Mineral-dependent and calcium-inhibited adsorption of radium-226 onto quartz and kaolinite in a natural sand substrate

1 Introduction

Radium-226 (²²⁶Ra) is a decay product of the uranium-238 (²³⁸U) series and is a radionuclide of environmental concern due to its long half-life (1,600 years), high specific activity (3.7 × 10¹⁰ Bq g^-1), and potential for accumulation in aquatic food chains because of its chemical similarity to Ca²⁺, an essential element for aquatic organisms (Jeffree, 1988; Jeffree and Simpson, 1986; 1984).

Mining activities can mobilize ²²⁶Ra by exposing excavated rock containing ²³⁸U or through milling of ore, which produces tailings containing U-series radionuclides. The leaching of ²²⁶Ra from mine tailings therefore represents an important pathway for the release of ²²⁶Ra to aquatic environments (Chellam and Clifford, 2002; Martin et al., 2003). Radium-226 may also be discharged in treated mining effluents not only from uranium mining and milling operations but from any mining activities involving ore bodies containing ²³⁸U (Proulx et al., 2025). Activity concentrations in mining effluent can reach up to 6,940 mBq L^-1 or 188 pg L^-1 (Proulx et al., 2025).

In natural surface waters, ²²⁶Ra occurs primarily as a divalent cation (²²⁶Ra²⁺) at background activity concentrations ranging between 0.5 and 20 mBq L^-1 (Fesenko et al., 2014; Proulx et al., 2025). Its aqueous speciation influences mobility, solubility, and potential availability to aquatic organisms. While ²²⁶Ra²⁺ may form aqueous complexes with ligands such as nitrate, chloride, and iodate under appropriate chemical conditions (Ames et al., 1983), its behaviour in freshwater is also strongly influence by co-precipitation with sulfate and carbonate minerals, particularly barite (BaSO₄(s)) owing to the close similarity in ionic radius and charge between ²²⁶Ra²⁺ and Ba²⁺(Langmuir and Riese, 1985; Zhang et al., 2014).

In addition to aqueous speciation and mineral precipitation, sorption is a dominant process controlling the environmental fate of dissolved ²²⁶Ra (Langmuir and Riese, 1985). Sorption refers to the interaction of an element with the surface of a solid, including suspended particles and bottom sediments (Beneš et al., 1984; Beneš and Strejc, 1986; Landa and Reid, 1983), and encompasses both surface adsorption and absorption into porous solids, while desorption refers to the reverse process (Riese, 1982). Radium-226 adsorption over a range of physical and chemical parameters is relatively well studied for common rock-forming minerals including quartz (Beneš et al., 1984), muscovite and albite (Beneš and Strejc, 1986) and clay minerals such as kaolinite, montmorillonite, bentonite and smectite (Beneš et al., 1985; Reinoso-Maset and Ly, 2016; Robin et al., 2017; Tachi et al., 2001). Laboratory experiments have demonstrated that ²²⁶Ra adsorption to quartz and kaolinite is very rapid, with the majority of the Ra²⁺ being adsorbed within 10 min (Beneš et al., 1984). Adsorption affinity differs substantially among mineral groups and is strongly influenced by mineral composition and cation-exchange capacity (CEC), with clay minerals generally exhibiting a greater affinity for ²²⁶Ra than quartz and other silicates (Ames et al., 1983; Beneš and Strejc, 1986; Robin et al., 2017). Because of its large ionic radius, the interaction of ²²⁶Ra with mineral surfaces is expected to be increasingly electrostatic or ionic relative to lighter alkaline earth metals (Langmuir and Riese, 1985). This conceptual understanding provides the mechanistic basis for representing ²²⁶Ra partitioning between aqueous and solid phases using cation-exchange and surface complexation models (Riese, 1982).

Beyond quartz and aluminosilicate minerals, Fe-(hydr)oxide phases including both poorly crystalline (amorphous) and crystalline minerals are also important sorbents for dissolved ²²⁶Ra (Chellam and Clifford, 2002; Nirdosh et al., 1990; Thiry and Van Hees, 2008). Laboratory experiments have demonstrated that ²²⁶Ra adsorption to Fe-(hydr)oxide is rapid, with most Ra²⁺ removed from solution within 10 min (Beneš et al., 1984). Iron (hydr)oxides commonly occur in freshwater environments as surface coatings on mineral particles and can play an important role in controlling ²²⁶Ra mobility even at low concentrations through electrostatic outer-sphere complexation with surface hydroxyl groups (Beneš et al., 1984; Chautard et al., 2020). Experimental studies of ²²⁶Ra sorption to Fe (hydr)oxides have focused primarily on minerals such as goethite and ferrihydrite (Ames et al., 1983; Sajih et al., 2014), as well as Fe-(hydr)oxide (Beneš et al., 1984). Consistent with these findings, Liu and Hendry (2011) reported that ²²⁶Ra associated with precipitates formed during neutralization of uranium mill raffinate was dominated by sorption to amorphous and crystalline Fe-(hydr)oxide phases.

The extent of ²²⁶Ra sorption depends not only on mineralogy but also on surrounding water chemistry. Radium-226 adsorption to quartz, kaolinite, Fe- and Mn-(hyr)oxides increases with increasing pH (Nirdosh et al., 1990; Robin et al., 2017; Thiry and Van Hees, 2008) and with decreasing liquid-to-solid ratio (Beneš et al., 1984). The presence of other cations (e.g., Ca²⁺, Ba²⁺, Mn²⁺) can also influence Ra²⁺ adsorption through competitive ion exchange (Brown et al., 2019; Robin et al., 2017). However, the immobilisation of ²²⁶Ra through, for example, sorption onto Fe- and Mn-(hydr)oxides and its co-precipitation with BaSO₄ is potentially reversible under changing geochemical conditions. Beneš et al. (1984) demonstrated that ²²⁶Ra adsorbed onto quartz via ion exchange in the electrical double layer can desorb as pH decreases. Similarly, adsorption of ²²⁶Ra is reversible from ferrihydrite and goethite (Sajih et al., 2014), and ²²⁶Ra can be quantitatively desorbed from kaolinite (Beneš et al., 1985) and quartz (Beneš et al., 1984), although desorption is more limited from minerals such as montmorillonite (Beneš et al., 1985) and certain hydrated metal oxides (Nirdosh et al., 1990).

Although numerous laboratory studies have quantified ²²⁶Ra sorption to individual mineral phases, comparatively few have examined the combined influence of mineralogy, Ca concentration, and organic matter under condition representative of freshwater toxicity testing or sandy aquatic sediments. Consequently, the relative important of quartz, clay minerals, competing Ca, and organic matter in controlling dissolved ²²⁶Ra remains poorly constrained. Proulx (2026) reported a 90% reduction in ²²⁶Ra activity concentrations during laboratory exposure experiments with freshwater fingernail clams, Sphaerium striatinum, with test solutions renewed every 48 h, indicating that this loss occurred within that interval. In these experiments, clams burrowed within a sand substrate and were fed algal cells, suggesting that the observed reduction in dissolved ²²⁶Ra may have results from sorption of ²²⁶Ra²⁺ onto either mineral phases present in the sand substrate or onto organic matter. Because the experiments were conducted in reconstituted soft water containing calcium, competitive interactions between Ca²⁺ and ²²⁶Ra²⁺ may also have influence adsorption processes. To investigate this hypothesis, the mineralogical composition of the sand substrate was first characterized to identify phases most likely to act as sorbents for ²²⁶Ra. This characterization guided the selection of quartz and kaolinite as representative mineral phases for further adsorption experiments.

The objectives of this study were to identify the mechanisms contributing to the loss of dissolved ²²⁶Ra observed during the fingernail clam exposure experiments. Specifically, the objectives were to: (1) evaluate the extent to which the sand substrate alone can remove dissolved ²²⁶Ra from solution in the absence of fingernail clams; (2) assess the influence of organic matter, represented by algal cells, on the removal of dissolved ²²⁶Ra from solution; (3) determine the effect of varying Ca²⁺ concentrations on ²²⁶Ra adsorption under conditions where competitive ion-exchange is expected to occur; and (4) compare the relative affinity of ²²⁶Ra for key mineral components of the substrate through batch sorption experiments using quartz and kaolinite. Understanding and predicting the mobility of trace radionuclides such as ²²⁶Ra is important for assessing the potential environmental impacts of uranium mine tailings and effluent on surrounding aquatic systems, as well as for broader applications such as water resource protection.

2 Materials and methods

2.1 Experimental solutions

Base stock solutions were prepared with reconstructed soft water (RSW) following the methodology of Clesceri (1998). The Ca content of the RSW solution (i.e., 7 mg L^-1) was modified by removing or adding calcium sulfate dihydrate to obtain three other base stock solutions with low (0.7 mg L^-1), moderate (35 mg L^-1) and high (70 mg L^-1) Ca concentrations (Supplementary Table S1). The target calcium concentrations were selected based on levels observed downstream of uranium mines and mills (Goulet et al., 2015). To confirm target Ca concentrations in the base stock solutions, along with concentrations of other major ions, a single water sample was collected. Samples were filtered using a 0.45 µm mesh syringe filters (Acrodisc HT Tuffryn Membrane, PALL New York) and sterile 20 mL luer-lock tip polyethylene-polypropylene syringe (Fisher Scientific), collected into pre-labelled polypropylene centrifuge tubes (15mL, Fisherbrand™), and preserved by acidifying them to 1% HNO₃ (trace metal grade). Dissolved Ca, Fe, Mg, Mn, K, and Na were analyzed at the CanmetMINING Analytical Laboratory (Ottawa, ON, Canada) using inductively coupled plasma optical emission spectrometry (ICP-OES, Synchronous Vertical Dual View, Agilent Technologies). Concentration of chloride (Cl) and sulfate anions was measured using ion chromatography (ICS, Dionex Inuvion IC System, ThermoFisher). Dissolved organic carbon (DOC) and inorganic carbon (DIC) content were measured using a total organic carbon analyzer (TOC-L, Shimadzu). Quality assurance/quality control by the CanmetMINING Analytical Laboratory included the collection of duplicates, spikes, and blanks. The Ca content of the stock solutions was confirmed to be 0.9 mg L^-1 (low Ca), 35 mg L^-1 (moderate Ca) and 70 mg L^-1 (high Ca).

The stock solutions were dosed with ²²⁶Ra to achieve nominal concentrations of 1, 5 and 50 Bq L^-1. Calibrated pipettes (Sartorious micropipetter) were used to dose the stock solutions with the appropriate aliquot from a 5 mL stock solution of radium chloride (RaCl₂) with an activity of 8 kBq (Stuart Hunt and Associates Ltd., Edmonton, AB, Canada). Water samples were collected from the stock solutions to confirm nominal concentrations of dissolved ²²⁶Ra. Samples were filtered using a 0.45 µm mesh syringe filters (Acrodisc HT Tuffryn Membrane, PALL New York) and sterile 20 mL luer-lock tip polyethylene-polypropylene syringe (Fisher Scientific), collected into pre-labelled polypropylene centrifuge tubes (15mL, Fisherbrand™), and preserved by acidifying them to 1% HNO₃ (trace metal grade). Sample masses were recorded using an electronic balance (AUW120, Shimadzu Corporation, Kyoto, Japan) with an accuracy of ±0.006 g to allow ²²⁶Ra activity to be expressed on a mass-normalized basis (Bq g^-1). Sample vials were sealed with parafilm and refrigerated until shipment to the Canadian Nuclear Laboratories (CNL, Chalk River, ON, Canada) for analysis of ²²⁶Ra by alpha spectrometry (detection limit = 0.0005 Bq L^-1) using the method outlined in Guérin et al. (2023). Activity concentrations were reported by CNL in millibecquerel per Gram of sample (mBq g^-1) and reported herein in becquerels per litre of water (Bq L^-1). The minimum detectable activity (MDA) ranged from 0.0002 to 0.0005 Bq, which was several orders of magnitude below measured concentrations.

2.2 Inorganic substrates

The inorganic substrate chosen for the adsorption experiments consisted of fine aquarium sand (CaribSea Inc. Super Naturals Premium Aquarium Substrate, color: Marine Sand/Moonlight Sand). This sand substrate was used previously in ²²⁶Ra toxicity experiments involving fingernail clams (Proulx, 2026). The sand substrate was characterized to determine its mineralogical composition and to identify potential adsorption sites for ²²⁶Ra. The approximate mineralogical composition and grain textures of an untreated sand sample were examined using back-scattered electron (BSE) imaging and energy-dispersive X-ray (EDX) spectroscopy on a TESCAN Mira3 field-emission gun scanning electron microscope (FEG-SEM) at the Geological Survey of Canada (Ottawa, Ontario). The data were collected using a 10 kV (BSE images) or 20 kV (EDX spectra) accelerating voltage. The substrates chosen for the batch sorption experiments consisted of natural kaolinite (Al₂O₃ 2SiO₂ 2H₂O, Sigma-Aldrich, Product Number 03584) and trace metal grade white quartz (≥99.995% SiO₂, Sigma-Aldrich, Product Number 204358). The particle size distribution of the substrates was estimated from optical microscope images by calculating the equivalent circular diameter (ECD) of algorithmically segmented particles from three areas of each subsample dispersed on a high-contrast background.

2.3 Geochemical modelling

2.3.1 Radium-226 speciation

Aqueous speciation was modelled for each stock solution using available thermodynamic data for dissolved ²²⁶Ra species (Table 1). Speciation of ²²⁶Ra was performed using the PHREEQC Interactive software (version 3.7.3) with the WATEQ4F thermodynamic database (Ball and Nordstrom, 1991). PHREEQC is a widely used geochemical modelling program for calculating aqueous speciation, reaction-path, and transport (Parkhurst and Appelo, 1999). While PHREEQC was used to evaluate inorganic aqueous speciation, the Windemere Humic Acid Model (WHAM version 7.0, Tipping (1994)) was additionally applied to assess potential complexation of ²²⁶Ra with dissolved organic matter (i.e., humic and fulvic acids ratio of 3:1). Stability constants provided in Table 1 were integrated into WHAM’s default thermodynamic database. Because stability constants for ²⁶Ra complexes with fulvic and humic substances are not available in the WHAM database, stability constants for barium (Ba²⁺) were used as a proxy. Stability constants were assumed as log K of 1.3 and 0.97 for humic and fulvic acids, respectively (Tipping, 1994). This substitution is commonly applied because Ba²⁺ and ²²⁶Ra²⁺ are both alkaline earth metals exhibiting similar chemical behaviour, occurring predominantly as divalent cations in aqueous systems, having comparable ionic radii (Ba²⁺ 142 p.m.; Ra²⁺ 148 p.m.), and similar electronegativities (Brown et al., 2022).

TABLE 1. Thermodynamic data used in speciation modelling.

2.3.2 Cation exchange and surface complexation models

Interactions between ²²⁶Ra and the mineral phases were modelled using both: (1) a cation exchange model for kaolinite and (2) a surface complexation model (SCM) for quartz, to account for different mechanisms by which ²⁶Ra may partition to mineral surfaces. Cation exchange modelling was applied to kaolinite using the EXCHANGE formalism in PHREEQC, which represents reversible electrostatic exchange of aqueous cations with negatively charged surface sites. This approach is appropriate for clay minerals such as kaolinite due to their permanent structural charge arising from isomorphic substitution within the crystal lattice, which gives rise to measurable CEC (Kumm and Gier, 2024). Exchange reactions were modelled under equilibrium conditions, allowing ²²⁶Ra²⁺ and competing cations (i.e., Ca, Mg, Na, and K) to partition between solution and exchange sites. Because an exchange selectivity constant for ²²⁶Ra is not widely available in standard thermodynamic databases, including WATEQ4F, the exchange reaction for ²²⁶Ra²⁺ was assigned the same equilibrium constant as Ba²⁺.

B a 2 + + 2 X − = B a X 2 , ⁡ log ⁡ K = 0.91

R a 2 + + 2 X − = R a X 2 , ⁡ log ⁡ K = 0.91

The total number of exchange sites (mol) was calculated from the reported CEC and the experimental solid mass according to Equation 1:where CEC is expressed in meq 100 g^-1 and m _*solid* is the mass of solid (g). The factor of 100 converts the CEC to the mass of solid used, and division by 1,000 converts milliequivalents to moles of charge.

n e x c h a n g e = C E C * m s o l i d 100 * 1 1000 (1)

Surface complexation modelling was also performed in PHREEQC using the SURFACE formalism to represent sorption of ²²⁶Ra to pH-dependent mineral surface functional groups. Surface complexation models describe sorption as the formation of chemical complexes between aqueous species and reactive surface groups, which are governed by acid–base and metal-binding equilibria (Dzombak and Morel, 1990; Stumm and Morgan, 1996). Surface protonation and deprotonation reactions are represented as: where ≡ S denotes a generic surface functional group involved in reactions between aqueous species and mineral surface (e.g., Kitamura et al., 1999; Nair et al., 2014; Rakshit et al., 2015; Szymanek et al., 2021). For quartz, these groups correspond primarily to silanol (≡ Si-OH), while for kaolinite they include aluminol (≡ Al-OH) and silanol sites (Angove et al., 1997). Surface complexation was simulated using a diffuse double layer (DDL) SCM. The approach was selected to maintain consistency with the literature-derived surface complexation constants, which were developed using the same DDL electrostatic formulation. Surface complexation constants were defined following literature parameterizations (Supplementary Table S2, Supplementary Material). Because surface complexation constants for ²²⁶Ra are not widely available in thermodynamic databases, constants for other alkaline earth metals (e.g., Ca²⁺, Sr²⁺, Ba²⁺) were used as proxies, reflecting their similar geochemical behaviour. This approach is consistent with the use of alkaline earth analogues to represent Ra²⁺ in geochemical modelling due to similarities in ionic radius, charge, and aqueous speciation (Serra-Ventura et al., 2025).

≡ S O H + H + = ≡ S O H 2 +

≡ S O H = ≡ S O − + H +

For the quartz substrate, specific surface area (SSA, m² g^-1) was estimated from measured ECDs, assuming spherical particles following standard geometric relationships between particle size and surface area (Stumm and Morgan, 1996), per Equation 2:where ρ is particle density (kg m^-3), obtained from the literature (Rühlmann et al., 2006), and d is particle diameter (m) derived from ECDs. This calculation assumes smooth, non-porous spherical particles and therefore represents an estimate of external surface area only. The number of surface sites was then calculated using Equation 3:where D are surface site densities, assigned based on literature values for mineral surface functional groups, and N _*A* is Avogadro’s number.

S S A = 6 ρ d (2)

n s i t e s = S S A * D * 10 18 N A (3)

For both exchange and surface complexation simulations, mineral phases were first equilibrated with the experimental solution composition in the absence of ²²⁶Ra to establish initial exchange or surface site occupancy. Radium-226 was then introduced at the experimental concentration of 50 Bq L^-1, and equilibrium partitioning between solution and solid phases was calculated. This sequential approach ensured mass-balanced simulation of ²²⁶Ra uptake.

2.4 Adsorption experiments

Adsorption test methods, modified from Nautilus Environmental (2021), aimed to quantify effects imparted by Ca concentrations, the sand substrate, and the addition of organic matter on measured dissolved ²²⁶Ra activity concentrations over 48 h. Four experiments were conducted, one for each of the Ca concentrations (0.9, 7, 35, 70 mg L^-1), each with the three ²²⁶Ra activities (1, 5 and 50 Bq L^-1), for a total of 12 treatment groups (i.e., ²²⁶Ra x Ca combinations). The design used a tiered approach (see Supplementary Figure S1) to assess the influence of the various components; solution only, solution + sand, and solution + sand + organic matter. Each series had 3 replicates, for a total of 108 test vessels across the 4 experiments. Experimental conditions, including the 48-h test duration, sand-to-water ratio, substrate, organic matter amendment, and test vessel configuration, were selected to replicate those used in the previous ²²⁶Ra toxicity experiments with fingernail clams (Proulx, 2026), thereby allowing the mechanisms responsible for the observed loss of dissolved ²²⁶Ra to be investigate under comparable exposure conditions. Sand substrates (described above) were rinsed three times with ultrapure water (18 MΩ cm) prior to the tests. The organic matter consisted of an algae mixture of 1 mL Instant Algae Shellfish Diet 1800 and 100 μL instant Algae Nano Diet 3,600 in 9 mL RSW, modified from (Salerno et al., 2018). Test vessels consisted of 800 mL plastic Nalgene beakers each containing 300 mL of solution. When added as a component, beakers also contained 150 g of sand and 300 μL of organic matter. Beakers were wrapped in black plastic to minimize light penetration and were loosely covered with parafilm to reduce evaporation.

In situ water quality parameters (water temperature (°C), dissolved oxygen (DO, %), pH, and specific conductivity (µS cm^-1)) were monitored daily using Hach meters (Hach HQ40 d pH/D.O., and HQ14 d conductivity) for a subsample (∼20 mL) which was collected from a randomly selected beaker, for each ²²⁶Ra x Ca treatment. Water subsamples were also collected from the water column of 1 of the 3 replicates, chosen at random, from each of the 12 treatments at hours 1, 3, 6, 9, 12, 24, and 48. Dissolved ²²⁶Ra activity concentrations were measured, following the same procedures as described above (see Section 2.1). Measurements of ²²⁶Ra in stock solution represented concentrations at time 0 h.

Given that dissolved ²²⁶Ra concentrations are expected to decline over the 48 h period, exposure concentrations were summarized using a time-weighted average (TWA). The TWA accounts for the variation in instantaneous concentration over time and provides a more representative estimate of exposure than nominal or single-time-point measurements when concentrations change during the test period (Belgers et al., 2011). The TWA was calculated using the trapezoidal integration method, which assumes a linear change in concentration between successive sampling times. For each experimental treatment, the area under the concentration-time curve was calculated and divided by the total elapsed time per Equation 4:where C _*i* and C _*i+1* are the measured dissolved ²²⁶Ra concentrations at sampling times t _*i* and t _*i+1* respectively, and t _*n* -t _*0* represents the total exposure duration of 48 h (Belgers et al., 2011). This approach integrates concentration over time and normalizes by the exposure period, yielding an average concentration that reflects temporal changes in dissolved ²²⁶Ra. Time-weighted averages (TWA) were calculated separately for each combination of experimental replicate, treatment concentration, and experimental condition.

T W A = ∑ i = 0 n − 1 C i + C i + 1 / 2 t i + 1 − t i t n − t 0 (4)

2.5 Batch sorption experiment

Batch sorption experiments were conducted to determine whether quartz and kaolinite provide binding sites for ²²⁶Ra. A substrates-to-solution ratio of 47 g per 94 mL in 250 mL plastic Nalgene beakers was selected to match the proportions used in the adsorption experiments and the proportions in the previously conducted toxicity tests using S. striatinum (Proulx, 2026). This ratio was intended to reproduce the laboratory exposure conditions rather than to simulate a field-equivalent sediment-to-water ratio in natural environments. Although mixing is commonly used in sorption experiments to ensure transport of ions to adsorption sites is not rate-limiting, the water column was intentionally left unmixed in the present study to replicate the undisturbed conditions used in the adsorption and toxicity experiments. Batch sorption experiments were conducted in triplicates for each of three initial ²²⁶Ra activity concentrations (1, 5 and 50 Bq L^-1 prepared in standard RSW). Subsamples were carefully collected from the water column of each replicate using a sterile 20 mL luer-lock tip polyethylene-polypropylene syringe (Fisher Scientific) to measure dissolved ²²⁶Ra (as described above) after 6, 12, 24 and 48 h.

2.6 Adsorption parameters and kinetic model

The fraction of solute adsorbed (fsorbed), or percent of dissolved ²²⁶Ra removed from solution, was calculated using Equation 5 (Nair et al., 2014; Raghav and Kumar, 2018):were C _*0* is the initial concentration at t = 0 and C _*t* is the measured concentration at time t. The activity concentration associated with the solid phase at equilibrium (C _*s*), which is equivalent to the equilibrium sorbed concentration (q _*e*) used in adsorption isotherm analyses, was calculated by mass balance according to Equation 6 (e.g., Abin-Bazaine et al., 2022):where C _*e* is the equilibrium concentration at t = 48, V is the solution volume (L), and m is the dry mass of sorbent (kg). The partitioning of ²²⁶Ra between the aqueous phase and solid surfaces was quantified, under test conditions, by calculating distribution coefficients (K _*d*). The distribution coefficient for a radionuclide is defined as the ratio of the activity concentration associated with the solid phase at equilibrium (C_s, Bq·kg^-1 dry mass) to the dissolved equilibrium activity concentration in solution (C _*e*, Bq·L^-1), as shown in Equation 7 (e.g., Goulet et al., 2022).

f s o r b e d = C 0 − C t C 0 (5)

C s = C 0 − C e * V m (6)

K d = C s C e (7)

Higher K _*d* value indicates greater partitioning to the solid phase, whereas lower K _*d* values indicate weaker adsorption and greater mobility in the aqueous phase. Distribution coefficients were calculated using ²²⁶Ra concentrations measured at the end of the adsorption experiments (48 h), corresponding to equilibrium, or near-equilibrium, conditions.

Finally, time-dependent changes in dissolved ²²⁶Ra concentrations (Equation 8) were described using first-order kinetic models in which concentration-time data were fit using an exponential approach to equilibrium model of the form per Equation 5 (e.g., Kajjumba et al., 2019; Liu and Shen, 2008):were k is the first-order rate constant. Rate constants (k) and associated standard errors (SE) were obtained by nonlinear least-squares regression. Large k values indicate faster removal of ²²⁶Ra from the water column, and faster equilibrium.

C t = C e + C 0 − C e e − k t (8)

2.7 Statistical analyses

The response variable used to assess the influence of sand and organic matter on ²²⁶Ra adsorption was the fraction ²²⁶Ra adsorbed, calculated for all sampling time points in both the adsorption experiments and the batch sorption experiments. To evaluate whether the loss of ²²⁶Ra from solution differed between Series A and B (i.e., the influence of sand) or between Series B and C (i.e., the influence of organic matter) in the absorption experiments, percent removal was compared using paired Wilcoxon signed-rank tests across matched experimental conditions (²²⁶Ra concentration, Ca concentration, and time). Agreement between series were further examined using Bland–Altman analysis, which assesses agreement between paired measurements (Bland and Atlman, 1986), with bias calculated as the mean paired difference and limits of agreement defined as bias ±1.96 standard deviations (SD).

The response variables used to assess the influence of Ca concentration on ²²⁶Ra adsorption were the sorbed activity (C _*s*), distribution coefficients (K _*d*), and first-order rate constant (k). Rate constants (k) were derived from first-order kinetic model fits, whereas sorbed activity (C _*s*) and distribution coefficients (K _*d*) were calculated from measured aqueous concentrations at 48 h. Statistical analyses of C _*s*, K _*d*, and k were conducted using data from treatments containing a solid phase only. Response variables were analysed using linear models including Ca concentration and nominal ²²⁶Ra activity as fixed factors. Given the observed non-linear response of these parameters to Ca, Ca was modelled using a second-order polynomial term. Interaction terms between Ca and ²²⁶Ra activity were included to assess whether the effect of Ca varied with ²²⁶Ra activity. Model significance was evaluated using analysis of variance (ANOVA).

The response variable used to assess the influence of the different mineral substrates (i.e., kaolinite and quartz) on ²²⁶Ra adsorption was the percent ²²⁶Ra removed from solution, calculated for all sampling time points in the batch sorption experiments. Differences in percent removal between mineral substrates were evaluated using a two-way ANOVA with substrate and time as fixed factors. Estimated marginal means were used to quantify differences between substrates and to evaluate contrasts at each sampling time. Finally, to assess whether sorption had stabilized by the end of the batch sorption experiments, paired sorbed ²²⁶Ra activity measured at 24 and 48 h was compared separately for quartz and kaolinite using Wilcoxon signed-rank tests. All statistical analyses were performed using the R software environment and associated base packages (R 4.6.1, R Core Team, 2026). For all statistical analyses, the level of statistical significance (alpha) was set at 0.05.

3 Results

3.1 Stock solution chemistry

Measured ²²⁶Ra activity concentrations for the stock solutions are outlined in Table 2. Measured dissolved ²²⁶Ra activities ranged from 0.68 to 1.16 Bq L^-1 (mean ± SD = 0.93 ± 0.2), 3.8–5.8 Bq L^-1 (4.8 ± 0.7), and 40–50 Bq L^-1 (46 ± 4.3) for the nominal 1, 5 and 50 Bq L^-1 treatments, respectively. Measured activities in the moderate Ca treatment were consistently lower than the nominal targets and the corresponding activities in the other Ca treatments. Analytical uncertainties were generally within ± 4%–15% depending on activity level. Major ion concentrations for stock solutions are outlined in Supplementary Table S3 (Supplementary Material). Measured Ca was 0.9 mg L^-1, 7.1 mg L^-1, 35.6 mg L^-1, and 72.3 mg L^-1in the low, RSW, moderate and high Ca treatments, respectively. Other water chemistry parameters were consistent among treatments, with mean ± SD of measured dissolved water chemistry variables across all 12 stock solutions (one sample per stock solution) was as follows: Cl 2.0 ± 0 mg L^-1, Fe 0.15 ± 0.1 mg L^-1,^, K 1.3 ± 0.1 mg L^-1, Mg 6.4 ± 0.1 mg L^-1, Na 18 ± 0.1 mg L^-1, Mn 0.03 ± 0.005 mg L^-1, SO₄ 34 ± 0.1 mg L^-1, DOC 19 mg L^-1 (RSW only) and DIC 8.6 ± 1.6 mg L^-1.

TABLE 2. Measured activity of ²²⁶Ra in stock solutions with varying Ca concentrations.

3.2 Radium-226 speciation

PHREEQC modelling indicated that ²²⁶Ra speciation was consistent across the range of tested ²²⁶Ra x Ca concentrations (Figure 1), with approximately 70% of total dissolved ²²⁶Ra occurring as free ²²⁶Ra²⁺ and the remaining fraction occurring as RaHCO₃ ⁺ (∼20%), RaSO₄ (∼6%) Move RaCl⁺ with other species in the next sentence and no need for less than 1%. Complexes with carbonate (RaCO₃) or hydroxide (RaOH⁺) were not predicted under the experimental conditions. The relative distribution among aqueous species remained unchanged across the range of nominal ²²⁶Ra activities (1–50 Bq L^-1) and calcium concentrations (0.9–70 mg L^-1), indicating that variations in experimental conditions did not alter aqueous ²²⁶Ra speciation (Figure 1). WHAM modelling indicated negligible (<0.001%) interaction between ²²⁶Ra and fulvic or humic acids under the experimental water chemistry conditions.

3.3 Sand characterization

The sand substrate was well sorted, with a mean grain size of 219 μm, and the dominant mineral was confirmed to be quartz (SiO2). Textural features (i.e., microfracture and pitting, Figure 2) were typically filled with a clay mineral. The EDX analysis of the clay-rich zones confirmed the presence of aluminosilicate with a high aluminum (Al) and low potassium (K) and magnesium (Mg) contents, consistent with kaolinite (Al₂Si₂O₅(OH)₄). Other minor impurities were also visible (e.g., ilmenite, zircon, and monazite). There was no visual or geochemical evidence of significant Fe or Mn (hydr)oxides (Supplementary Table S4, Supplementary Material). The geochemical results confirmed the presence of Fe (0.034%) in the sand; however, it appeared to be associated with ilmenite (FeTiO3; Ti 0.037%), large grains of which were identified by EDX, and not Fe coatings. The Mn concentration (<0.010%) was low compared to the Al concentration (0.124%) and may be associated with other minor unidentified clays in the microfractures.

3.4 Geochemical models

The mineral properties used in the exchange and surface complexation models, including measured particle sizes, literature-derived parameters (e.g., CEC, particle density, site density), and calculated values (e.g., specific surface area and total site concentrations), are provided in Section S1 of the Supplementary Material. Both the kaolinite and quartz substrates exhibited relatively uniform particle sizes, with mean ECDs of 14 ± 7 µm (n = 300) for kaolinite and 188 ± 40 µm (n = 146) for quartz.

Exchange modelling predicted strong partitioning of ²²⁶Ra to kaolinite, with approximately 95% of the added ²²⁶Ra associated with exchange sites and only 5% remaining in solution under the simulated conditions. Although ²²⁶Ra exhibited a strong affinity for kaolinite relative to its initial concentration, exchange-site occupancy remained dominated by Ca²⁺ and Mg²⁺ because of their substantially greater abundance.

The SCM using the DDL approach predicted substantially weaker sorption of ²²⁶Ra to quartz. Approximately 27% of the added ²²⁶Ra was associated with the quartz surface, while approximately 72% remained in solution. Quartz surface speciation was likewise dominated by Ca-bearing surface complexes and unoccupied surface sites, with ²²⁶Ra occupying only a trace fraction of the available surface sites.

3.5 Tiered adsorption experiments

3.5.1 Supporting water quality

During the adsorption experiments, water quality conditions remained stable across treatments (Supplementary Table S6, Supplementary Material). Temperature was between 21 °C and 22 °C, dissolved oxygen remained high (91%–96% saturation), and pH was circumneutral in all treatments (generally 7.0–7.3), with a slight decrease observed in the low-calcium treatment (6.8–7.2). Specific conductivity reflected the intended calcium gradient, increasing from the low Ca and RSW treatments (∼128–182 μS cm^-1), to the moderate Ca (∼309–337 μS cm^-1) and high Ca (∼486–523 μS cm^-1) treatments, with only minor fluctuations within each treatment.

3.5.2 Stability of dissolved ²²⁶Ra

Stability of ²²⁶Ra in the solution-only treatments remained stable over the 48-h exposure period across all ²²⁶Ra and Ca combinations (Figure 3). Measured activity concentrations showed no decline, indicating negligible loss of ²²⁶Ra from solution in the absence of solid substrates or organic matter. Across the three nominal ²²⁶Ra treatments (1, 5, 50 Bq L^-1), observed concentrations ranged from 0.75 to 0.86 Bq L^-1, 4.1–4.8 Bq L^-1, and 40–50 Bq L^-1in RSW (Ca = 7 mg L^-1). Similar results were observed for the calcium modified treatments, with variability within each treatment generally ≤10%, which was within analytical uncertainty (±4%–15%). The TWA activity concentrations for solution only treatments were estimated to range from 0.8 to 1.2, 3.7–5.5, and 33–52 Bq L^-1 for the 1, 5, and 50 Bq L^-1 nominal treatments, respectively (Figure 4). These values reflect the biologically relevant exposure, by integrating temporal changes in ²²⁶Ra over the 48 h.

3.5.3 Effect of mineral substrate on ²²⁶Ra activity concentrations

In the low Ca (0.9 mg L^-1) and RSW (7 mg L^-1) treatments the addition of natural sand resulted in rapid removal of dissolved ²²⁶Ra from solution, with 46% and 25% of dissolved ²²⁶Ra removed within the first hour, increasing to 49% and 43% by 9 h, respectively. Thereafter, removal increased more gradually toward apparent equilibrium. By 48 h, total ²²⁶Ra removal reached 83% in the low Ca (0.9 mg L^-1) treatment and 75% in the RSW (7 mg L^-1) treatment. A comparable pattern was observed under moderate and high Ca concentrations (Figure 3). In the moderate Ca treatment (35 mg L^-1), 29% of dissolved ²²⁶Ra was removed within the first hour, increasing to 61% by 48 h. In the high Ca treatment (70 mg L^-1), 9% of dissolved ²²⁶Ra was removed within the first hour, increasing to 58% by 48 h. The TWA activity concentrations for the sand only treatment were estimated to range from 0.4 to 0.7, 1.3–2.8, and 13–41 Bq L^-1 for the 1, 5, and 50 Bq L^-1 nominal treatments, respectively (Figure 4).

The addition of sand resulted in a rapid and significant reduction in dissolved ²²⁶Ra concentrations relative to solution-only controls (Wilcoxon signed-rank test, p < 0.001; Figure 3). Relative to solution-only controls, the addition of sand increased ²²⁶Ra removal from solution by a median of 38.4% across matched experimental conditions. This effect was significant overall and within all ²²⁶Ra and Ca treatments tested, with median paired differences ranging from 20.7% to 46.2% ²²⁶Ra concentrations and from 25.1% to 49.1% among Ca concentrations.

3.5.4 Effect of organic matter on ²²⁶Ra activity concentrations

The addition of organic matter (solution + sand + organic matter) resulted in a small but consistent additional reduction in dissolved ²²⁶Ra concentrations relative to sand-only treatment (Wilcoxon signed-rank test, p < 0.001; Figure 3). Across all matched experimental conditions, median removal increased from 40% in Series B to 46% in Series C at 9 h and from 60% to 70% by 48 h. Similar patterns were observed across calcium treatments. In the low Ca (0.9 mg L^-1) treatment, 39% of dissolved ²²⁶Ra was removed within the first hour when exposed to sand and organic matter compared with 46% when exposed to sand only, increasing to 80% and 83% by 48 h, respectively. In the RSW (7 mg L^-1) treatment, 31% of dissolved ²²⁶Ra was removed within the first hour when exposed to sand and organic mater compared with 25% when exposed to sand only, increasing to 74% and 75% by 48 h. In the moderate Ca treatment (35 mg L^-1), removal increased from 29% when exposed to sand only to 50% when exposed to sand and organic matter by 48 h. Similarly, in the high Ca treatment (70 mg L^-1), removal increased from 58% when exposed to sand only to 60% when exposed to sand and organic matter by 48 h. Overall, the presence of organic matter increased ²²⁶Ra removal by a median of approximately 6% across matched experimental conditions. The TWA activity concentrations when exposed to sand and organic matter were estimated to range from 0.4 to 0.7, 1.3–2.5, and 14–26 Bq L^-1 for the 1, 5, and 50 Bq L^-1 nominal treatments, respectively (Figure 4).

3.5.5 Effect of calcium on ²²⁶Ra activity concentrations

The mass of ²²⁶Ra sorbed to the solid phase at equilibrium (C _*s*) was estimated when exposed to sand only and when exposed to sand and organic matter by mass balance using aqueous concentrations measured at 48 h (Table 3). Estimated C _*s* values ranged from 0.73 to 85 Bq kg^-1 when exposed to sand only and from 0.55 to 84 Bq kg^-1 when exposed to sand and organic matter, depending on ²²⁶Ra activity and Ca concentration (Table 3). The quadratic model indicated that C _*s* was strongly dependent on nominal ²²⁶ Ra (F _1,20 = 75.7, p < 0.001), whereas Ca concentration (p = 0.68), and the Ca x ²²⁶Ra interaction (p = 0.59) were not significant. At the lowest nominal ²²⁶Ra concentration, sorbed activity generally decreased with increasing Ca concentration; however, at higher nominal ²²⁶Ra concentrations the response was non-monotonic. Although C _*s* varied among Ca treatments, no consistent directional trend with increasing Ca concentration was evident.

TABLE 3. Estimated sorption parameters for ²²⁶Ra in adsorption experiments, including initial (C _*0*) and equilibrium (C _*e*) aqueous activities, fraction sorbed (fsorbed), equilibrium solid-phase activity (C _*s*), distribution coefficients (Kd), and first-order rate constants (k, reported in h-1 with associated standard errors (SE)) for Series B (solution + sand) and Series C (solution + sand + organic matter) across Ca treatments (0.9–70 mg L-1) and nominal ²²⁶Ra activities (1, 5, 50 Bq L-1).

Volume (V) and mass (g, dry weight).

Distribution coefficients (K _*d*), calculated from aqueous concentrations at 48 h, ranged from 0.3 to 11.8 L kg^-1 when exposed to sand only and 0.6–12.0 L kg^-1 when exposed to sand and organic matter (Table 3). The quadratic model indicated that Ca had a significant non-linear effect on K _*d* (F_2,18 = 4.72, p = 0.023), whereas ²²⁶Ra concentration (p = 0.61) and the Ca x ²²⁶Ra interaction (p = 0.87) were not significant. Overall, K _*d* tended to be lower at higher Ca concentrations, although the relationship was non-linear, with the clearest reduction observed at the lowest nominal ²²⁶Ra concentration.

Time-dependent changes in dissolved ²²⁶Ra were modelled using first-order kinetics to obtain comparative rate constants (k) when exposed to sand only and when exposed to sand and organic matter. Model fits captured the rapid initial removal phase followed by slower asymptotic behaviour (Supplementary Figure S2). Estimated rate constants (k) ranged from 0.06 to 0.89 when exposed to sand only, and 0.04 to 1.01 when exposed to sand and organic matter (Table 3). The quadratic model indicated that Ca had a significant non-linear effect on k (F _2,18 = 5.11, p = 0.018), whereas ²²⁶Ra concentration (p = 0.98) and the Ca x ²²⁶Ra interaction (p = 0.50) were not significant. Sorption was generally rapid at low Ca, slowed substantially at 35 mg L^-1 Ca, and increased again at 70 mg L^-1 Ca, although remained slower than at the low Ca concentration.

3.6 Radium-226 sorption to quartz and kaolinite

The results of the batch sorption experiment demonstrated that ²²⁶Ra²⁺ sorption differed significantly between quartz and kaolinite substrates (Figure 5; Table 4). Differences in percent removal between substrates were highly significant across all sampling times (two-way ANOVA, substrate effect: p < 0.001), with estimated marginal mean differences ranging from 83.6% to 86.8% between kaolinite and quartz at 6–48 h (p < 0.001 for all pairwise contrasts). In treatments containing kaolinite, dissolved ²²⁶Ra²⁺ concentrations declined rapidly within the first 6 h, with sorption exceeding 90% across all three ²²⁶Ra treatments (mean ± SD = 96 ± 2%). Specifically, ²²⁶Ra²⁺ removal at 6 h was on average 95%, 98%, and 97% for initial nominal concentrations of 1, 5, and 50 Bq L^-1, respectively. Following this rapid initial removal, dissolved ²²⁶Ra²⁺ concentrations remained stable, suggesting that equilibrium or near-equilibrium conditions were reached rapidly for kaolinite treatments. Consistent with this observation, there was no significant difference in sorbed ²²⁶Ra activity between paired 24 and 48 h for either kaolinite (Wilcoxon signed-rank, p = 0.313) or quartz (Wilcoxon signed-rank, p = 0.773).

TABLE 4. Estimated sorption parameters for ²²⁶Ra in batch sorption tests, including initial (C _*0*) and equilibrium (C _*e*) aqueous activities, fraction sorbed (fsorbed), equilibrium solid-phase activity (C _*s*), and distribution coefficients (Kd) at nominal ²²⁶Ra activities (1, 5, 50 Bq L-1).

Mean K _*d* values for kaolinite were 51, 49 and 62 L kg^-1 for the nominal 1, 5, and 50 Bq L^-1 treatments, respectively. In contrast, ²²⁶Ra sorption to quartz was minimal (K _*d* < 0.6 L kg^-1) and showed limited temporal progression. Within the first 6 h, there was minimal sorption (11% ± 6%). At 12 h, ²²⁶Ra removal remained low (0%–17%), with little change observed at 24 h (2%–21%). By 48 h, ²²⁶Ra removal remained low (0%–25%), with mean sorption 0%, 22%, and 10% for initial nominal concentrations of 1, 5, and 50 Bq L^-1, respectively.

4 Discussion

The objective of this study was to determine the main sorbing phases controlling dissolved ²²⁶Ra²⁺ activity concentrations in sandy aquatic environments and to evaluate the influence of calcium and organic matter on ²²⁶Ra partitioning. The results demonstrate that ²²⁶Ra²⁺ removal from solution under the experimental conditions of this study was governed primarily by interactions with kaolinite contained within the sand substrate, followed by organic matter and quartz, which exerted minimal influence.

4.1 Limited interaction between ²²⁶Ra and organic matter

The addition of particulate organic matter (algal cells) produced a small but consistent additional removal of ²²⁶Ra. Despite the fact that the organic matter amendment represented only a minor fraction of the total system volume (300 µL in 300 mL), the observed increase in ²²⁶Ra removal (∼6%) was statistically significant. However, the magnitude of this effect remained small relative to the much larger influence of kaolinites and calcium concentration on ²²⁶Ra partitioning under the oxic, circumneutral conditions examine in this study. This interpretation is consistent with WHAM speciation modelling, which indicated negligible interaction between ²²⁶Ra and fulvic or humic acids under the experimental conditions. These findings are also consistent with previous studies demonstrating weak association between ²²⁶Ra²⁺ and organic phases compared to mineral substrates (Rachkova et al., 2010; Thiry and Van Hees, 2008). Thiry and Van Hees (2008) reported that ²²⁶Ra²⁺ exhibits low affinity for organic matter due to its large ionic radius, low hydrolysis tendency, and preference for inorganic sorption sites such as clays. Unlike multivalent transition metals, ²²⁶Ra²⁺ does not form strong surface complexes with functional groups commonly present in organic matter, such as carboxyl or phenolic sites, limiting its incorporation into organic matrices (Brown et al., 2022).

More broadly, field observations further support the limited role of organic matter in directly controlling ²²⁶Ra mobility. Organic matter in surface tailings or sediments can enhance biological oxygen demands and promote reducing conditions (Imafidon et al., 2025; Martin et al., 2003; Xu and Tsang, 2022). Under such conditions, Fe³⁺-reducing microorganisms can dissolve Fe-(hydr)oxides and release associated ²²⁶Ra (Landa et al., 1991; Stackelberg et al., 2018; Martin et al., 2003; Kartick, 2015), while sulfate-reducing bacteria may promote dissolution of Ra-BaSO₄ minerals (Fedorak et al., 1986; Landa et al., 1991; McCready et al., 1980). While ²²⁶Ra sorbed to clay minerals (e.g., kaolinite) and quartz is not sensitive to redox conditions, reducing environments may indirectly increase ²²⁶Ra mobility through the reductive dissolution of Fe-(hydr)oxides, which are known to strongly sorb ²²⁶Ra (Chellam and Clifford, 2002; Nirdosh et al., 1990; Thiry and Van Hees, 2008). Similarly, ²²⁶Ra incorporation into sulfate minerals (e.g., Ra–BaSO₄) may also be mobilized under reducing conditions (Martin et al., 2003). Consequently, under the oxic, circumneutral conditions examined in this study, the direct influence of organic matter on ²²⁶Ra partitioning was limited.

4.2 Mineralogical control on ²²⁶Ra²⁺ sorption in sandy substrates

The addition of sand resulted in rapid and statistically significant removal of dissolved ²²⁶Ra²⁺ from solution, with 75%–85% removal observed after 48 h under low-calcium conditions. However, subsequent mineral-specific experiments demonstrated that this sorption was attributable almost entirely to kaolinite rather than quartz, the dominant component of the sand used in this experiment. The mineral-specific experiments also indicated that sorption had stabilized by 48 h, indicating that the calculated distribution coefficients (K _*d*) correspond to near-equilibrium partitioning under the experimental conditions. Sorption of ²²⁶Ra²⁺ to kaolinite was rapid and extensive (>90% removal), with distribution coefficients ranging from 38 to 49 L kg^-1, whereas sorption to quartz was negligible (Kd < 0.6 L kg^-1). These experimental observations were consistent with model predictions. Exchange modelling for kaolinite predicted approximately 94% sorption of ²²⁶Ra, in good agreement with the observed removal of 95%–98%, supporting the importance of cation exchange processes. In contrast, the SCM for quartz predicted ∼38% sorption, which was higher than the observed range (0%–25%), but still correctly indicated substantially weaker interaction relative to kaolinite. The overprediction for quartz likely reflects uncertainties associated with the simplified SCM parameterization, including the use of literature-derived surface properties, estimated specific surface area, and surrogate equilibrium constants for ²²⁶Ra. Together, these results confirm that the presence of reactive clay minerals such as kaolinites within an otherwise quartz-dominated substrate can dominate ²²⁶Ra partitioning.

The strong affinity of kaolinite for ²²⁶Ra is consistent with previous laboratory studies reporting enhanced adsorption of ²²⁶Ra²⁺ to clay minerals relative to quartz and other framework silicates (Beneš et al., 1985; Reinoso-Maset and Ly, 2016; Robin et al., 2017). Clay minerals possess higher surface area, permanent negative structural charge arising from isomorphic substitution, and greater cation exchange capacity than quartz (Davis and Kent, 1990), facilitating sorption of divalent cations such as ²²⁶Ra²⁺. In contrast, quartz exhibits limited sorption capacity due to its low surface charge density and absence of permanent charge sites, particularly at circumneutral pH (Stumm and Morgan, 1996). The relatively consistent K _*d* values observed across the tested ²²⁶Ra concentration range suggest that sorption sites were not saturated under the experimental conditions, consistent with partitioning controlled by ion-exchange processes at trace concentrations. These findings highlight the importance of mineralogical composition in controlling ²²⁶Ra mobility under the experimental conditions examined in this study.

4.3 Influence of calcium on ²²⁶Ra²⁺ sorption kinetics and partitioning

Calcium concentration exerted a significant influence on both ²²⁶Ra partitioning (K _*d*) and sorption kinetics (k), whereas no significant effect was observed for the mass of ²²⁶Ra sorbed to the solid phase (C _*s*). Distribution coefficients generally decreased with increasing Ca concentration, particularly at the lowest ²⁶Ra loading, indicating competitive inhibition of ²²⁶Ra sorption by Ca²⁺. The absence of a significant effect on C _*s* suggests that calcium primarily influenced the partitioning of ²²⁶Ra between the aqueous and solid phases rather than the total amount sorbed under the experimental conditions. This behaviour is consistent with the chemical similarity between Ca²⁺ and Ra²⁺, both of which are alkaline earth metals with similar ionic radii and sorption mechanisms (Langmuir and Riese, 1985; Riese, 1982). Competition for sorption sites on kaolinite and other mineral surfaces reduces the availability of binding sites for Ra²⁺, resulting in lower sorbed fractions and greater persistence of ²²⁶Ra in solution.

The influence of calcium on sorption kinetics was non-linear, with intermediate Ca concentrations producing slower apparent sorption rates compared to both low and high Ca treatments. The partial recovery in sorption rates at 70 mg L^-1 Ca relative to 35 mg L^-1 Ca was unexpected and suggests that additional processes may influence ²²⁶Ra sorption under increasing calcium concentrations. One possible explanation is that increased ionic strength associated with higher calcium concentrations may compress the electrical double layer, facilitating faster transport of ²²⁶Ra to mineral surfaces. However, competition between Ca²⁺ and Ra²⁺ for sorption sites ultimately limits the extent of ²²⁶Ra partitioning to mineral surfaces. Because these processes were not directly evaluated in the present study, the mechanism responsible for the observed non-linear kinetic response remains uncertain and warrants further investigation. Nevertheless, the overall influence of calcium observed here is consistent with previous studies demonstrating that major cations such as Ca²⁺ play an important role in controlling ²²⁶Ra sorption and mobility in natural waters (Beneš et al., 1985; 1984; Langmuir and Riese, 1985).

4.4 Implications for ²²⁶Ra mobility and availability to aquatic organisms in sandy environments

The results of this study have implications for predicting ²²⁶Ra mobility and availability to aquatic organisms inhabiting sandy substrates. The presence of reactive clay minerals such as kaolinite is likely to substantially reduce dissolved ²²⁶Ra concentrations, thereby potentially reducing dissolved exposure and the bioavailable fraction to aquatic organisms. Conversely, environments dominated by quartz-rich sands with low clay content may exhibit limited sorption capacity, potentially resulting in greater ²²⁶Ra mobility and dissolved concentrations. These findings are particularly relevant when studying ²²⁶Ra toxicity to benthic invertebrates such as fingernail clams (Sphaeriidae), which inhabit sandy and fine-grained substrates and are primarily exposed to ²²⁶Ra via overlying water, with additional exposure from interstitial water in sediments and through sediment ingestion (Proulx, 2026; Wiramanaden et al., 2015). Greater partitioning of ²²⁶Ra to sediments would be expected to reduce uptake from the overlying water column. However, uptake through sediment ingestion or contact with sediment-bound particles may also contribute to overall exposure, particularly in ²²⁶Ra-impacted environments such as uranium mine tailings and downstream aquatic systems (Goulet et al., 2022).

One limitation of the application of our study to natural sandy sediments is that Fe- and Mn-(hydr)oxide phases were largely absent from our experimental substrate. These minerals are recognized as important sorbents for ²²⁶Ra because of their high surface area and strong affinity for divalent cations. Previous studies have identified Fe-(hydr)oxides as important sinks for ²²⁶Ra in sediments and uranium mine tailings (Landa and Reid, 1983; Liu and Hendry, 2011). Accordingly, natural sediments containing both clay minerals and Fe- and Mn-(hydr)oxide phases would be expected to exhibit greater ²²⁶Ra sorption than the sandy substrate evaluated in the present study. However, in natural systems, accumulation of organic matter increases biological oxygen demand and creates reducing conditions. Under such conditions, Fe³⁺-reducing microorganisms can dissolve Fe-(hydr)oxides and release associated ²²⁶Ra (Landa et al., 1991; Stackelberg et al., 2018; Martin et al., 2003; Kartick, 2015), while sulfate-reducing bacteria may promote dissolution of Ra-BaSO₄ minerals (Fedorak et al., 1986; Landa et al., 1991; McCready et al., 1980). Under such reducing conditions, sorption to Fe-(hydr)oxides and Ra-BaSO₄ may become less important, leaving clay minerals as the main retention mechanism of ²²⁶Ra in sandy sediments.

5 Conclusion

Overall, this study demonstrates that ²²⁶Ra partitioning under the experimental conditions examined was controlled primarily by mineralogical composition and aqueous geochemistry, particularly the presence of reactive clay minerals and competing major cations such as Ca. Organic matter played a limited direct role in ²²⁶Ra sorption under the oxic, circumneutral conditions of this study. Collectively, these findings improve our understanding of the factors controlling ²²⁶Ra partitioning in sandy substrates and provides a framework for predicting ²²⁶Ra mobility and potential availability to aquatic organisms in natural and contaminated aquatic environments.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fenvc.2026.1912040/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Roxana Suehring, Toronto Metropolitan University, Canada

Reviewed by: Chaoran Li, Nanjing Normal University, China

Noha Imam, National Institute of Oceanography and Fisheries (NIOF), Egypt