Multi-cation exchanges involved in cesium and potassium sorption mechanisms on vermiculite and micaceous structures

Vermiculite and micaceous minerals are relevant Cs+ sorbents in soils and sediments. To understand the bioavailability of Cs+ in soils resulting from multi-cation exchanges, sorption of Cs+ onto clay minerals was performed in batch experiments with solutions containing Ca2+, Mg2+, and K+. A sequence between a vermiculite and various micaceous structures has been carried out by conditioning a vermiculite at various amounts of K. Competing cation exchanges were investigated as function of Cs+ concentration. The contribution of K+ on trace Cs+ desorption is probed by applying different concentrations of K+ on Cs-doped vermiculite and micaceous structures. Cs sorption isotherms at chemical equilibrium were combined with elemental mass balances in solution and structural analyses. Cs+ replaces easily Mg2+  > Ca2+ and competes scarcely with K+. Cs+ is strongly adsorbed on the various matrix, and a K/Cs ratio of about a thousand is required to remobilize Cs+. Cs+ is exchangeable as long as the clay interlayer space remains open to Ca2+. However, an excess of K+, as well as Cs+, in solution leads to the collapse of the interlayer spaces that locks the Cs into the structure. Once K+ and/or Cs+ collapse the interlayer space, the external sorption sites are then particularly involved in Cs sorption. Subsequently, Cs+ preferentially exchanges with Ca2+ rather than Mg2+. Mg2+ is extruded from the interlayer space by Cs+ and K+ adsorption, excluded from short interlayer space and replaced by Ca2+ as Cs+ desorbs.


Introduction
Subsequently to the nuclear contamination in the Fukushima prefecture, numerous studies have been devoted to the understanding of radiocesium mobility to predict its impact in the environment and thus anticipate potential risks on health (Klika et al. 2007;Nakao et al. 2014;Liu et al. 2004;Kasar et al. 2020;Siroux et al. 2021). To counteract the bioavailability of cesium in soils, K fertilization was recommended in Japan (Kato et al. 2015) to decrease Cs incorporation by plants by acting on competing rule between Cs + and K + in plants as well as in soils (Kato et al. 2015;Kubo et al. 2015Kubo et al. , 2017. Nevertheless, the transfer of Cs + to plants is highly dependent on the clay mineralogy, especially as soils that are enriched in micaceous minerals and vermiculite (Kubo et al. 2018). Weathering micaceous minerals are able to release K + to reduce Cs + uptake by plants. On the other hand, biotite, vermiculite, and aluminous smectite are known to strongly adsorb Cs + compared to other cations (Sakalidis et al. 1988;Cornell 1993;Maes et al. 1999;Mukai et al. 2016;Durrant et al. 2018;Wissocq et al. 2018;Ogasawara et al. 2019;Kitayama et al. 2020). Furthermore, illite and partially vermiculitized biotite sorb Cs + more strongly than vermiculite (Sawhney 1972;Brouwer et al. 1983;Maes et al. 1999;Kitayama et al. 2020;Latrille and Bildstein 2022). Indeed, these clay minerals provide selective sorption sites that enhanced Cs + sorption (Brouwer et al. 1983;Park et al. 2019), which in some cases appear as Cs fixation sites (Kikuchi et al. 2015). Cs + sorption is mainly controlled by an ion exchange process on the surfaces of clay mineral (Mishra et al. 2015).
Due to isomorphic substitutions in tetrahedral and octahedral layers and hydroxyl groups with amphoteric character Responsible Editor: Tito Roberto Cadaval Jr at the particle edges, these minerals hold different types of sorption sites. Generally, sorption sites are structurally located on the basal planes of the phyllosilicate layer, on the crystal border, and in the interlayer space. Weathered micaceous minerals and in particular illite exhibit frayed edges which refer to the transition zones between collapsed and open interlayer space. Since Sawhney (1972), particular reactive properties have generally been attributed to this specific space. Due to its configuration, it can only accept cations with low hydration energy or small size cations, notably K + and Cs + , compared to Ca 2+ and Mg 2+ . In addition, Cs + can deeply penetrate inside the closed interlayer space by ion exchange Okumura et al. 2018).
The adsorption process depends on the Cs + concentration, the solution ionic strength, the competing cation, and the pH (Fuller et al. 2014;Missana et al. 2014). This involves the compensating cation placed on the negative surface charges of phyllosilicates and the cations in solution. At trace concentrations, Cs + is easily replaced by Mg 2+ Tamura et al. 2015) on vermiculite, while Cs + is more strongly adsorbed on illite (Wissocq et al. 2018) but remains exchangeable (Latrille and Bildstein 2022). At high Cs + concentrations, it has been assumed that weakly hydrated Cs + sorbed to the frayed edges of micaceous minerals is dehydrated leading to a partial collapse of the interlayer (Mukai et al. 2016;Sawhney 1972;Kikuchi et al. 2015;Park et al. 2019). Likewise, Cs + incorporated in the vermiculite interlayer by ion exchange with Mg 2+ , Ca 2+ , and Na + induces subsequent interlayer collapse in vermiculite (Sawhney 1972;Lee 1973Lee , 1974Kogure et al. 2012;Dzene et al. 2015;Yin et al. 2017;Kitayama et al. 2020). Cesium trapped in the interlayer space is therefore excluded from desorption Motokawa et al. 2014a, b;Yamamoto et al. 2019) or is weakly displaced by hydrated divalent cations (Sawhney 1964;Yin et al. 2017). Only Cs + adsorbed on planar and edge sites remain exchangeable (Dzene et al. 2015;Yin et al. 2017).
Cesium sorbed on micaceous minerals is generally exchanged in small amounts with competitive cations such as K + and NH 4 + (Liu et al. 2004;Ohnuki and Kozai 2013) despite their similar chemical properties. This indicates that the minerals are more selective for Cs + than K + and NH 4 + . NH 4 + extracts smaller amounts of Cs + than hydrated Mg 2+ and Ca 2+ ; K + is the most efficient compared to the other cations (Yin et al. 2017). Moreover, dehydrated K + originally found in the interlayer space of micaceous minerals remains non-exchangeable unlike those of expanding phyllosilicate. Only the weathering process partially releases interlayer cations from micas Sawhney 1972). Then, superimposed on the chemical ion exchange, a physical mechanism leads to the capture of cations with low hydration potential into the mineral. The competition between K + and Cs + is associated with their location in the mineral structure and Cs + sorption properties of micaceous minerals differ depending on their degree of weathering (Kitayama et al. 2020;Murota et al. 2020). Consequently, the sorption properties of pristine minerals deviate from those which are weathered in soils.
To understand the bioavailability of radionuclides in soils, knowledge of sorption-desorption behavior in soil-water systems remains crucial (Kasar et al. 2020). At chemical equilibrium, the reversibility of the adsorption process can be validated by desorption under the same physicochemical conditions. However, Cs + desorption properties of clay minerals are frequently deduced from forced ion exchange accounting chemical extractants (Brouwer et al. 1983;Dzene et al. 2015;Yin et al. 2017;Mukai et al. 2018;Kitayama et al. 2020;Murota et al. 2020), which qualifies a different reaction pathway of the adsorption process. The partitioning between solid and solution, usually named Kd, is finally recommended to verify the adsorption reversibility (Wissocq et al. 2018;Kasar et al. 2020). The same Kd value is expected for both adsorption and desorption reactions under the same physicochemical conditions. This implies a control of the chemical exchange at equilibrium by respecting the mass action law and thus the cations mass balance in the system. The sorption of Cs + on clay mineral is often studied in a context of a single cation pair exchange or far from a natural context. Although cation adsorption/desorption depends on competitions with surrounding cations in solution, especially major cations, few studies have been focused on the sorption of Cs + in simultaneous competition with several cations in soil solution (Kasar et al. 2020).
The aim of this study is to further understand the adsorption and desorption behavior of Cs + on micaceous minerals and vermiculite in a chemical environment close to a natural soil. A plant nutritive solution was selected, containing the major cations Ca 2+ , K + , and Mg 2+ and deficient in trace elements. The analogous chemical behavior of K + and Cs + leading to their competitions was highlighted by using solutions containing various K + concentrations. In order to mimic the transition from weathered mica to vermiculite, a vermiculite was saturated with K + at various degrees of interlayer closing. Their exchange properties were then monitored. Finally, the adsorption/desorption behavior of Cs + is investigated in multi-cation solutions. This involves simultaneous competition of Cs + with Ca 2+ , Mg 2+ , and K + . Cesium concentration isotherms are acquired in batch experiments performed under controlled physicochemical conditions at thermodynamic equilibrium. Desorption experiments performed with the same cation species allow to test the reversibility of Cs + adsorption. Cation mass balances are checked for each chemical condition to highlight the cation competitions and selectivities that occur during Cs + sorption. Then K + competition on Cs + desorption is studied by imposing various amounts of K + on a trace Cs + -doped vermiculite and a trace Cs + -doped K-intercalated vermiculite. The K + concentrations were selected to mimic a K amendment in the fields, a depletion of K + in the solution due to plant uptake or weathering event, and to provide insight into the location of Cs + in the mineral structure. Subsequent chemical disturbances in solution were recorded through cation mass balances. Finally, the chemical results coupled with structural XRD analyses led to the proposal of a reaction scheme for the adsorption and desorption of Cs + in multications solutions. The structural location of the sorption sites is discussed.

Materials and conditioning process
A commercial vermiculite (Sigma-Aldrich) was selected for this study. The chemical analysis of the raw vermiculite gives the following structural formula Si 2.905 Al 1.021 Fe 0.075 Fe 0.311 Mg 2.624 Ti 0.06 O 10 (OH) 2 K 0.204 Ca 0.034 Mg 0.201 . XRD analysis revealed a pristine vermiculite associated with a vermiculite/mica mixed layer. The CEC determined by NH 4 + exchange at pH 7 is 109.6 ± 1.2 meq/100 g compared to the theoretical CEC estimated to 166 meq/100 g. The nonexchangeable K + by NH 4 + is consistent with a mixed layer vermiculite/mica. The raw vermiculite was initially finely crushed and suspended in ultrapure water overnight and then sonicated in a tank and sieved to isolate the 25-20-µm particles. The conditioning steps lead to imposing the initial composition of compensating cations on the negative charges of the solids. This imposes the amount of K + , Ca 2+ , and Mg 2+ associated with the vermiculite and consequently defines the different conditioned vermiculites S(x)V in the transition from micaceous vermiculite to true vermiculite. Then, the conditioned solids are chemically equilibrated with the S(x) solutions used for the adsorption/desorption experiments. The composition of the S(x) solutions is based on the nutrient solutions for Arabidopsis thaliana cultivation (Conn et al. 2013) and simplified to avoid NH 4 + , Na + , and micronutrients competitions with Cs + in the systems. The conditioning and experimental solutions were prepared by dissolving salts and diluting them with ultrapure water (milli-Q, 18.2 MΩ.cm −2 , Millipore®). Salts of KCl (VWR Chemical), Ca(NO 3 ) 2 4H 2 O (VWR Chemical), Mg(SO 4 ) 7H 2 O (VWR Chemical), and Mg(NO 3 ) 2 6H 2 O (VWR Chemical) were used. Acidification was done with ultrapure nitric acid (Chem-Lab) and phosphoric acid (Sigma-Aldrich) diluted to appropriate concentrations where needed to achieve the targeted pH value to 5.6 in S(x) solutions under conditions optimized for plant growth. The 20-25-µm vermiculite fraction was partitioned in five batches, each equilibrated with five solutions containing 2.5 10 −2 mol/L Ca and 3.75 10 −2 mol/L Mg and varying by their K + concentrations at 0, 0.125, 1.25, 6.25, and 12.5 10 −2 mol/L K. For the conditioning process, batches of 1.5 g of vermiculite were dispersed in 120 mL of solutions with a solid/solution ratio of 12.5 g/L, end to end shaken for 3 days and centrifuged at 10,596 g during 40 min to isolate the supernatant, which is replaced by the same conditioning solution. These operations were repeated three times. Three similar operations were successively realized with the solutions used for experiments prepared with the same solution composition but diluted 12.5 times, named S(x) solution (x = 0, 0.1, 1, 5, or 10 K mmol/L concentration) with a cation composition detailed in Table 1. This step contributes to equilibrate the solids with the operating solutions. The solids were finally rinsed with ethanol to eliminate residual salts. After centrifugation, the five conditioned materials were left to air-dry for 2 days.
The cation exchange capacities (CEC) of the conditioned vermiculites are deduced from the compensating cations exchanged by NH 4 + at pH 7. Fifty mg of dried conditioned vermiculite was dispersed in 4 mL of 0.1 mol/L NH 4 Cl, end to end shaken for 1 week and centrifuged at 33,000 g during 40 min. The cation concentrations of the supernatants were analyzed by ionic chromatography, and low K concentrations were refined by ICP-AES measurements (Activa, Horiba Jobin Yvon).

Cs adsorption/desorption at equilibrium
The effects of Cs + concentration were investigated by sorption isotherms for each conditioned vermiculite following the approach described in Wissocq et al. (2018) and Siroux et al. (2021). For all these batch experiments, 70 mg of dried conditioned vermiculite was dispersed in 6.1 mL of S(x) solution, leading to a solid/solution of 11.4 g/L. Each batch suspension was spiked with 137 Cs by adding 30 µL of a source ( 137 CsCl in HCl 0.1 mol/L) containing an activity of 7000 Bq, corresponding to 4.6 10 −8 mol/L of Cs introduced. For the higher concentrations required, stable isotope (CsCl solution, Fluka Ultra) was added to achieve a solute concentration ranged between 10 −7 and 10 −2 mol/L. Protons introduced by the radiotracer were compensated by the addition of deaerated 0.02 mol/L Ca(OH) 2 solution freshly prepared. The pH was naturally buffered by conditioned vermiculites and checked. The suspensions were orbital shaken during 14 days, sufficient contact time to reach chemical equilibrium (based on Dzene et al. (2015) study, and verified by kinetics measurement in SI1), and centrifuged at 33,000 g (Beckman Optima LE-80 K Ultracentrifuge) for 45 min. The supernatants were collected for analysis of electrolyte concentration and 137 Cs counting and to check the pH at equilibrium. In order to verify the reversibility of the Cs + adsorption process and the cations exchanges occurring during the desorption process in accordance with Eq. 1, desorption experiments were carried out on the slurry resulting from the adsorption step. S(x) solutions free of Cs + were used to exchange the adsorbed Cs + with cations in solutions respecting the pH, the solid/solution ratio, and the contact time used in the adsorption step. Then, 6 mL of the same S(x) solution used for the adsorption step was introduced each batch. Suspensions were orbital shaken during 14 days and centrifuged at 33,000 g during 45 min. 137 Cs and cations (Ca 2+ , Mg 2+ , K + ) in the supernatants collected after centrifugation were analyzed by gamma counting and ion chromatography respectively. The activity of the residual adsorption solution, not recoverable from the slurry, was taken into account to define the initial activity of 137 Cs adsorbed on the vermiculites for the desorption step. The total Cs + concentration is deduced from the 137 Cs activity by assuming no isotopic fractionation between 137 and 133 Cs, as previously verified by Wissocq et al. (2018). In addition, no natural Cs amount was measured on raw vermiculite. The pH was also checked at equilibrium.
The effects of K + concentrations on Cs + desorption were investigated by batch experiments on conditioned vermiculites S(0)V and S(5)V doped in 137 Cs at 7.6 10 -8 mol/L (6.62 10 −6 mol/kg of adsorbed Cs). For this purpose, 70 mg of dried conditioned and doped vermiculite were dispersed in 6 mL of S(0) solution with the addition of KCl solutions at different concentrations. For K-free vermiculite (S(0) V), the introduced KCl concentrations were ranged from 3 10 −6 mol/L to 5.9 10 −2 mol/L, whereas KCl concentrations were ranged between 1.5 10 −4 and 4.9 10 −2 mol/L for conditioned vermiculite S(5)V. The suspensions were orbital shaken during 14 days and centrifuged as described above. The supernatants were collected for analysis of electrolyte concentration, 137 Cs counting and to check the pH at equilibrium.
Aqueous cation concentrations were measured by ion chromatography (standards Chem-Lab). The measurement uncertainties were 5 10 −6 mol/L 2σ (i.e., 10 −6 eq, equivalent ( +) charge). The radioactivity of 1 mL supernatant was measured by gamma counting (Perkin Elmer Wizard 3). pH values were determined using a combined glass pH electrode (Mettler Toledo) equipped with an Ag/ AgCl reference electrode. The electrode was calibrated with buffer solutions at pH 4.0, 7.0, and 12.0. Uncertainties were calculated by error propagation and established at ± 0.2 pH unit.
Cations adsorption and desorption by exchange with cations in solution at chemical equilibrium are expressed by Eq. 1: with X − i n − N n+ and X − i m − M m+ the adsorbed cations on the X − i sorption sites, and M m+ and N n+ , the cations in solution. These reactions are described by the apparent exchange constants according to the mass action law, such as Eq. 2: where K * is the apparent exchange constants, i is the type of sorption site, [] is the concentration of species in solution (mol/L) or sorbed species (mol/kg of dry solid), and γ is the activity coefficient of species in solution.
Both Cs adsorption and desorption isotherm results were expressed in terms of distribution coefficient K d , which is defined by the ratio of the adsorbed Cs concentration ( [Cs] ads,eq in mol.kg −1 ) over the Cs concentration in solution ([Cs + ] sol,eq in mol.L −1 ). Thanks to 137 Cs radioisotope, K d can be also estimated by the Cs activity initially introduced (A ini in Bq) and measured in solution at equilibrium (A sol,eq in Bq), as expressed in Eq. 3: The mass balance expressed the change in chemical composition that occurs in solution during the cation exchange process. Each concentration of cation in solution was converted to equivalent of cation (mole of positive charge) in the system. The mass balance of each batch (Δeq) was therefore calculated by the equivalent of cation at equilibrium in solution subtracted from the equivalent of cation initially introduced by the S(x) solutions. For the adsorption experiments, the introduced solutions S(x) were added with Cs + , while for the desorption experiments, the S(x) solutions are free of Cs + . Negative values express an uptake onto the solid. Conversely, positive values represent a release into the solution.
Experimental errors were hence calculated for each batch following the propagation error theory. The variance of a G function of different x i variable can be calculated from the variances of the variables x i using this expression: is the partial derivative of G with respect to x i and σ 2 is the variance of x i and σ ij the covariance of the x i and x j variables. If these variables are independent, the covariance term is then equal to zero.

Crystal structure analysis
X-ray diffraction (XRD) analyses were performed using a XRD 5000 INEL powder X-ray diffractometer using Cu Kα radiation, equipped with a CPS120 curve detector Si/Li. The XRD patterns were scanned between 3° and 110° 2θ angle at 30 kV and 30 mA with a Cu-target tube at room temperature of 22 °C. The patterns were cumulated every few seconds during 10 min. Analyses were performed on oriented clay deposits to characterize the cation exchanges occurring in the interlayer space. The peak for (001) reflection was scanned to detect shifts due to occupancy change (Yin et al. 2017). This reflection gives the interlayer distance of vermiculite structures. The basal spacings (d (001) ) are ranged between 14.3 to 14.9 Å for mixed interlayer occupancy by Ca 2+ and Mg 2+ , whereas d (001) spacing is reduced to 10.4 Å for K + occupancy or 10.8 Å to 11.4 Å for Cs + occupancy in vermiculite (Brindley and Brown 1980;Dzene et al. 2015;Kogure et al. 2012). For this purpose, the XRD patterns were recorded on samples resulting from Cs + desorption experiments. Vermiculite slurries were dispersed in pure water, dropped on glass slides and air dried to obtain oriented preparations.

Exchange capacity of conditioned vermiculites
The conditioning of vermiculites with different S(x) solutions leads to imposing the composition of compensating cations to the vermiculite negative charges. Moreover, it allows the selection of different intercalated K structures in a sequence between vermiculite and K-depleted micaceous structures. The analyses of the cations extracted with NH 4 Cl at pH 7 were performed on each conditioned vermiculite. The sum of these exchangeable cations leads to the CEC ( Table 2). The vermiculite conditioned without K + (S(0)V) indicates the highest CEC at 127 cmol( +)/kg. K + is largely extracted from the original vermiculite by the conditioning process. This is confirmed by XRD analysis that shows a (001) reflection with a d value at 14.6 Å, indicating a major occupancy of the interlayer space by di-hydrated Ca 2+ and Mg 2+ (Fig. 1). As K + increases in the conditioning solutions, a decrease in the CEC of the conditioned vermiculites is observed. Indeed, K + appears to be very difficult to extract with NH 4 + : the exchangeable K + represents less than 1% of the extractable cations for S(0)V to S(5)V with a maximum of 1.4% for S(10)V. Moreover, with the increasing of the amount of K + in the conditioning media, the intensity of the (001) reflection at 14.6 Å decreases and peaks open towards the large angles (6 to 6.5° 2θ angle), while a second broad reflection at 12.5 Å appears and grows in intensity (Fig. 1). Therefore, some K + is incorporated into the interlayer space of vermiculite during the conditioning process to form vermiculite/ mica mixed layers.
Based on the maximal value of CEC obtained from S(0) V (Table 2), the measured CEC differences between S(0)V and the other S(x)V express the amount of non-exchangeable K + incorporated in the mineral structure. This K + accounts for 11, 18, 27, and 33% of the CEC for S(0.1)V to S(10)V, respectively. While the Mg/Ca ratios was maintained at 1.5 in each S(x) solution, it decreased from 2.96 to 1.53 for the extractable cations of the S(0)V to S(10)V structures. This  indicates the affinity of vermiculite for Mg 2+ is higher than Ca 2+ in the absence of K + . As the amount of K + increases in the structure, the amount of Mg 2+ decreases, while the incorporated Ca 2+ varies slightly. Then, K + replaces Mg 2+ more easily than Ca 2+ during its incorporation into the structure (Robin et al. 2015).

Cs sorption behavior on vermiculite and micaceous structure in multi-cation media
Adsorption and desorption isotherms of Cs + (Fig. 2) were performed on the five conditioned vermiculites in function of Cs concentration in the multi-cation S(x) solutions at chemical equilibrium. The repartition of Cs + between solid and liquid is expressed by the Kd value. Similarly, to illite and smectite, the adsorption of Cs + on vermiculite is nonlinear with the Cs + concentration in solution (Durrant et al. 2018;Wissocq et al. 2018;Latrille and Bildstein 2022). Regardless of the matrix and the solution composition, as the Cs + concentration increases in solution, the Cs adsorption K d decreases to reach a local minimum between 2 and 4 10 −6 mol.L −1 (− 5.7 and − 5.4 in log scale) and then increases to a local maximum between 0.38 and 2.4 10 −4 mol/L Cs (− 4.4 and − 3.6 in log scale) depending on the solution, before decreasing again at the highest Cs + concentrations. The highest adsorption of Cs + occurs on vermiculite in K-free solution (S(0)) at Cs + concentration below 2 10 −6 (− 5.7 in log scale) and up to 10 −5 mol/L (− 5 in log scale). At trace concentrations of Cs + (< 10 −7 mol/L, < − 7 in log scale), the higher the amount of K + in the conditioned vermiculites and their respective conditioning solution, the lower the Kd value (for K + less than 1 mmol/L). Above 5 mmol/L K + in the media, the Kd of trace Cs + becomes slightly larger. All matrices give a similar lowest Kd value at about 10 −6 mol/L Cs + in solution. Conversely, the highest K loaded vermiculite (S(10)V) records the lowest Kd of Cs + adsorption above 10 −5 mol/L Cs + .
The reversibility of adsorption process is verified when the apparent exchange constant for adsorption reaction is the inverse of the apparent exchange constant for desorption reaction at chemical equilibrium ( K * ads = 1 K * des ), according to the mass action law (Eqs. 1 and 2). As the major cation concentration remains constant for both reactions, especially at low Cs + concentration, the Kd of Cs + adsorption is similar to the Kd of Cs + desorption (Eq. 3) (Durrant et al. 2018;Wissocq et al. 2018). This means that the desorption reaction must have released the right Cs + amount into the liquid phase to reach equilibrium. This does not imply that the total amount of adsorbed Cs + must be desorbed at equilibrium. At a Cs + concentration below 10 −7 mol/L (< − 7 in log scale), Cs + adsorption is reversible until 5 mmol/L of K + in solution (Fig. 2). Only 4.3 to 5.6% of the adsorbed Cs + is released during the desorption step. This suggests that Cs could be desorbed over a long period of time (Murota et al. 2016). For K + concentration higher than 5 mmol/L, the adsorption becomes partially reversible as shown by the desorption Kd slightly higher than adsorption Kd for a K + concentration at 10 mmol/L (S(10)). Between 10 −7 and 5 10 −5 mol/L (− 7 and − 4.3 in log scale) of Cs + in solution at equilibrium, reversible adsorption is the lowest, with very close Kd values whatever the K + concentration in the medium. At higher Cs + concentration (> 5 10 −5 mol/L), partial reversible adsorption is observed with the highest adsorption/desorption Kd ratio for S(0)V. This ratio decreases with increasing K + contents in solids and solutions. Thus, the adsorption capacity of Cs + provided by sorption sites located in the interlayer space is reduced due to their occupancy by K + , in accordance with the CEC (Table 2) and the (001) reflection distances decrease with the K + incorporation in the structure (Fig. 1). These non-linear isotherms are comparable to those of illite and smectite (Wissocq et al. 2018), except for the increase in Kd for a Cs + concentration above 10 −5 mol/L. They emphasize the involvement of several sorption sites distinguished by their selectivity towards Cs + , as seen on Ca-vermiculite (Latrille and Bildstein 2022). Referred to this latter study, at trace concentration, a low capacity and high selectivity site contributes mainly to Cs + adsorption. This type of sorption site also records a strong selectivity for K + given the decrease in Kd of Cs + as K + increases in both solid and solution. It should be noted that the concentration of K + must be at least a thousand times higher than that of Cs for this competition to occur. As the capacity of this site to load Cs + is saturated, the Kd of Cs + decreases drastically. Subsequently, one or several sites with low Cs affinity for Cs + continue to take up Cs + , which explains the decrease in the Kd value. At concentrations above 10 −5 mol/L, the high uploading of Cs + may result from the presence of one or several high capacity sites. Lee (1973Lee ( , 1974 and Latrille and Bildstein (2022) observed a similar trend onto vermiculite saturated Ca 2+ , Mg 2+ , or Na + and explained it by the penetration of Cs + into the interlayer space depending on the compensating cation. In the case of several cations involved in the uptake of Cs + by conditioned vermiculites, the cation mass balance of the solution compositions gives the variation of the amounts of elements (in equivalents) in the solution compared with the initial solution S(x). As a result, this balance was used to evaluate the ongoing reactional processes. The focus was made on the cation mass balances obtained on the adsorption and desorption solutions resulting from equilibrium between S(0)V and S(5)V vermiculites and the S(0) and S(5) solutions, respectively (Figs. 3 and 4). Indeed, these two sorption conditions are selected to highlight two Cs + adsorption behaviors depending on the rate of K + intercalated in the vermiculite structure. In the initial state, compensating cations on S(0)V and S(5)V are mainly Ca 2+ and Mg 2+ , and K + appears as poorly exchangeable in both vermiculites structures ( Table 2). Note that the chemical compositions of the S(0) and S(5) solutions (Table 1)     express the mobility of Cs + taking into account the cations selectivity and chemical competitions. The mass balances of the cations involved in cation exchanges with Cs + on S(0.1) V, S(1)V, and S(10)V are displayed in SI2. The significant Δeq are delimited by the limit of quantification by ion chromatography and the cumulated mass balance uncertainties, fixed at 10 −6 equivalent ( +) charge for Ca 2+ , Mg 2+ , and K + and the experimental uncertainty to 4 10 −13 Eq. (2σ) calculated for Cs + . As cation exchange with H + can occur for sorption of Cs + onto clay minerals (Wissocq et al. 2018;Durrant et al. 2018), the pH of the adsorption and desorption solutions was checked on the solutions along the isotherms at equilibrium. The pH value did not vary significantly around 5.6 (initial solutions pH) except for the adsorption experiment performed with the highest Cs + concentration (SI3) in which pH drops to 5.1. However, this pH value variation, although significant, associates a negligible amount of protons that could be involved in the exchanges (i.e., 3.7 10 −8 mol of H + ) with Ca 2+ , Mg 2+ , and K + . Moreover, H + slightly affects the adsorption/desorption behavior of Cs + (Latrille and Bildstein 2022) in this pH range.
In the adsorption step (Fig. 3), the cations content of solutions at equilibrium are compared with those of S(x) solutions with addition of Cs + . The adsorptions of Cs + on S(0) V and S(5)V at a concentration below 10 −4 mol/L (− 4 in log scale) of Cs + (Fig. 3) indicate that the composition of the solutions remains mostly similar to the initial solutions. The Δeq of Ca 2+ and Mg 2+ in solution are not significant (ca. 1 10 −6 eq) for K + -free media, whereas the significant release of Mg 2+ and Ca 2+ is of few extents (ca. 2 10 −6 eq) in media containing K + . The adsorption of Cs + in trace level does not significantly vary the cationic composition of the solutions. Regardless of the amounts of Cs + uptake until 2 10 −5 mol/L (− 4.7 in log scale) of Cs + concentration in the solution, Ca 2+ and Mg 2+ tend to exchange equally with Cs + . Moreover, the Cs + adsorbed on S(0)V is of the same amount as on S(5)V (Fig. 3). The amounts of K + and H + remain constant in solution whatever the concentration of Cs + introduced; therefore, they do not contribute to the exchange with Cs + (SI2). This corroborates the fact that the K + incorporated in the vermiculite structure such as S(5)V remains non-exchangeable with Cs + . Beyond 10 −4 mol/L (− 4 in log scale) of Cs + , Cs + largely depletes from solution and is compensated by Mg 2+ > Ca 2+ desorption. Larger adsorption of Cs + occurs on K-free vermiculite in accordance with its larger CEC than on the vermiculite with K + incorporated. At the highest Cs + concentration, Cs + compensates 74% of the K-free vermiculite CEC compared to 81% of the vermiculite with K + incorporated. The mass balances of cations remain equilibrated under all conditions tested. This further advocates for a cation exchange process to describe the adsorption of Cs + . The cation exchanges observed on the intermediate K + intercalated vermiculite, S(0.1)V and S(1) V in SI2, indicate the same trends: below 10 −4 mol/L Cs + , the releases of Mg 2+ and Ca 2+ are of few extent and become progressively significant with increasing K + in the media. This suggests an easier desorption of Mg 2+ and Ca 2+ in the presence of K + even though the K + concentration remains constant in solution. This should be related to a slight structural change that will be discussed below.
In the desorption step (Fig. 4), the cation composition in the solutions at equilibrium is compared with those of initial S(x) solutions (Table 1). No significant adsorption of K + , Ca 2+ , and Mg 2+ (< 1 10 −6 eq) is recorded as compensation for the release of trace Cs + . This expresses that the media remains mainly at equilibrium during the exchange with Cs + . The concentrations of Cs + released are too low to disturb the major concentration of Ca 2+ , Mg 2+ , and K + in solution under these experimental conditions. In addition, trace Cs + desorbed from S(0)V is of same amount as that from S(5) V. When the Cs + concentration becomes competitive with the major cations, above 3 10 −5 mol/L Cs + (− 4.5 in log scale), the release of Cs + is accompanied by the release of Mg 2+ , which are compensated by the adsorption of Ca 2+ for K-free vermiculite (Fig. 4a). For K-intercalated structures (Fig. 4b), the Cs + release is mainly compensated by Ca 2+ and to a lesser extent by K + (> 1 10 −6 eq). This suggests a higher selectivity of the sites loading Cs + for Ca 2+ than Mg 2+ or a lack of sites accessibility for Mg 2+ . At high Cs + concentration, the more K + is initially intercalated in the structure the higher the amount of Cs + desorbed. Indeed, the amount of Cs + released from S(5)V (5.4% of the CEC) is greater than that from S(0)V (1.2% of the CEC) although its CEC is lower (Figs. 4 and SI2). Thus, it appears that Cs + desorption is easier in K-intercalated than in K-free structures of the vermiculite. This suggests a competition between Ca 2+ and K + with Cs + or a structure collapse of the interlayer space inhibiting the Cs + desorption as previously observed in many studies Dzene et al. 2015;Latrille and Bildstein 2022). As the intercalated K + is higher in the structure, Mg 2+ and Ca 2+ replace Cs + (see S(10)V in SI2). This argues for desorption of Cs + from easily accessible sorption sites.
Two sorption behaviors of Cs + are revealed by chemical analyses of solutions in equilibrium with vermiculites at different rates of K + incorporation. As expected, the K + incorporated into the structure is not significantly displaced by Cs + during the Cs + adsorption/desorption process. The major cation composition of the solutions remains stable during Cs + adsorption/desorption in trace concentration due to the concentration difference between the major cations and Cs + involved in the exchange. The reversible adsorption process is verified by the Kd of Cs + values (Fig. 2). The low ratios between the amounts of Cs + desorbed in solution and those adsorbed reflect the high selectivity of Cs + towards the structures. When the concentration of Cs + exceeds 10 −5 mol/L, the cationic competition between Ca 2+ , Cs + , and Mg 2+ takes place: Cs + mainly replaces Mg 2+ for adsorption, while Ca 2+ mainly replaces Cs + for desorption. Cs + sorption sites appear more available for cation exchange in a K-incorporated structure than in a K-free structure.

K competition with Cs on desorption process
In order to better understand the competition of K + towards Cs + , the desorption of Cs + was studied by imposing different amounts of K + on Cs + doped K-free vermiculite (S(0)V) and K-intercalated vermiculite (S(5)V). For this purpose, the vermiculites were doped with Cs + in trace amount (6.6 10 −6 mol/ kg) and dispersed in solutions (S(0) and S(5), respectively) to which various concentrations of K + were added.
For the K-free vermiculite doped with Cs + in trace, Fig. 5 shows the mass balance of cations Ca 2+ , Mg 2+ , K + , and Cs + ( Fig. 5a and b) and the evolution of the interlayer space (Fig. 5c) as a function of the K + concentration in solution at equilibrium. When the S(0) solution is used, the result of Cs + desorption is the same as the desorption isotherm (Fig. 4a). The amount of Cs + desorbed (Fig. 5b) remains equal until 0.4 mmol/L K + introduced without any significant exchange with Ca 2+ and Mg 2+ , and the interlayer space remains fully open (i.e., d (001) reflection at 14.6 Å in Fig. 5c). The desorption of Cs + increases with increasing K + introduced between 0.7 and 5 mmol/L (Fig. 5b). The K + uptake is firstly balanced by Mg 2+ and then Ca 2+ released into solution. The interlayer spaces distances remain at 14.6 Å although K + introduced in solution represents at most 35% of the CEC of S(0)V. However, the adsorbed K + represents only 5.3% of the CEC; this explains the absence of a second broad reflection at 12.5 Å characterizing a vermiculite /mica mixed layer. A large amount of K + is then required to extract most 4% of the adsorbed Cs + from the vermiculite (i.e., 2.65 10 −7 mol/kg of Cs). This represents 3.7 to 27.9 times more than the amended K + fertilizer (250 mg/kg soil K 2 O) on Japanese soils (Kubo et al. 2015;Kato et al. 2015). Above 10 mmol/L K + introduced (Fig. 5a), K + is largely loaded onto the mineral, and Mg 2+ > Ca 2+ are released into solution, while no desorption of Cs + occurs. From 10 mmol/L K + introduced (i.e., 12.5 times less than conditioning solution for S(10)V), the structural analysis records the decrease in intensity and the spreading of the peak at 14.6 Å in favor of a peak at 12.5 Å, characteristic of a vermiculite /mica mixed layer . The latter shifts towards 12 Å and spreads towards the large angles until it reaches 10.8 Å for the highest K + concentration. The collapse of the interlayer space is thus evidenced with this structure change, as reported by Kogure et al. (2012). This collapse is then related to the adsorption of K + corresponding to 11.4% of the available sites. For the initially K + -loaded vermiculite (S(5)V), the mass balances of Ca 2+ , Mg 2+ , K + , and Cs + cations versus K + concentration in solution at equilibrium are given in Fig. 6a to c. This indicates a release of K + balanced by an uptake of Ca 2+ and Mg 2+ from solution when the contact solution initially has a lower K + concentration than the K(5) solution (Fig. 6b). In this case, the higher the K + concentration, the lower the exchange with Ca 2+ and Mg 2+ . Thus, the exchange Mg 2+ > Ca 2+ , mainly with K + , as desorption of Cs + remains unchanged with decreasing K + concentration of the solution. This results in the opening of the interlayer space of vermiculite ( Fig. 6c), as expressed by an increase in the intensity of diffraction peak at 14.6 Å and a slight decrease in the intensity of the mixed layer peak at 12.5 Å (d (001) at 0 mmol/L K + compared with that at 5 mmol/L K + ). Consequently, among the interlayer spaces, those containing exchangeable K + seem to be particularly affected by this exchange. When the S(5) solution is used, the result of Cs + desorption is the same as the desorption isotherm (Fig. 4b). When the K + concentration becomes higher than of the S(5) solution (> 5 mmol/L K), the solid adsorbs K + and releases first Mg 2+ and then Ca 2+ . Then Cs + desorption decreases until it stops between 10 and 25 mmol/L K + concentration. As long as the Cs + desorption occurs, no change in the interlayer space dimensions is recorded (see d (001) at 10 mmol/L K in Fig. 6c). When Cs + desorption no longer occurs, the characteristic mixed layer peaks at 12.5 Å, 12 Å (7.5° 2θ), and the broad reflection at 10 Å (8.6° 2θ) indicate an increasing and mixed proportion of micaceous layer. This again provides evidences for the collapse of the interlayers. Meanwhile, adsorbed K + accounts for only 2.33% of the CEC of S(5)V for a K + concentration of 25 mmol/L. Hence, K + intercalates and locks at least the vermiculite interlayer space, inhibiting the release in solution of Cs + mainly located at this structural site. The role of K + on trace Cs + desorption is evidenced by the mechanically scavenging of Cs + in the mineral structure of vermiculite and K-incorporated vermiculites. In addition, chemical competition between K + and Cs + is highlighted when the vermiculite interlayer space remains fully open and K + is introduced in the medium. Then selectivity of Cs + at trace level of vermiculite and micaceous structures is then higher than that of K + .

Mechanism of Cs sorption on vermiculite and micaceous structures
In order to establish the sorption mechanism of Cs + on vermiculite and micaceous structures in multi-cation media, the cation exchange analyses were complemented with structural analyses. Indeed, the Cs + sorption isotherms and the corresponding cation mass balances showed that cation exchange takes place differently depending on the K + occupancy in the interlayer space of conditioned vermiculites. The types of structural sites, interlayer sites, and external sites involved in the Cs + sorption are then deduced from structural analyses by XRD (Fig. 7). These latter were performed on batch slurries from Cs + desorption experiments at varying Cs + concentrations for the K-free vermiculite and the K-intercalated vermiculites (Figs. 2,4,and SI2).
Two distinct behaviors are highlighted as a function of the Cs + concentration in contact with the different clay structures, synthetized in Fig. 8. At trace Cs level (until 10 −6 mol/L of Cs), whatever the conditioned vermiculite, during Cs + adsorption and desorption, the original structure remains (Fig. 7) with only slight changes. Indeed, the (001) reflection of the K-free vermiculite expresses only a single peak around 14.6 Å. Likewise, the XRD reflections of the K-incorporated vermiculites show two peaks (Fig. 1). In the case of reversible adsorption, the (001) reflection intensity at 14.6 Å decreases without significant enhancement of the mixed layer peak at 12.5 Å (except for S(10)V), and both reflections slightly shift to larger angles. This may be related to a slight dehydration taking place in the interlayer space (Dzene et al. 2015). The (001) reflection intensities are higher at 10 −6 mol/L Cs + than at 10 −9 mol/L Cs + . This reflects the reincorporation of hydrated cations into the interlayer space subsequently to the desorption of Cs + trace. This effect appears in larger amounts at 10 −6 mol/L Cs + than at 10 −9 mol/L Cs + , confirmed by the cation mass balances, and is made possible by the sufficiently opened interlayer space up to a distance of 12.5 Å. However, the intensity and the positions of the peak do not recover the original values; hence, the cation mixture has a whole lower hydration state in the interlayer space. Therefore, the ratio between Ca 2+ to Mg 2+ in the interlayer space occupancy has changed. As the concentration of Cs + increases above 10 −5 mol/L in solution, the intensity of the vermiculite peak at 14.6 Å decreases, shifts to the large angles 2θ, and finally disappears in favor of mixed layer peak at 12.5 Å (Fig. 7) or less regardless of the initial mineral structure. This observation supports the hypothesis of a collapse of the interlayer space subsequent to the Cs + incorporation and explains the partial reversibility of Cs + sorption. The more K loaded into the vermiculite is initially, the lower the d (001) value after Cs + adsorption/ desorption and the higher the amount of released Cs (i.e., 48.8 to 11.9% for S(10)V compared to 15.8 to 1.7% for S(0) V between 10 −5 and 10 −3 mol/L Cs). Moreover, the Kd of Cs + adsorption (Fig. 2) decreases with the increase of the charged K + in the structure, and the reversibility of Cs + adsorption increases. Second, the desorption of Cs + involves the role of external sites and the interlayer space when it is sufficiently opened. In K + free media, the more Cs + is absorbed, the more the interlayer spaces collapse and the less Cs + is released. Since K + occupies the Cs + selective sorption sites and is strongly incorporated into the structure, it is not replaced by Cs + . When K + disturbs the chemical equilibrium, the interlayer distance reduces to 10 Å (Figs. 5 and 6) upon Cs + adsorption. Such a small distance is not achieved by the introduction of Cs + . Then, K + is incorporated into the interlayer space by exchange with Ca 2+ and Mg 2+ and then locks the specifically adsorbed Cs + in trace to sites with high selective and low capacity. This specific site should also be present on the basal structural site, by considering a sheet merging as shown by the disordered stacking on the XRD pattern (broad reflection peak between 12 and 10 Å)  (Motokawa et al. 2014a, b). Similarly, for Cs + (Motokawa et al. 2014a, b), the collapse of the interlayer space after K + incorporation can provide new planar adsorption sites for Cs + . This is in agreement with the higher Cs + desorption ability of K-incorporated vermiculites. Due to its high exchange capacity, the interlayer space allows for a large amount of Cs + to be loaded and remains free as long as the interlayer is sufficiently opened. The reaction scheme of the K + impact on Cs + exchange with Ca 2+ , Mg 2+ , and K + on vermiculite and K-intercalated vermiculites is displayed in Fig. 9.
The involvement of the specific structural site, usually named frayed edge site (FES) (Sawhney 1972), was investigated by artificially reducing the interlayer distance by incorporating K + during the conditioning process. By reducing the CEC of vermiculite with K + fixation, a sequence between true vermiculite and vermiculitized mica is mimicked. Consequently, it was expected to increase the occurrence of FES by leaving available cations in the space near the edge of the layer (Akemoto et al. 2020). High selectivity and low capacity sites are usually assimilated to the FES. Thus, this type of site should be evidenced with Cs + behavior at trace level. Partial reversibility of Cs + sorption in trace level (Fig. 1) is evidenced only on the most K-incorporated vermiculite (S(10)V). Consistent with the structural analysis (Fig. 7c), the reduction of the interlayer space leads to limiting the diffusion of the adsorbed trace Cs + in the dehydrated interlayer space below 12.5 Å. Thus, Cs + is adsorbed on the basal, edge, and interlayer space larger than 12.5 Å for this K-intercalated vermiculite. Furthermore, the Kd desorption is similar to the Kd adsorption for other experimental conditions. Moreover, trace Cs + adsorption decreases with K + structural incorporation, while the selectivity for Cs + is expected to increase. These results indicate that the mixed layers at 12.5 Å, created by K + incorporation, are not related to an enhancement of structural of the FES site, which is responsible of the high affinity and the irreversibility of the sorption. There is also no evidence that the FES is actually created under these operating conditions and that the FES holds the chemical reactivity of the high selectivity and low capacity site.
Sequences of Cs + exchange taking into account the involvement of structural sites can be proposed by combining the results. As discussed above, K + in solution does not interact mainly in the ion exchange with Cs + , since the matrices are equilibrated with the major cations in solution and intercalated K + is mainly not exchangeable. The chemical composition of the solutions reflects the sequence of cation exchange: at first, Mg 2+ is exchanged with Cs + , when Cs + starts to compete with Ca 2+ , the latter is released in solution. This explains the dehydration and then the reduction of the interlayer distance, in accordance with the hydration enthalpy of the cations ordered as Mg 2+ > Ca 2+ > K + > Cs + , inversely to their cation radius (Noyes 1962;Yin et al. 2017). Consequently, K + and Cs + promote the decrease of the interlayer distance and are easily dehydrated, leading to the interlayer collapse (Sawhney 1972, Okumura et al. 2018, Motokawa et al. 2014a. Some of the Cs + is mechanically locked in the interlayer space (Kikuchi et al. 2015). When the interlayer space opens, Mg 2+ and Ca 2+ occupy both interlayer space and external sites. Because the hydration energy of Mg 2+ is higher than that of Ca 2+ (Yin et al. 2017), Ca 2+ reincorporation is favored over Mg 2+ , which is consistent with the cation exchanges observed at high Cs + concentration. Once the interlayer space is closed or reduced to 12 Å, hydrated Mg 2+ appears too large to enter in shorter space (Okumura et al. 2018), which is evidenced by the cation exchanges scavenging that occurs in the desorption step, particularly for K-incorporated vermiculites. Because Cs + adsorption results in the closure of the interlayer space, only a small percentage of the adsorbed Cs + can desorb. This Cs + appears adsorbed on less affine sites. Ca 2+ easily replaces desorbed Cs + but Mg 2+ is strongly limited (Fig. 4 and SI2). The affinity of Mg 2+ is therefore lower than that of Ca 2+ for all sorption sites. In addition to K + incorporated in the structure, the interlayer spaces have shorter distances. Consequently, desorbed Cs + appears to come mainly from external sites (basal and edge site) when interlayer space is closed due to Cs + adsorption. This interpretation is supported by the study of Robin et al. (2015) on beidellite. These authors founded preferential sorption of Ca 2+ on the external sites rather than Mg 2+ . The reversibility of Cs + adsorption is also dependent on the crystal structure. K + /Cs + competition occurs for a thousand-fold higher concentration of K + than Cs + . Indeed, a large amount of K + is required to easily desorb Cs + at trace concentration (above 7 10 −4 mol/L K in excess). In addition, Cs + is locked in the interlayer space with the incorporation of K + . Consequently, K + has a high selectivity for vermiculite and micaceous structures but remains lower than that of Cs + .

Conclusions
The transfer of Cs + to plants is strongly dependent on the clay mineralogy of the soil, especially as soils are enriched in micaceous minerals and vermiculite. To understand the bioavailability of Cs radionuclides in soils, the sorption-desorption behavior of Cs + on micaceous minerals and vermiculite was studied in a chemical environment close to a natural soil. For this purpose, a sequence between vermiculite and weathered mica was produced by saturating a vermiculite with K + at various contents. The experiments involve simultaneous exchange of Cs + with Ca 2+ , Mg 2+ , and K + at thermodynamic equilibrium. By coupling sorption isotherms, cation mass balances, and structural analysis by XRD, the conceptual reaction scheme and the involvement of structural sites on the adsorption and desorption of Cs + on clay minerals in multi-cations solutions are inferred. Next, the competition of K + on the desorption of Cs + is characterized by imposing different amounts of K + on trace Cs + doped vermiculite and K-intercalated vermiculite. The concentrations of K + were selected to mimic a K + amendment in the fields, a depletion of K + in the solution due to plant uptake or weathering event, and to provide insight into the location of Cs + on the mineral structure.
Two sorption behaviors of Cs + have been demonstrated. At trace level, Cs + exchanges with Mg 2+ and Ca 2+ in a reversible adsorption process. The external sites and interlayer sites remain available during the exchange process regardless of the amount of K + incorporated in the structure. When Cs + competes with the major cations, it replaces firstly Mg 2+ and then Ca 2+ . An excess of Cs + , as well as K + , leads to a reduction and to a collapse of the interlayer space. In this case, Mg 2+ is excluded from the interlayer space and the Cs + located in this structural site is blocked and becomes non-exchangeable. The residual exchangeable Cs + is then allocated to the external sites. Less Cs + is loaded on the micaceous structure than on the vermiculite. This is due to their lower cation exchange capacity resulting from K + incorporation. In contrast, Cs + adsorption on the micaceous structure is more reversible, suggesting the dominant involvement of external sorption sites. Trace Cs + is seldom removed by K + . Cs + is strongly adsorbed on the various matrix, and a K + /Cs + ratio in solution of about one thousand is needed to remobilize Cs + . The present study provides strong evidence for the fact that interlayer sorption is responsible for the incomplete reversibility of Cs + sorption on clay minerals. In the case of soil contamination by Cs + , K + amendment leads to only a weak enhancement of Cs + desorption, more pronounced for vermiculite than for micaceous minerals. The amount of Ca 2+ and Mg 2+ in natural systems will therefore affect the bioavailability Cs + . Indeed, the multi-cation exchange study on Cs adsorption/ desorption on clay minerals highlighted that Mg 2+ favors Cs + uptake from solution by clay minerals, while Ca 2+ limits the desorption of Cs + by reducing the interlayer spatial distance. This suggests that the post-accident amendment mixing different major cations, K + , Ca 2+ , and Mg 2+ should be more effective in stabilizing Cs on soil particles and limiting Cs uptake by plants.