz et al.

Journal of Environmental Radioactivity 213 (2020) 106133

levels of the trophic chain (Konovalenko et al., 2014). To date, little is known about the environmental behaviour and dispersion of radioactive Se and Sn isotopes (i.e., at ppt levels) in aquatic environments, partic­ ularly in continent-ocean transition systems with highly dynamic salinity and turbidity gradients. Speciation determines the mobility of trace elements and their interaction/exchange between dissolved and particle phases. The most studied inorganic species of Se (Se(IV) and Se(VI)) are water soluble and present as oxyanions in environmental conditions (i.e., circumneutral pH, oxic media; He et al., 2010). From these species, Se(VI) generally shows lower adsorption than Se(IV), and respective complexation mechanisms include the formation of outer-sphere (physisorption) and inner-sphere (chemisorption) complexes (e.g., Martínez et al., 2006; Jordan et al., 2011). In natural sediments presenting low content in organic matter (e.g., <20%; Coppin et al., 2009; Tolu et al., 2014), several mineral phases are involved in Se sorption (e.g., apatite, iron oxides, iron sulphides and clay minerals; Duc et al., 2003; He et al., 2010; Mitchell et al., 2013). In addition, natural systems present other major anions in solution (i.e., sulphate or phosphate, especially in brackish/saline environments), which can further influence Se sorp­ tion/desorption processes (particularly for Se(VI), e.g., Fujikawa and Fukui, 1997; He et al., 2010). In contrast, there is a lack of knowledge concerning the environ­ mental behaviour and sorption of inorganic Sn to mineral phases, mainly due to the relative importance of its organic species (e.g., tributyltin, etc.) and the focus of environmental research on their ecological im­ plications in aquatic systems (Rüdel, 2003). Studies focusing on the main redox species of inorganic Sn (Sn(II) and Sn(IV)) include batch experiments showing high sorption onto mineral phases of soils and rocks (e.g., Al/Fe-(hydr)oxides, bentonite, tuff and granodiorite; Tick­ nor and McMurry 1996; Oda et al., 1999; Kedziorek et al., 2007; Nakamaru and Uchida 2008), synthetic materials (e.g., nano crystalline calcium hydroxyapatite; Ghahremani et al., 2017), and cementitious systems (e.g., ettringite and CSH phases Bonhoure et al., 2003). It is considered that Sn(II) is easily oxidised into Sn(IV) and that Sn(IV) has low mobility due to SnO2(s) stability (S�eby et al., 2001). In environ­ mental conditions, these species are mostly present in neutral forms (e. g., Sn(OH)4) and may show relevant proportions of hydroxo complexes (e.g., Sn(OH)-5, Sn(OH)26 ) at pH > 7.5 (S� eby et al., 2001). The aim of this study is to understand the water-particle distribution of inorganic Se and Sn with natural/environmental particles in con­ trasting salinity and turbidity conditions, as observed in the Gironde ^ne River (hosting 4 NPPs and one Estuary (hosting 2 NPPs) and the Rho nuclear fuel reprocessing facility). The obtained results are then inte­ grated in qualitative simple radionuclide dispersion scenarios antici­ pating potential accidental releases.

2. Material and methods 2.1. Areas of study The Gironde Estuary is a major continent-ocean transition system in Europe with an estuarine surface area of 635 km2 at high tide and an associated watershed of 81 000 km2 (Fig. 1a). It is geographically delimited by the confluence of two main rivers: the Garonne River (freshwater discharges from 302 m3 s 1 to 880 m3 s 1) and the Dor­ dogne River (discharging between 119 m3 s 1 and 477 m3 s 1; 1959–2017, DIREN, 2018). The semidiurnal tidal cycle of 12 h 25min extends the marine influence to the upstream limit of the dynamic tide at La R� eole on the Garonne River (Fig. 1a). Average annual solid discharges €fer et al., of 2–4 Mt enter the Gironde Estuary (Castaing, 1981; Scha 2002). Average Suspended Particulate Matter (SPM) concentrations are ~100 mg L 1 and the strong tidal hydrodynamics develop a Maximum Turbidity Zone (MTZ) of up to 6 Mt of SPM with SPM concentrations of >1000 mg L 1 (Castaing, 1981; Sottolichio and Castaing, 1999). This MTZ is subjected to seasonal migrations between the estuary mouth and the fluvial estuary, upstream the city of Bordeaux (Sottolichio and Castaing, 1999). Massive expulsion of the MTZ to the coastal area only occurs for specific hydrological conditions (i.e., high water discharge during several weeks and high tidal coefficient; Castaing and Allen, 1981; Doxaran et al., 2009). Coastal dispersion through the estuarine plume shows two density layers (surface vs. bottom), which can settle off the Gironde Estuary mouth and/or be transported north-westwards or southwards according to the seasonal coastal currents (Froidefond et al., 1998; Jouanneau et al., 1999). In the Gironde Estuary, water residence times vary from ~18 days during high discharge conditions to 86 days in low discharge conditions, whereas average residence time of SPM is 1–2 years (Castaing and Jouanneau, 1979; Jouanneau and Latouche, 1981). ^ne River is one of the largest rivers in Europe (i.e., 816 km The Rho long, Fig. 1b) and since 1968 (i.e., construction of the Aswan dam in the River Nile; Ollivier et al., 2010) the main fluvial system contributing freshwater to the Mediterranean Sea. Its watershed of 98 800 km2 pro­ vides an average freshwater discharge of 1700 m3 s 1 (i.e., at Beaucaire, 60.5 km from the river mouth) and the contained nutrients support 20–70% of the coastal primary production of the west Mediterranean Sea (Lefevre et al., 1997; Ludwig et al., 2009). Annual solid discharges vary from 2 to 20 Mt (average 4.6 Mt y 1), of which 70% occur during flood events (Pont, 1997; Pont et al., 2002; Marion et al., 2010). Nevertheless, the SPM concentrations at Beaucaire range from 10 to ~ ez, 2005; Poulier et al., 2019) and particle trans­ <200 mg L 1 (Peri� an port in general is highly influenced by 21 hydroelectric dams con­ ^ne River (Poulier et al., 2019). structed along the main course of the Rho At 50 km from the river mouth, the river splits into two branches,

Fig. 1. Studied sites. Map of (a) the Lot-Garonne-Gironde fluvial-estuarine system and (b) the Rh^ one River. Main cities (squares), sampling sites (circles), rivers and kilometric points (KP) are also shown. 2

T. Gil-Díaz et al.

Journal of Environmental Radioactivity 213 (2020) 106133

^ne” (52 km long, draining 85% of the water forming “le Grand Rho ^ne” (60 km long; Ollivier et al., discharge during floods) and “le Petit Rho 2010). At the shore, a complex intrusion process takes place as the freshwater flows into the marine waters in the absence of tidal influence. Flocculation and aggregation of particles form a pro-delta of fine-grained deposits and sand bars (Aloisi and Monaco, 1975), showing average sediment accumulation velocities of 20–50 cm y 1 and sub­ jected to physical resuspension events (i.e., mainly by storms, waves and winds; Marion et al., 2010). The associated suspended plume (e.g., ~30 mg L 1 SPM) shows a 1–2 m thickness and seasonal, well-defined freshwater-seawater stratification of 10–20 units of difference with surrounding seawater, extending up to 20–30 km offshore (Broche et al., 1998). This plume is initially oriented in a southeast direction but generally rotates westwards moving along the Mediterranean coast due �n ~ ez, 2005). to Coriolis acceleration (Peria

2.4. Experimental design Radionuclide solutions of 113Sn (SnCl4 in 6 M HCl, >99% purity, Eckert & Ziegler) and 75Se (H2SeO3 in 0.1 M HCl, >99% purity, Eckert & Ziegler) were diluted, independently, in freshwater and seawater matrices several weeks before the experiments in order to homogenise/ stabilise the radionuclides. The high acidity from 113Sn was corrected with NaOH solution (1 M, Merck®) and 75Se was oxidised to Se(VI) with H2O2 (30% Merck®). Batch experiment slurries were prepared by mix­ ^ne River ing wet bulk sediments from the Gironde Estuary and the Rho with filtered water from the Garonne River (S ¼ 0) and the Arcachon Bay (S ¼ 32), simulating three different SPM concentrations: 10 mg L 1 (i.e., 10.7 � 0.4 mg L 1), 100 mg L 1 (i.e., 105 � 4 mg L 1), and 1000 mg L 1 (i.e., 973 � 60 mg L 1). The number of sample replicates was N ¼ 5 for 10 mg L 1, N ¼ 6 for 100 mg L 1 and N ¼ 3 for 1000 mg L 1. Adsorption kinetic investigations started when the standard solu­ tions of 113Sn (nominal 1000 Bq mL 1, equivalent to ~3 ng L 1) and 75Se (nominal 100 Bq mL 1, equivalent to ~0.3 ng L 1) were added, together, into the slurries of 30 mL. Aliquots were sampled at 0 h, 2 h, 4 h, 6 h, 24 h and 48 h. Experimental blanks, i.e., spiked freshwater and seawater with no sediments, were also followed over time to verify adsorption onto reactor walls. Desorption experiments were performed with the slurry samples containing 100 mg L 1 of sediments, after centrifugation and emptying the initial water content. The desorption conditions at 100 mg L 1 of sediments were considered to be the best compromise between experimental conditions (e.g., concentrations, expected element behaviour, analytical limits of detection, etc.) and environmental representativeness. Out of the 6 replicates, three were used for desorption experiments in unspiked freshwater and the other three in unspiked seawater for 48 h, sampling at 0 h, 2 h, 4 h, 24 h and 48 h. This means that sediments that were previously exposed to 75 Se/113Sn adsorption in freshwater conditions were desorbed in clean freshwater (N ¼ 3) and seawater (N ¼ 3). The similar treatment was applied to samples previously exposed to 75Se/113Sn in seawater. The potential contribution of radionuclide desorption from the tube walls during the desorption experiments was also checked under the same experimental conditions from the blank tubes. Batch experiments were performed in acid-washed 50 mL poly­ propylene (PP) tubes with sporadic manual shaking at room tempera­ ture (~28 � C) and neutral pH (7.0–7.2). Sampling consisted of collecting 1.5 mL of homogeneous sample at each sampling time, followed by centrifugation at 4000–6000 rpm during 20 min for SPM-liquid sepa­ ration. Supernatant water was then recovered with a 1 mL pipette and stored in 10 mL light-density polyethylene (LDPE) Kautex™ gamma vials with 9 mL HCl 2% (Merck®, to avoid Sn precipitation in HNO3; Weber, 1985) for analysis.

2.2. Sample collection Freshwater from the Garonne River and seawater from the Arcachon Bay, were filtered through 0.45 μm Teflon filters (FHLC, Merck Millipore Ltd.) and stored at 4 � C. Wet sediments from the Garonne River banks were collected manually during low-tide at Portets, ~30 km upstream ^ne River from Bordeaux (PK-30, Fig. 1a). Wet sediments from the Rho ^ne en were collected at the SORA station (Station Observatoire du Rho ^ne) by the IRSN (Institut de Arles, ~40 km upstream le Grand Rho Radioprotection et de Sûret�e Nucl�eaire, Cadarache) with active, continuous pumping from a floating arm at ~50 cm below the water surface into a sediment trap (TAS; Masson et al., 2018). Water content of wet sediments was evaluated, before preparing the experimental slur­ ries, by comparing the masses of precise volumes of wet and dry sedi­ ment aliquots (oven dried at 70 � C). Suspended sediments of the Gironde Estuary show characteristic contents of particulate organic carbon (POC) between 0.05 and 1.5% and grain sizes of mainly silts (i.e., grain size diameters ranging between ~3 and ~30 μm for surface SPM, increasing up to ~470 μm in estuarine bottom sediments; Doxaran et al., 2003; Coynel et al., 2016) whereas the sediments collected at SORA ^ne River show POC contents varying from 1.5 to 3.5% station in the Rho (Thollet et al., 2018) and grain-sizes of 10–15 μm (in the silt range). 2.3. Sediment characterisation Major sample mineralogy (>1–5 wt%) of sediments from both sites was characterized from un-oriented oven-dried (50 � C) sediment pow­ der with X-Ray Diffraction (XRD Bruker D8 DISCOVER, AGW-KIT, Germany) using Cu Kα radiation (40 kV/40 mA). Diffractograms were recorded in an angular range from 2� to 82� 2θ with step width 0.02� and counting time of 0.4 s/step. Results were compared to international peak lists of minerals (Crystallography Open Database QualX and the Crystal Structure Data Base for Minerals WWW-MinCryst) to identify the major mineralogical composition. The minerals were identified using the EVAsoftware programme (Bruker) and were quantified with a Rietveld approach as implemented in the TOPAS software (Bruker). All quanti­ fications have to be considered with a relative uncertainty of �10–20%. Further identification of clay minerals was performed on textured preparations of sediment aliquots (Eiche et al. unpublished). Therefore, unconditioned sediment samples were put in slurry with a 10% ammonia solution, dispersed in an ultrasound bath for 15 min, gravi­ metrically separated in suspension (1 h) and left to dry onto glass plates for analyses. From each sample, three glass plates were prepared in parallel. One aliquot was measured without further treatment, the sec­ ond aliquot was exposed to ethylene glycol vapour atmosphere at 60 � C for 24 h and the third sample aliquots was further burned at 550 � C for 3 h (muffle oven). Finally, diffractograms were recorded in an angular range from 2� to 22� 2θ.

2.5. Radiotracer gamma analyses Both 75Se and 113Sn activities were analysed in the initial matrices (experimental background levels) and in the experimental batch kinetic samples using a High Purity Germanium detector (HPGe, ORTEC®, INEKIT, Germany). The activity of 75Se was directly measured with the 264.7 keV energy line, whereas that of 113Sn was measured indirectly from its daughter isotope, 113mIn, at the 391.7 keV line. The gamma detector was calibrated with a certified Multinuclide Standard Solution (2 M HCl, Eckert & Ziegler) and showed detection limits of 0.03 Bq mL 1. All gamma results were mathematically corrected to a unique date to overcome the decrease in signal related to radionuclide decay (i. e., half-lives of 120 d for 75Se and 115 d for 113Sn). 2.6. Solid/liquid distributions and adsorption percentages The ratio between the solid/liquid distributions of radiotracers in the batch experiments was calculated in comparison to the dissolved con­ centrations in the blank solutions (Eq. (1)). Ratios for both adsorption 3

T. Gil-Díaz et al.

Journal of Environmental Radioactivity 213 (2020) 106133

and desorption experiments were calculated and compared. These values were not similar during the experimental timeframe of 48 h, justifying the use of “Rd” (e.g., solid/liquid concentration ratio) in this study instead of “Kd” (i.e., equilibrium solid/liquid partitioning per se). In fact, the term Kd is generally reserved for fast, reversible and solute independent distribution coefficients whereas the term Rd should be used for empirical measurements (McKinley and Alexander, 1992). Nevertheless, many environmental studies use “Kd” instead of “Rd” for both field and experimental distribution coefficients, thus, both types of ratios will be used and compared in the discussion section. Sorption percentages were also calculated (Eq. (2)). Rd ¼

Ablank d

Asample d

^ne SPM showed the following major minerals: the signal. The Rho phyllosilicates (~23%, mainly illite and muscovite with 18%, 1% kaolinite, 4% chlorite, and <0.5% smectite), plagioclase/feldspars (~14%), carbonates (~28%) and quartz (~33%). In this case, the amorphous minerals accounted for ~13% of the signal. Clay minerals such as kaolinite, illite, micas (e.g., muscovite), chlorite and smectite (e. g., montmorillonite) were readily observed in the texturised prepara­ tions (Supplementary materials, Fig. S1). The results of major mineral­ ^ne SPM fall within the ranges already ogical composition from the Rho reported by Pont et al. (2002). 3.2. Kinetic adsorption experiments

(1)

Asample ⋅SPM d

Natural radioactive background present in the Gironde Estuary and ^ne River sediment slurries had no relevant impact on 75Se activities Rho and contributed with activities of 0.30 Bq mL 1 to the 113Sn signals. Experimental blanks suggest constant 75Se concentrations along both, sorption and desorption experiments for all water matrices (Fig. 3a). However, blanks of 113Sn showed 50–90% loss of Sn during the exper­ iment, more important in freshwater than seawater matrices (Fig. 3b). Results from 48 h Se adsorptions showed rather similar 75Se activ­ ities in sample slurries and average blanks (Table 1). The resulting estimated Se adsorptions for both Gironde and Rh^ one sediment types were �5%, falling within the experimental error. Estimated solid/liquid partitioning for the 1000 mg L 1 sediment conditions showed Rd values < 85 L kg 1 for both freshwater and seawater matrices in Gironde and ^ne sediment types (Table 1). Rho Given the 113Sn losses observed in the experimental blanks, only the results from 113Sn adsorption obtained during the first 4–6 h in all slurries were considered and averaged. Adsorption percentages were calculated by comparing the obtained dissolved 113Sn activities in each experimental slurry to the average activities found in the blanks (Table 2) in order to correct for the ~50% Sn loss likely attributed to adsorption to container walls (though low; Weber, 1985) in the first 6 h of the experiment. Loss of Sn by precipitation is unlikely given that the spike concentrations were ~50 fold (cassiterite) and ~300 fold (SnO2(am)) undersaturated (i.e., PHREEQC calculations showing satu­ ration indices of 1.70 for cassiterite and 2.5 for amorphous SnO2). This approach assumes that experimental conditions causing Sn losses are comparable between slurries and blanks. Adsorption of 113Sn in all sediment concentrations was fast, as average activities from the first 4–6 h of the adsorption kinetics compared to average blank activities showed >90% adsorption for both sediments (Table 2). Estimated average sol­ id/liquid partitioning showed freshwater Rd values ranging from 187 000 to 7 590 000 L kg 1 and seawater Rd values ranging from 95 700 to 2 990 000 L kg 1 for both sediments (Table 3).

!

sorptionð%Þ ¼ 1

Asample d ⋅100 Ablank d

(2)

where “A” stands for activities (i.e., 75Se or 113Sn), index “d” referring to the dissolved phase and “SPM” for the concentration of suspended particle matter used in each condition (i.e., 10, 100 and 1000 mg L 1). Particle grain-sizes have an impact on the reactivity/affinity of ele­ ments to the particle phase (e.g., Loring and Rantala, 1992). This means that Rd values also change with grain-size (i.e., the smaller the particle, the higher the surface area, thus, on small particles higher sorption and greater Rd may be expected than for larger particles). However, the bank sediments used during the experiments of the Gironde Estuary showed average grain-sizes of 12 � 1.5 μm (i.e., within the range of surface suspended sediments; Doxaran et al., 2003), and the results will be further discussed in the context of scenarios of suspended particles (SPM). This means that the reported Rd values for the Gironde Estuary represent the expected Se and Sn solid/liquid partitioning at environmentally-representative spike concentrations and solid/liquid matrices. 3. Results 3.1. Bulk sediment mineralogy The XRD analyses performed in aliquots of the original sediments showed similar major mineralogy composition though in different pro­ portions per SPM type. Diffractograms have been normalised to the quartz signal to compare both sediments (Fig. 2). The Garonne/Gironde SPM shows a predominance of phyllosilicates (i.e., accounting for ~50% of relative abundance, with illite and muscovite dominating the clay fraction with 34%, along with 9% smectite, 6% kaolinite and <1% chlorite), quartz minerals (~25% relative abundance), (Ca-) plagio­ clases and (K-) feldspar (~15%), with relatively few carbonate minerals (calcite/dolomite in <10%). Amorphous minerals account for 36% of

Fig. 2. XRD diffractograms. XRD patterns from bulk samples in the Gironde Estuary (KP-30) and Rh^ one River (TAS). Abbreviations: Cal ¼ Calcite (carbonate, CaCO3); Chl ¼ Chlorite (Fe(II) containing silicate, (Fe,Mg,Al)6(Si,Al)4O10(OH)8); Feld ¼ Feldspars (tectosilicate, (K,Na)AlSi3O8–CaAl2Si2O8); Ill ¼ Illite (phyllosi­ licate, (K,H3O)(Al,Mg,Fe)2(Si,Al)4O10[(OH)2,(H2O)]); Kao ¼ Kaolinite (phyllosilicate, Al2Si2O5(OH)4); Smec ¼ Smectite (e.g., Montmorillonite, phyllosilicate (½Ca, Na)(Al,Mg,Fe)4(Si,Al)8O20(OH)4⋅nH2O); Musc ¼ Muscovite (phyllosilicate, KAl2(AlSi3O10)(OH,F)2); Q ¼ Quartz (SiO2). 4

T. Gil-Díaz et al.

Journal of Environmental Radioactivity 213 (2020) 106133

Fig. 3. Radiotracer blank kinetics. Temporal distribution of (a) 75Se and (b) 113Sn experimental blank activities in freshwater (blue triangles) and seawater (blank diamonds) during sequential adsorption and desorption slurry kinetic experiments (~125 h in total). Nominal spike concentrations were 100 Bq mL 1 for 75Se and 1000 Bq mL 1 for 113Sn. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) Table 1 one Average activities from the 48 h kinetic experiments of75Se sorption in sediment slurries (10 mg L 1, 100 mg L 1, 1000 mg L 1) with the Gironde Estuary and Rh^ River sediments in freshwater and seawater matrices. Maximum solid/liquid partitioning (Rd in L kg 1) for 1000 mg L 1 conditions are also shown. Activities (Bq mL 1)

Gironde Estuary sediments

Batch conditions:

Average

SD

Adsorption

Average

SD

Adsorption

88 87 87 84 84.1b 86 87 86 83 61.2

2 2 0 1 1 2 0 1 -

– 2% 2% 5% – 0% ( 1%) 0% 4% -

88 90 91 89 9.00 85 87 87 85 15.7

1 1 2 1 1 0 0 1 -

– 0% 0% 0% – – 0% 0% 0% –

75

Freshwater Se

Seawater75Se

a b

Blank 10 mg L 1 100 mg L 1 1000 mg L 1 Max. Rd Blank 10 mg L 1 100 mg L 1 1000 mg L 1 Max. Rd

Rh^ one River sediments

( 2%)a ( 4%)a ( 1%)a ( 2%)a ( 3%)a

Negative values in brackets correspond to activities in blanks lower than samples, within the analytical uncertainty. Maximum solid/liquid partitioning is calculated for 1000 mg L 1 as lower SPM ratios are within the uncertainty.

Table 2 Average activities from the first 4–6 h kinetic experiments of113Sn sorption in sediment slurries (10 mg L 1, 100 mg L Rh^ one River sediments in freshwater and seawater matrices. Activities (Bq mL 1)

Gironde Estuary sediments

Batch conditions: Freshwater113Sn

Seawater113Sn

Blank 10 mg L 1 100 mg L 1 1000 mg L 1 Blank 10 mg L 1 100 mg L 1 1000 mg L 1

1

, 1000 mg L 1) with both Gironde Estuary and

Rh^ one River sediments

Average

SD

Adsorption

Average

SD

Adsorption

511 7 19 2 601 43 24 3

115 1 11 0 103 15 6 2

– 99% 96% 100% – 93% 96% 100%

157 13 3 1 471 15 12 5

43 9 0 0 55 7 9 4

– 92% 98% 99% – 97% 97% 99%

Desorption of 113Sn was both sediment-dependent and salinityindependent (Fig. 5). Except for some anomalous data, potentially related to Sn desorption from tube walls (Fig. 5a,c), 113Sn desorption from the Garonne River sediments was 5–10 Bq mL 1, corresponding to 8–13% of the spiked 113Sn. Almost 10-fold lower 113Sn desorption was obtained from the Rh^ one sediments (Fig. 5b,d), desorbing <1% of the originally spiked 113Sn. In all cases, Rd values at the end of 48 h of desorption were not equivalent to the respective Rds of adsorption (section 3.2.), i.e., desorption Rds were always in average 60–80% in the ^ne River sediments Garonne River sediments and in 70–95% in the Rho higher than the original adsorption Rds for SPM concentrations of 100 mg L 1 (Table 3). These differences were slightly lower when adsorption and desorption conditions were alike, e.g., when 113Sn was adsorbed onto SPM in freshwater conditions, the resulting Rd of desorption in seawater was ~70% higher than the adsorption Rd but only ~60% higher for desorption of the same particles in freshwater. Desorption Rds

3.3. Kinetic desorption experiments Desorption of 75Se from blank tube walls ranged between 0.5 and 0.7 Bq mL 1 in freshwater (corresponding to <0.7% of the spiked 75Se) and below detection limits in seawater. Desorption of 113Sn from blank tube walls in freshwater and seawater showed activities of ~60 Bq mL 1 and ~15 Bq mL 1, respectively, corresponding to 10–40% and <6% of the observed losses from the experimental blanks. Results from 75Se desorption are considered as indicative given (i) the previously observed, very low 75Se adsorption in the slurries, and (ii) the low activities registered during desorption tests (Fig. 4a and b), close to the expected contribution of the spiked solution from the previous adsorption conditions, left behind in the experiment tube and diluted with the clean water during desorption (“estimated residual”). There­ fore, if any 75Se adsorbed into the studied sediments, <2% of the orig­ inally spiked 75Se was desorbed in clean water (fresh or seawater). 5

T. Gil-Díaz et al.

Journal of Environmental Radioactivity 213 (2020) 106133

Estuary, 106-fold variations in concentrations of dissolved Se induce changes of one order of magnitude in the Kd values. However, these differences seem to fall within the uncertainty ranges of Kd (i.e, one to two orders of magnitude; Ciceri et al., 1988). Further studies should focus on better understanding and quantifying Kd values at represen­ tative concentrations as laboratory studies can represent/replicate anthropogenic releases into the environment as “added” or “spiked” elements (vs inherent element behaviour from field studies), funda­ mental for environmental risk assessment. In general, Se is a relatively mobile element in environmental com­ partments and may show an unreactive character in estuarine salinity and turbidity gradients (Seyler and Martin, 1991). Total dissolved Se concentrations along estuarine salinity gradients mostly range 0.04–2.00 μg L 1 and show linear distributions, negatively-related with salinity (e.g., Measures and Burton, 1978; van der Sloot et al., 1985; Cutter, 1989; van den Berg et al., 1991; Seyler and Martin, 1991; van den Berg, 1993). Few studies report its non-conservative, generally seasonal, behaviour in estuarine systems, greatly related to biological activities (i.e., low-salinity or mid-estuarine maximum, e.g., Cutter, 1989; Yao et al., 2006; Chang et al., 2016). Our kinetic experiments suggest that, in the absence of biological mediated processes, about 5%– ~20% of the natural Se concentrations (i.e., ranging between 0.3 ng L 1 and 100 μg L 1, with Rd ranging between 10 and 630 L kg 1) may adsorb to natural SPM from the Gironde Estuary presenting solid/liquid (SPM) ratios between 10 and 1000 mg L 1. This observation implies that <20% of Se will be associated to SPM, thus, transported/immobilized in marine sediments in this continent-ocean transition system. Sorption isotherms of stable Se species in natural soils show both Langmuir and Freundlich behaviours, depending on the nature of the particle phase (Dhillon and Dhillon, 1999). This means that Se can adsorb onto different mineral phases depending on the particle composition. According to Tan et al. (1994), based on Langmuir models with oxides and monolayered silicates, sorption of Se(IV) follows the sequence: oxides (Fe2O3 > MnO2 > Al2O3) > 1:1 layer silicates (e.g., kaolinite) > 2:1 layer silicates (e.g., muscovite/biotite, chlorite, smec­ tite). Martínez et al. (2006) and Rovira et al. (2008) also observed Se(VI) outer-sphere complexation onto natural Fe mineral phases such as magnetite, hematite and goethite. Nevertheless, this adsorption from the dissolved into the particle phase was generally <5% that of the original dissolved concentrations (Martínez et al., 2006; Rovira et al., 2008), i.e., in a similar range as the results presented in this work. This low adsorption observed for Se may be related to (i) the low spike concen­ trations, and/or (ii) the low content of Fe minerals (<3%) in both Gar­ ^ne sediments (Gil-Díaz et al., 2019b). Once in the onne/Gironde and Rho seawater endmember, desorption of Se from mineral phases may follow the inverse sequence (silicates > MnO2 > Fe2O3) and account for up to 10% of the original dissolved concentration (Kharkar et al. 1968), i.e., ^ne sediments. <2% for the case of the Garonne/Gironde and Rho Initial environmental concentrations of Se in the water matrices used

Table 3 Average solid/liquid ratios for the first 4–6 h after adsorption of113Sn in sedi­ ment slurries (10 mg L 1, 100 mg L 1, 1000 mg L 1) of the Gironde Estuary and Rh^ one River sediments in freshwater and seawater matrices. Average solid/ liquid ratios at the end of the desorption experiments with previous slurries of 100 mg L 1 are included. Minimum and maximum values are shown in brackets. Rds (kg L 1)

Gironde Estuary sediments

Rh^ one River sediments

Batch conditions:

Average (min – max)

Average (min – max)

7 590 000 (6 390 000–9 320 000) 259 000 (160 000–632 000) 321 000 (267 000–401 000) 1 300 000 (939 000–2 070 000) 245 000 (194 000–330 000) 221 000 (137 000–557 000) 677 000 (462 000–1 260 000)

1 100 000 (619 000–3 460 000) 587 000 (526 000–664 000) 187 000 (146 000–260 000) 2 990 000 (2 060 000–5 350 000) 373 000 (208 000–1 550 000) 95 700 (52 000–505 000) 2 000 000 (1 630 000–2 570 000)

834 000 (600 000–1 370 000)

2 300 000 (1 930 000–2 830 000)

1 300 000 (943 000–2 110 000)

10 400 000 (8 050 000–14 600 000) 4 810 000 (3 310 000–8 730 000)

Adsorption of113Sn in freshwater

10 mg L

1

1

100 mg L

1

1000 mg L Adsorption of113Sn in seawater

10 mg L

1

100 mg L 1000 mg L

SPM from adsorption in freshwater

SPM from adsorption in seawater

1 1

100 mg L 1 (desorption in freshwater) 100 mg L 1 (desorption in seawater) 100 mg L 1 (desorption in freshwater) 100 mg L 1 (desorption in seawater)

1 130 000 (881 000–1 580 000)

from 113Sn adsorbed in freshwater conditions were always lower than those from 113Sn adsorption in seawater. 4. Discussion 4.1. Selenium estuarine reactivity Radiotracer sorption kinetics of Se at ultra-trace levels (~0.3 ng L 1) showed negligible adsorption, with estimated Rd (or Kd) values < 85 L ^ne River sediments. A similar kg 1 for both the Gironde Estuary and Rho study on Se adsorption kinetics performed with stable 77Se(VI) for the same Gironde Estuary sediments (i.e., 100 and 1000 mg L 1) and con­ trasting water matrices at nominal concentrations of 100 μg L 1 of Se showed Kds of ~315 L kg 1 in freshwater and of 400–630 L kg 1 in seawater adsorptions (Gil-Díaz et al., 2019a). These observations sug­ gest that, in environmental conditions representative of the Gironde

Fig. 4. Se radiotracer desorption kinetics. Desorption kinetics of 75Se in freshwater (fw, blue) and seawater (sw, black) conditions from Gironde Estuary (filled circles) and Rh^ one River (empty squares) sediments previously adsorbed in (a) freshwater and (b) seawater conditions. Error bars correspond to standard deviations between repli­ cates (N ¼ 3). The estimated maximum contribution by left-over solution from previous adsorption experiment, potentially influencing 75Se activities during desorption (red dotted line), is also shown. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

6

T. Gil-Díaz et al.

Journal of Environmental Radioactivity 213 (2020) 106133

Fig. 5. Sn radiotracer desorption ki­ netics. Desorption kinetics of 113Sn in freshwater (fw, blue) and seawater (sw, black) conditions from Gironde Estuary (filled circles) and Rh^ one River (empty squares) sediments previously adsorbed in (a,b) freshwater and (c,d) seawater conditions. Error bars correspond to standard deviations between replicates (N ¼ 3). The estimated maximum contribution from left-over solution from previous adsorption spiked water, potentially influencing 113Sn activities during desorption (red dotted line), is also shown. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

for this experimental work were 0.14 � 0.03 μg L 1 in freshwater and ~0.31 μg L 1 for seawater (Gil-Díaz et al., 2019a). Consequently, the 75 Se radiotracer spikes (~0.3 ng L 1) to the water matrices were 103-fold lower than environmental concentrations. The expected con­ centrations of potential Se radionuclide releases by hypothetical nuclear power plant accidents also would be orders of magnitude lower than environmental concentrations. For example, the Fukushima Dai-ichi accident discharged into the coastal environment 137Cs activities ranging between 10 Bq L 1 and 68 000 Bq L 1 (i.e., equivalent to ~0.003 ng L 1 and 20 ng L 1 of 137Cs; Buesseler et al., 2017). Although no such data exists for Se or Sn radionuclides, one may reasonably as­ sume that the expected releases would be in the ultra-trace concentra­ tion levels. In addition, long-term storage repositories may release long-lived Se-isotopes, which belong to the instant-release fraction like e.g. Cs- and I-isotopes (i.e., they belong to the inventory of safety-relevant radionuclides showing rapid release from fuel materials when the canister is breached; Garisto et al., 1989). Accordingly, and given that equilibrium conditions were reached in <48 h, the results obtained from the adsorption experiments suggest that dissolved ^ne radioactive Se isotopes released into the Gironde Estuary or the Rho River would mostly remain in solution within the estuarine reaches, independent from the turbidity and salinity gradients, continuously diluted with seawater during estuarine mixing, before reaching the coastal waters. The temporal scale of this discharge to the coastal waters would depend on the residence times of water masses in these aquatic systems (e.g., differences between dry and wet seasons in the Gironde Estuary show contrasting water residence times of 86 days and 18 days, respectively). The potential radiological risk to aquatic organisms due to beta and gamma emissions of dissolved Se-radioisotopes is associated to its potential bioavailability (i.e., Se biomethylation, Amouroux and Donard, 1997; nutrient-type behaviour; Measures and Burton, 1980; Cutter and Cutter, 1995) and to external exposure (expected to show

dose coefficients of one order of magnitude higher than those of internal exposure; Ulanovsky et al., 2017). 4.2. Tin estuarine reactivity Sorption of Sn at ultra-trace levels (~3 ng L 1) showed efficient adsorption of >90% of spiked Sn for both, the Gironde Estuary and the ^ne River sediments, generally increasing with sediment concentra­ Rho tion. The applied spiked concentration matches the orders of magnitude of environmental total dissolved Sn freshwater concentrations in the Garonne-Gironde fluvial-estuarine system (average 9.79 � 0.97 ng L 1 at La R�eole, Pougnet, 2018), in other river systems (6–10 ng L 1; Weber, 1985; van den Berg et al., 1991), and in the Atlantic Ocean (0.36–3.56 ng L 1; Byrd and Andreae, 1982, 1986). In this study, experimental distributions between the dissolved and particulate phases in the first 4–6 h of contact produced average Rd (or Kd) values ranging from 95 700 to 7 590 000 L kg 1. Noteworthy, Rd values after 48 h of desorption in clean water matrices were higher than those registered during adsorption. This difference suggests that adsorption is partially irreversible within the timeframe of the experi­ ments, i.e., desorption is much slower that adsorption. These trends of adsorption vs desorption have also been observed in similar batch ex­ periments performed with radioactive Fe- and Cr-isotopes (in contrast to intermediate reversibility for other elements like Zn and Cd, especially in seawater; Hatje et al., 2003). In fact, desorption seems to be faster (i) in clean freshwater, independent of the SPM nature and original adsorption matrix, and (ii) when both adsorption and desorption ex­ periments are performed in the same matrices (e.g., desorption of Sn in freshwater is faster than in seawater for particles that have been previ­ ously exposed to freshwater during adsorption). Furthermore, desorp­ tion in the Rh^ one River sediments was slower than for the Garonne River sediments. In any case, these values fall within the range of natural Kd 7

T. Gil-Díaz et al.

Journal of Environmental Radioactivity 213 (2020) 106133

values found in the Garonne River at the La Reole site (average 1 580 000 L kg 1) and in the Gironde Estuary (63 100 to 1 580 000 L kg 1; Pougnet, 2018). These results suggest that field observations and labo­ ratory experiments reflect similar solid/liquid partitioning for Sn, with high Sn particle affinities widely independent from the salinity gradient. The few published studies report the conservative geochemical behav­ iour of Sn along estuarine salinity and turbidity gradients, despite the difficulties in evaluating dissolved inorganic Sn concentrations in the presence of higher-concentrated dissolved organo-tin species (Andreae et al., 1983; Byrd and Andreae, 1986; van den Berg et al. 1991). Some studies report a non-conservative behaviour, proposing dissolution ki­ netics of atmospheric Sn-containing particles and estuarine scavenging as potential explanations to the deviation of the conservative trend (Andreae et al., 1983). The specific interactions of inorganic Sn species with mineral phases are still unknown. Some authors suggest that Sn is capable of replacing Al in octahedral lattice sites (Ringwood et al., 1979). This means that Sn can interact with silicates and clay minerals, quite abundant in the Garonne/Gironde and Rh^ one sediments (Fig. 2, Fig. S1). This statement is also in accordance with studies reporting high contents of inherited Sn in the residual fraction of natural sediments, e.g., ~90% of total Sn for sediments of the Changjiang River Estuary (Duan et al., 2014), ~86% and ~75% for the Gironde Estuary and the River Rh^ one, respectively (Gil-Díaz et al. unpublished). The influence of crystalline Fe mineral carrier phases cannot be discarded, even though crystalline Fe oxides (e. g., goethite) may adsorb Sn(IV) with Kds ranging from 27 000 to 47 000 L kg 1 at circumneutral pH (Ticknor and McMurry, 1996). A lower in­ fluence of amorphous Fe and Mn mineral phases can be suspected from parallel selective extractions performed in aliquots of the same sedi­ ments, showing 10–18% of total inherited Sn associated to the amor­ phous Fe/Mn oxides fraction, compared to ~14% (KP-30, Gironde SPM) ^ne SPM) in the acid-soluble fraction (i.e., including and ~25% (TAS, Rho phyllosilicates and sulphides, amorphous and crystalline Fe/Mn oxides and carbonates carrier phases; Huerta-Diaz and Morse, 1990; Gasparon and Matschullat, 2006). Nevertheless, several mineral phases could explain the observed differences between studied sites during desorp­ tion of spiked Sn, despite the general low desorption (i.e., <15%, Fig. 5). In fact, the high content in swellable clay minerals (e.g., smectites, thus, potential higher exchange with other dissolved elements) and the rela­ tively high amount of amorphous phases (i.e., weaker interactions) in the Garonne/Gironde sediments could enhance desorption of Sn whereas the greater content in illite/muscovite and calcareous minerals ^ne River could favour Sn retention in the in the sediments of the Rho particle phase. Nevertheless, silicate minerals may show different Kds for Sn(IV) species (e.g., smectite-containing solids such as bentonite and argillite range 720–12 000 L kg 1 whereas Na-illite in 0.1 M NaClO4 solution may show Kd values of ~400 000 L kg 1; Kedziorek et al., 2007; Bradbury and Baeyens, 2009). Mineral phases showing high affinity for Sn would contribute first to adsorption processes whereas those showing low sorption affinity are the first to contribute to desorption. All of this suggests that phyllosilicates may be an example of minerals showing high sorption affinity for Sn (i.e., explaining the general/comparable Kds between sites during adsorption), and that exchange from amor­ phous phases is an example of a low affinity sorption process. The results on 113Sn adsorption and desorption kinetics suggest that, in the hypothetical case of a nuclear power plant accidental release to the aquatic environment, most of the radioactive Sn-isotopes (from direct fission or as activation products; Kleykamp, 1985; Le Petit et al., 2014) may be rapidly adsorbed onto SPM and partly exported to the sediment, whatever the hydrological condition and related SPM con­ centrations. In case of any changes in the water composition (e.g., par­ ticles exposed to non-contaminated waters due to tidal influence), based on the experimental results, we can assume that the relatively slow re-equilibration kinetics matching the new conditions will not be ach­ ieved in environmental timescales (e.g., tidal cycles). Accordingly, Sn radioisotope fate will directly depend on sediment dispersion and

storage. This implies that Sn radioisotope effects to biota will be mostly ^ne River, the linked to the bioaccessibility of particulate Sn. In the Rho fate of Sn radioisotopes would mainly depend on the site of emission and the downstream morphology of the river (e.g. storage structures favouring sedimentation) and hydrology (floods may favour transport to the Mediterranean). Radionuclide dispersion between the pro-delta and the adjacent shelf are currently being assessed by numerical modelling approaches (Estournel et al. unpublished). Long Sn half-lives implying long environmental persistence of Sn radioisotopes, together with long particle residence times (1–2 years or more) in the Gironde Estuary and the upstream migration of the MTZ towards the city of Bordeaux during summer must be considered when developing Sn radionuclide disper­ sion and radioprotection scenarios in the Gironde Estuary. 5. Conclusion The present work is the first to address the biogeochemical behaviour of Se and Sn simulating contrasting salinity and turbidity conditions as they may occur in the river-sea continuum of the Garonne-Gironde ^ne River. The experimentally fluvial-estuarine system and the Rho determined solid-liquid partitioning, expressed by Rd values, are a prerequisite for the conceptual development of dispersion scenarios and management strategies anticipating potential accidental radionuclide releases. Different series of sorption experiments with stable and radioactive isotopes suggest that laboratory-Rd values of soluble ele­ ments such as Se may vary with the spiked concentrations, but elements presenting high particulate affinities (e.g. Sn) may show more inde­ pendent, constant solid/liquid partitioning. Further work on Rd values under contrasting environmental conditions (including salinity, turbidity and redox conditions) is needed to develop precise models of radionuclide dispersion in aquatic systems. The observed dominance of the dissolved fraction in Se radionuclide partitioning, independent from salinity and even in highly turbid con­ ditions (>1000 mg L 1) as those in the Gironde Estuary, clearly suggests that Se radionuclide dispersion would follow water transport towards the sea and that interactions with or uptake by living organisms should be considered/expected. In contrast, Sn radionuclide releases would rapidly and partly irreversibly enter the particulate phase in both sys­ tems, independent from their respective sediment loads, and then follow the respective sediment dynamics. Sediment dynamics in the Gironde Estuary imply Sn radionuclide retention within the estuary, potentially during several years, and upstream migration into the reaches of the Bordeaux agglomeration during the dry seasons. Potential expulsion to the coastal ocean would depend on combined effects of seasonal fresh­ ^ne River water discharges and tidal cycles. The scenario for the Rho suggests that Sn radionuclide dispersion will depend on the site of po­ tential release, and implies possible retention of Sn radionuclides in reservoir sediments or discharge-related transport to the Mediterranean coast. Low reactivity and desorption efficiency suggest that transport and persistence of Sn radionuclides along the pro-delta and continental shelf in the Gulf of Lions may involve various sedimentation/erosion cycles. Declaration of competing interest The authors do not declare any conflicts of interest. Acknowledgements This work was supported by French government funds managed by the National Research Agency in the frame of the Investments in the Future program with the reference number ANR-11-RSNR-0002. The authors gratefully acknowledge Franck Gîner from the IRSN in Cadar­ ^ne ache for providing wet sediments from the SORA station in the Rho River, Beate Oetzel, Rosa Danisi and Kirsten Drüppel from AGW-KIT for the XRD analyses and interpretations, Christian Marquardt from INE-KIT 8

T. Gil-Díaz et al.

Journal of Environmental Radioactivity 213 (2020) 106133

for buying the radioactive standard solutions, Alexandra Coynel from TGM-UBx for the granulometry measurements and Markus Lenz from FHNW for quantifying environmental concentrations of dissolved Se in freshwater samples from La R�eole. We also acknowledge the access to the INE-KIT controlled area to perform the radiotracer experiments.

illumination conditions effects by use of reflectance ratios. Appl. Opt. 42 (15), 2623–2634. Doxaran, D., Froidefond, J.M., Castaing, P., Babin, M., 2009. Dynamics of the turbidity maximum zone in a macrotidal estuary (the Gironde, France): observations from field and MODIS satellite data. Estuar. Coast Shelf Sci. 81 (3), 321–332. Duan, L.Q., Song, J.M., Yuan, H.M., Li, X.G., Li, N., Ma, J.K., 2014. Distribution, chemical speciation and source of trace elements in surface sediments of the Changjiang Estuary. Env. Earth Sci. 72 (8), 3193–3204. Duc, M., Lefevre, G., Fedoroff, M., Jeanjean, J., Rouchaud, J.C., Monteil-Rivera, F., Dumonceau, J., Milonjic, S., 2003. Sorption of selenium anionic species on apatites and iron oxides from aqueous solutions. J. Environ. Radioact. 70 (1–2), 61–72. Eichenberger, E., 1986. The interrelation between essentiality and toxicity of metals in the aquatic ecosystem. Met. Ions Biol. Syst. 20, 67–100. Froidefond, J.M., Jegou, A.M., Hermida, J., Lazure, P., Castaing, P., 1998. Variability of the Gironde turbid plume by remote sensing, effects of climatic factors. Oceanol. Acta 21 (2), 191–207. Fujikawa, Y., Fukui, M., 1997. Radionuclide sorption to rocks and minerals: effects of pH and inorganic anions. Part 2. Sorption and speciation of selenium. Radiochim. Acta 76 (3), 163–174. Garisto, N.C., Vance, E.R., Stroes-Gascoyne, S., Johnson, L.H., 1989. Instant-release Fractions for the Assessment of Used Nuclear Fuel Disposal (AECL–9892). Canada. Gasparon, M., Matschullat, J., 2006. Trace metals in antarctic ecosystems: results from the Larsemann hills, East Antarctica. Appl. Geochem. 21 (9), 1593–1612. Ghahremani, D., Mobasherpour, I., Salahi, E., Ebrahimi, M., Manafi, S., Keramatpour, L., 2017. Potential of nano crystalline calcium hydroxyapatite for Tin (II) removal from aqueous solutions: equilibria and kinetic processes. Arabian J. Chem. 10, S461–S471. Gil-Díaz, T., Sch€ afer, J., Keller, V., Eiche, E., Dutruch, L., M€ oßner, C., Lenz, M., Eyrolle, F., 2019. Tellurium and selenium sorption kinetics and solid fractionation under contrasting estuarine salinity and turbidity conditions. Chem. Geol. 532, 119370. Gil-Díaz, T., Sch€ afer, J., Filella, M., Dutruch, L., Bossy, C., 2019. Fractionation of inherited and spiked antimony (Sb) in fluvial/estuarine bulk sediments: unexpected anomalies in parallel selective extraction protocols. Appl. Geochem. 108, 104386. Hatje, V., Payne, T.E., Hill, D.M., McOrist, G., Birch, G.F., Szymczak, R., 2003. Kinetics of trace element uptake and release by particles in estuarine waters: effects of pH, salinity, and particle loading. Environ. Int. 29 (5), 619–629. He, Z.L., Shentu, J., Yang, X.E., 2010. Manganese and selenium. In: Hooda, P.S. (Ed.), Trace Elements in Soils. Wiley, London, pp. 481–496. Huerta-Diaz, M.A., Morse, J.W., 1990. A quantitative method for determination of trace metal concentrations in sedimentary pyrite. Mar. Chem. 29, 119–144. Jordan, N., Foerstendorf, H., Weiß, S., Heim, K., Schild, D., Brendler, V., 2011. Sorption of selenium (VI) onto anatase: macroscopic and microscopic characterization. Geochem. Cosmochim. Acta 75 (6), 1519–1530. J€ org, G., Bühnemann, R., Hollas, S., Kivel, N., Kossert, K., Van Winckel, S., Gostomski, C. L.V., 2010. Preparation of radiochemically pure 79Se and highly precise determination of its half-life. Appl. Radiat. Isot. 68 (12), 2339–2351. Jouanneau, J.M., Latouche, C., 1981. The Gironde estuary. Contrib. Sedimentol. 10. Jouanneau, J.M., Weber, O., Cremer, M., Castaing, P., 1999. Fine-grained sediment budget on the continental margin of the Bay of Biscay. Deep-Sea Res. Part II Top. Stud. Oceanogr. 46 (10), 2205–2220. Kedziorek, M.A., Bourg, A.C., Giffaut, E., 2007. Hydrogeochemistry of Sn (IV) in the context of radioactive waste disposal: solubility and adsorption on MX-80 bentonite and Callovo-Oxfordian argilite. Phys. Chem. Earth, Parts A/B/C 32 (8–14), 568–572. Kharkar, D.P., Turekian, K.K., Bertine, K.K., 1968. Stream supply of dissolved silver, molybdenum, antimony, selenium, chromium, cobalt, rubidium and cesium to the oceans. Geochim. Cosmochim. Ac. 32 (3), 285–298. Kleykamp, H., 1985. The chemical state of the fission products in oxide fuels. J. Nucl. Mater. 131 (2–3), 221–246. Konovalenko, L., Bradshaw, C., Kumblad, L., Kautsky, U., 2014. Radionuclide transfer in marine coastal ecosystems, a modelling study using metabolic processes and site data. J. Environ. Radioact. 133, 48–59. Lefevre, D., Minas, H.J., Minas, M., Robinson, C., Williams, P.L.B., Woodward, E.M.S., 1997. Review of gross community production, primary production, net community production and dark community respiration in the Gulf of Lions. Deep Sea Res. Part II Top. Stud. Oceanogr. 44 (3–4), 801–832. Le Petit, G., Douysset, G., Ducros, G., Gross, P., Achim, P., Monfort, M., Raymond, P., Pontillon, Y., Jutier, C., Blanchard, X., Taffary, T., 2014. Analysis of radionuclide releases from the Fukushima Dai-Ichi nuclear power plant accident Part I. Pure Appl. Geophys. 171 (3–5), 629–644. Loring, D.H., Rantala, R.T.T., 1992. Manual for the geochemical analyses of marine sediments and suspended particulate matter. Earth Sci. Rev. 32 (4), 235–283. Ludwig, W., Dumont, E., Meybeck, M., Heussner, S., 2009. River discharges of water and nutrients to the Mediterranean and Black Sea: major drivers for ecosystem changes during past and future decades? Prog. Oceanogr. 80 (3), 199–217. Marion, C., Dufois, F., Arnaud, M., Vella, C., 2010. In situ record of sedimentary processes near the Rh^ one River mouth during winter events (Gulf of Lions, Mediterranean Sea). Cont. Shelf Res. 30 (9), 1095–1107. Martínez, M., Gim� enez, J., De Pablo, J., Rovira, M., Duro, L., 2006. Sorption of selenium (IV) and selenium (VI) onto magnetite. Appl. Surf. Sci. 252 (10), 3767–3773. Masson, M., Angot, H., Le Bescond, C., Launay, M., Dabrin, A., Mi�ege, C., Le Coz, J., Coquery, M., 2018. Sampling of suspended particulate matter using particle traps in the Rh^ one River: relevance and representativeness for the monitoring of contaminants. Sci. Total Environ. 637, 538–549. McKinley, I.G., Alexander, W.R., 1992. Constraints on the applicability of ‘in-situ distribution coefficient’values. J. Environ. Radioact. 15 (1), 19–34.

Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.jenvrad.2019.106133. References Abe, Y., Iizawa, Y., Terada, Y., Adachi, K., Igarashi, Y., Nakai, I., 2014. Detection of uranium and chemical state analysis of individual radioactive microparticles emitted from the Fukushima nuclear accident using multiple synchrotron radiation X-ray analyses. Anal. Chem. 86 (17), 8521–8525. Aloisi, J.C., Monaco, A., 1975. La s� edimentation infralittorale. Les prodeltas nordm�editerran�eens. Comptes Rendus Acad�emie Science Paris 280, 2833–2836. Amouroux, D., Donard, O.F., 1997. Evasion of selenium to the atmosphere via biomethylation processes in the Gironde estuary, France. Mar. Chem. 58 (1–2), 173–188. Andreae, M.O., Byrd, J.T., Froehlich, P.N., 1983. Arsenic, antimony, germanium, and tin in the Tejo Estuary, Portugal: modeling a polluted estuary. Environ. Sci. Technol. 17 (12), 731–737. Asai, S., Hanzawa, Y., Okumura, K., Shinohara, N., Inagawa, J., Hotoku, S., Suzuki, K., Kaneko, S., 2011. Determination of 79Se and 135Cs in spent nuclear fuel for inventory estimation of high-level radioactive wastes. J. Nucl. Sci. Technol. 48 (5), 851–854. Asai, S., Toshimitsu, M., Hanzawa, Y., Suzuki, H., Shinohara, N., Inagawa, J., Okumura, K., Hotoku, S., Kimura, T., Suzuki, K., Kaneko, S., 2013. Isotope dilution inductively coupled plasma mass spectrometry for determination of 126Sn content in spent nuclear fuel sample. J. Nucl. Sci. Technol. 50 (6), 556–562. Bonhoure, I., Wieland, E., Scheidegger, A.M., Ochs, M., Kunz, D., 2003. EXAFS study of Sn (IV) immobilization by hardened cement paste and calcium silicate hydrates. Environ. Sci. Technol. 37 (10), 2184–2191. Bradbury, M.H., Baeyens, B., 2009. Sorption modelling on illite Part I: titration measurements and the sorption of Ni, Co, Eu and Sn. Geochem. Cosmochim. Acta 73 (4), 990–1003. Broche, P., Devenon, J.L., Forget, P., de Maistre, J.C., Naudin, J.J., Cauwet, G., 1998. Experimental study of the Rhone plume. Part I: physics and dynamics. Oceanol. Acta 21 (6), 725–738. Buesseler, K., Dai, M., Aoyama, M., Benitez-Nelson, C., Charmasson, S., Higley, K., et al., 2017. Fukushima Daiichi–derived radionuclides in the ocean: transport, fate, and impacts. Ann. Rev. Marine Sci. 9, 173–203. Byrd, J.T., Andreae, M.O., 1982. Tin and methyltin species in seawater: concentrations and fluxes. Science 218 (4572), 565–569. Byrd, J.T., Andreae, M.O., 1986. Geochemistry of tin in rivers and estuaries. Geochem. Cosmochim. Acta 50 (5), 835–845. � l’oc�ean des suspensions estuariennes: cas de la Gironde, Castaing, P., 1981. Le transfert a vol. 12. Original typescript. Castaing, P., Allen, G.P., 1981. Mechanisms controlling seaward escape of suspended sediment from the Gironde: a macrotidal estuary in France. Mar. Geol. 40 (1–2), 101–118. Castaing, P., Jouanneau, J.M., 1979. Temps de r�esidence des eaux et des suspensions dans l’estuaire de la Gironde. Journal Recherche Oc� eanographie IV, 41–52. Chang, Y., Zhang, J., Qu, J., Zhang, G., Zhang, A., Zhang, R., 2016. The behavior of dissolved inorganic selenium in the Changjiang Estuary. J. Mar. Syst. 154, 110–121. Ciceri, G., Traversi, A.L., Martinotti, W., Queirazza, G., 1988. Radionuclide partitioning between water and suspended matter: comparison of different methodologies. Stud. Environ. Sci. 34, 353–375. Coppin, F., Chabroullet, C., Martin-Garin, A., 2009. Selenite interactions with some particulate organic and mineral fractions isolated from a natural grassland soil. Eur. J. Soil Sci. 60 (3), 369–376. Coynel, A., Gorse, L., Curti, C., Schafer, J., Grosbois, C., Morelli, G., Ducassou, E., Blanc, G., Maillet, G.M., Mojtahid, M., 2016. Spatial distribution of trace elements in the surface sediments of a major European estuary (Loire Estuary, France): source identification and evaluation of anthropogenic contribution. J. Sea Res. 118, 77–91. Cutter, G.A., 1989. The estuarine behaviour of selenium in San Francisco Bay. Estuar. Coast Shelf Sci. 28 (1), 13–34. Cutter, G.A., Cutter, L.S., 1995. Behavior of dissolved antimony, arsenic, and selenium in the Atlantic Ocean. Mar. Chem. 49 (4), 295–306. Cuvin-Aralar, M.L.A., Furness, R.W., 1991. Mercury and selenium interaction: a review. Ecotoxicol. Environ. Saf. 21 (3), 348–364. Dhillon, K.S., Dhillon, S.K., 1999. Adsorption–desorption reactions of selenium in some soils of India. Geoderma 93 (1–2), 19–31. � DIREN, 2018. National Hydrographic Databank. Minist�ere de l’Ecologie, du � D�eveloppement Durable et de l’Energie. http://www.hydro.eaufrance.fr/. (Accessed 19 June 2018). Doxaran, D., Froidefond, J.M., Castaing, P., 2003. Remote-sensing reflectance of turbid sediment-dominated waters. Reduction of sediment type variations and changing

9

T. Gil-Díaz et al.

Journal of Environmental Radioactivity 213 (2020) 106133 S�eby, F., Potin-Gautier, M., Giffaut, E., Donard, O.F.X., 2001. A critical review of thermodynamic data for inorganic tin species. Geochem. Cosmochim. Acta 65 (18), 3041–3053. Seyler, P., Martin, J.M., 1991. Arsenic and selenium in a pristine river-estuarine system: the Krka (Yugoslavia). Mar. Chem. 34 (1–2), 137–151. Sottolichio, A., Castaing, P., 1999. A synthesis on seasonal dynamics of highlyconcentrated structures in the Gironde estuary. Comptes Rendus Acad. Sci. - Ser. IIA Earth Planet. Sci. 329 (11), 795–800. Tan, J.A., Wang, W.Y., Wang, D.C., Hou, S.F., 1994. Adsorption, volatilization, and speciation of selenium in different types of soils in China. In: Frankenberger, W.T., Benson, S. (Eds.), Selenium in the Environment. Marcel Dekker, New York, pp. 47–68. Thollet, F., Le Bescond, C., Lagouy, M., Gruat, A., Grisot, G., Le Coz, J., Coquery, M., Lepage, H., Gairoard, S., Gattacceca, J.C., Ambrosi, J.-P., Radakovitch, O., 2018. Observatoire des S�ediments du Rh^ one, Irstea. Last accessed on the 04/11/19. https: //dx.doi.org/10.17180/OBS.OSR. Ticknor, K.V., McMurry, J., 1996. A study of selenium and tin sorption on granite and goethite. Radiochim. Acta 73 (3), 149–156. Tolu, J., Thiry, Y., Bueno, M., Jolivet, C., Potin-Gautier, M., Le H�echo, I., 2014. Distribution and speciation of ambient selenium in contrasted soils, from mineral to organic rich. Sci. Total Environ. 479, 93–101. Ulanovsky, A., Copplestone, D., Batlle, J.V.I., 2017. ICRP publication 136: dose coefficients for non-human biota environmentally exposed to radiation. Ann. ICRP 46 (2), 1–136. van den Berg, C.M.G., Khan, S.H., Daly, P.J., Riley, J.P., Turner, D.R., 1991. An electrochemical study of Ni, Sb, Se, Sn, U and V in the estuary of the Tamar. Estuar. Coast Shelf Sci. 33 (3), 309–322. van den Berg, C.M., 1993. Complex formation and the chemistry of selected trace elements in estuaries. Estuaries 16 (3), 512–520. van der Sloot, H.A., Hoede, D., Wijkstra, J., Duinker, J.C., Nolting, R.F., 1985. Anionic species of V, as, Se, Mo, Sb, Te and W in the scheldt and rhine estuaries and the southern Bight (north sea). Estuar. Coast Shelf Sci. 21 (5), 633–651. Weber, G., 1985. The importance of tin in the environment and its determination at trace levels. Die Bedeutung des Zinns in der Umwelt und seine Bestimmung im Spurenbereich. Fresenius’ Z. für Anal. Chem. 321 (3), 217–224. Yao, Q.Z., Zhang, J., Qin, X.G., Xiong, H., Dong, L.X., 2006. The behavior of selenium and arsenic in the Zhujiang (Pearl river) estuary, South China sea. Estuar. Coast Shelf Sci. 67 (1–2), 170–180. Zeng, H., 2009. Selenium as an essential micronutrient: roles in cell cycle and apoptosis. Molecules 14 (3), 1263–1278.

Measures, C.I., Burton, J.D., 1978. Behaviour and speciation of dissolved selenium in estuarine waters. Nature 273 (5660), 293. Measures, C.I., Burton, J.D., 1980. The vertical distribution and oxidation states of dissolved selenium in the northeast Atlantic Ocean and their relationship to biological processes. Earth Planet. Sci. Lett. 46 (3), 385–396. Mitchell, K., Couture, R.M., Johnson, T.M., Mason, P.R., Van Cappellen, P., 2013. Selenium sorption and isotope fractionation: iron (III) oxides versus iron (II) sulfides. Chem. Geol. 342, 21–28. Morewitz, H.A., 1981. Fission product and aerosol behavior following degraded core accidents. Nucl. Technol. 53 (2), 120–134. Nakamaru, Y., Uchida, S., 2008. Distribution coefficients of tin in Japanese agricultural soils and the factors affecting tin sorption behavior. J. Environ. Radioact. 99 (6), 1003–1010. Oda, C., Shibata, M., Ikeda, T., 1999. Experimental Studies for Sorption Behavior of Tin on Bentonite and Rocks, and Diffusion Behavior of Tin in Compacted Bentonite (No. JNC-TN–8400-99-073). Japan Nuclear Cycle Development Inst. Ollivier, P., Hamelin, B., Radakovitch, O., 2010. Seasonal variations of physical and chemical erosion: a three-year survey of the Rhone River (France). Geochem. Cosmochim. Acta 74 (3), 907–927. ~ ez, R., 2005. Modelling the transport of suspended particulate matter by the Rhone Peri� an River plume (France). Implications for pollutant dispersion. Environ. Pollut. 133 (2), 351–364. � proximit� Pont, D., 1997. Les d�ebits solides du Rh^ one a e de son embouchure : donn� ees r� ecentes (1994–1995). Rev. Geogr. Lyon 72, 23–33. Pont, D., Simonnet, J.P., Walter, A.V., 2002. Medium-term changes in suspended sediment delivery to the ocean: consequences of catchment heterogeneity and river management (Rh^ one River, France). Estuar. Coast Shelf Sci. 54 (1), 1–18. Pougnet, F., 2018. Etat de la qualit�e des eaux de l’estuaire de la Gironde: cas du cadmium et des butyl�etains. Doctoral dissertation, Universit� e de Bordeaux. Poulier, G., Launay, M., Le Bescond, C., Thollet, F., Coquery, M., Le Coz, J., 2019. Combining flux monitoring and data reconstruction to establish annual budgets of suspended particulate matter, mercury and PCB in the Rh^ one River from Lake Geneva to the Mediterranean Sea. Sci. Total Environ. 658, 457–473. Ringwood, A., Kesson, S.E., Ware, N.G., Hibberson, W., Major, A., 1979. Immobilisation of high level nuclear reactor wastes in SYNROC. Nature 278 (5701), 219. Rovira, M., Gim� enez, J., Martínez, M., Martínez-Llad� o, X., de Pablo, J., Marti, V., Duro, L., 2008. Sorption of selenium (IV) and selenium (VI) onto natural iron oxides: goethite and hematite. J. Hazard Mater. 150 (2), 279–284. Rüdel, H., 2003. Case study: bioavailability of tin and tin compounds. Ecotoxicol. Environ. Saf. 56 (1), 180–189. Sch€ afer, J., Blanc, G., Lapaquellerie, Y., Maillet, N., Maneux, E., Etcheber, H., 2002. Tenyear observation of the Gironde tributary fluvial system: fluxes of suspended matter, particulate organic carbon and cadmium. Mar. Chem. 79 (3–4), 229–242.

10