Development, Optimization, and Validation of Radium-226 Measurement in Oyster, a Sentinel Organism by Mass Spectrometry

: Radium-226 ( 226 Ra) measurement in living organisms, such as the American oyster ( Cras-sostrea virginica ), is an analytical challenge: the matrix complexity and the extremely low Ra levels require a puriﬁcation/preconcentration step prior to its quantiﬁcation. In this study, 5 g of dry oyster soft tissues and 1.6 g of shell were both mineralized, preconcentrated on an AG50W-X8 and a strontium-speciﬁc resin, and measured using Inductively Coupled Plasma Mass Spectrometry (ICP-MS). The volumes of digestate used in the method for both matrices were optimized to reach a high preconcentration factor without any risk of oversaturating the columns. Out of the 50 mL of digestates, 48 mL and 2.5 mL were determined as optimal volumes for soft tissues and shell, respectively. To obtain a higher preconcentration factor and a lower limit of quantiﬁcation (LOQ) for shell samples, three aliquots of 2.5 mL digestate were run on three different sets of resins and, ultimately, combined for Ra analysis using ICP-MS. LOQs of 7.7 and 0.3 fg/g (260 and 11 µ Bq/g) were achieved for the oyster shell and soft tissues, respectively. The new protocols were applied on relevant samples: oyster soft tissues and shell from New Brunswick, Canada, and different types of reference materials, such as IAEA-470, oyster soft tissue and IAEA-A-12, and animal bones. 226 Ra recovery of 105 ± 3% ( n = 6) was achieved for IAEA-A-12 (animal bones), the closest available reference material to shell with a recommended value for 226 Ra. Resin performances were investigated using 226 Ra standard solution and real samples: each set of columns could be used more than 100 times without any signiﬁcant reduction in Ra preconcentration efﬁciency. Although the method proposed and validated in this work was developed for oysters, it could easily be applied to other matrices by adjusting the volume of digestate run on the resins to avoid their oversaturation.


Introduction
Radium (Ra) is a radioactive element issued from the uranium and thorium decay series.This radionuclide has 34 isotopes, but only 4 of them ( 226 Ra, 228 Ra, 223 Ra and 225 Ra) have a sufficiently long half-life to be persistent in the environment [1].Being part of the alkaline earth elements, radium follows similar environmental, biological, and physiological pathways to calcium and magnesium [2].Ra bioaccumulation and biomagnification occur in living organisms [3], where it can trigger different bone and pulmonary diseases, and different types of cancer [4].
Assessing the risk associated with radium exposure requires, at an ecosystem scale, the quantification of this radionuclide at ultra-trace levels in multiple samples of different matrices (water, soil, sediments, plants, living organisms and other biotas).To minimize the number of samples analyzed, sentinel organisms are often targeted to reflect the level of contamination in an ecosystem [5].In the case that the excretion rate is lower than the uptake rate, bivalves bioconcentrate contaminants in their internal soft tissues and shells.Contaminants can then be either measured in soft tissues and/or shells to provide the organism with an integrated response to exposure [6].For instance, bivalves, such as mussels, are already successfully used in different programs to monitor heavy metals [7], including radium [6].The American oyster (Crassostrea virginica) is also a confirmed sentinel for heavy metals, such as mercury, zinc, and arsenic [8,9], and has been recently proposed for Ra monitoring [10,11].From an analytical standpoint, as oyster soft tissues and shells greatly differ in terms of composition, the determination of radium in both organic and inorganic matrices represents a challenge.Ra measurement for these types of samples is usually conducted using radiometric techniques, such as γ and α spectrometers [6,12], or mass spectrometry techniques, such as Inductively Coupled Plasma Mass Spectrometry (ICP-MS) [13,14].Each technique category has its advantages and flaws.Radiometric instruments measure the activity of the specific α, β or γ emissions of the desired isotope.All Ra isotopes can be simultaneously measured using these techniques; however, this requires lengthy ingrowth periods and/or prolonged counting times [15,16].For mass spectrometry instruments, the elements are measured by separating them based on their mass/charge ratio and by counting their ionized atoms.If the analysis is restricted to a single isotope ( 226 Ra) in most samples, mass spectrometer instruments present the advantage of having a shorter analysis time and usually requiring a smaller sample size [15].
For the analysis of radionuclides in solid samples like oyster soft tissues or shells, the matrix decomposition is usually achieved via dry-ashing, fusion, or acid digestion techniques [17].For dry-ashing, samples are heated at a high temperature, up to 800 • C, to decompose all the organic materials.In fusion, samples are heated with fusion substrates until the mixture is molten and clear, then it is dissolved in nitric acid (HNO 3 ) or hydrochloric acid (HCl) [17].For acid digestion, strong acids are combined with the sample and heated to ensure its complete dissolution.Depending on the matrix itself, different acids and mixtures can be used.Amongst them, nitric acid is used for ores [18]; aqua regia is used for fish [19]; and a mixture of fluoric, nitric and hydrochloric acids is used for sediment [20].Once the matrix is mineralized, radium is extracted and preconcentrated to enable its quantification using either radiometric or mass spectrometry instruments.Coprecipitation, adsorption, chromatography or other separation techniques are usually involved.In the case of mass spectrometry methods using ICP-MS in particular, a combination of multiple purification/preconcentration methods can be followed to enhance the elimination of potential interfering elements [21][22][23].For instance, two ion exchange resins, such as AG50W-X8 and Sr specific resins, are combined to measure 226 Ra in natural waters using ICP-MS [21].The AG50W-X8 resin eliminates all major elements, such as calcium, magnesium, sodium, and strontium, that interfere with the plasma of the ICP-MS, while the Sr-specific resin eliminates the rest of the strontium and barium, preventing the onset of a potential polyatomic interference at a m/z of 226 [21].
The performance of the analytical measurements greatly varies depending on the sample matrix, the sample preparation, as well as the instrument used.For radiometric instruments, a limit of detection (LOD) of 55 fg/g (2.0 Bq/kg) was achieved for 226 Ra in edible parts of fish [24], while a 226 Ra concentration as low as 0.20 fg/g (0.0075 Bq/kg) was reported in blue mussel (Mytilus edulis) soft tissues [25].Using a mass spectrometer, LODs of 2.8 fg/g (0.10 Bq/kg) and 0.40 fg/g (0.015 Bq/kg) were measured, respectively, in sediment [26], a matrix with a high inorganic content, and in human brain [27], a matrix enriched with organic compounds.Even with an absolute method for Ra measurement, a reoptimization and validation of the method is required for each matrix used.In this study, we have focused on developing a new approach to measure 226 Ra in two contrasting matrices (oyster shell and soft tissues) using ICP-MS based on AG50W-X8 and Sr-specific resin exchange capacities.If radiometric methods already existed for these two matrices [10,11], we do believe that we could achieve comparable, if not better, analytical performances using ICP-MS without a lengthy ingrowth period and higher sample output.The method was optimized and validated for both mineral and organic oyster matrices (shell and soft tissues), enabling its possible extrapolation to other types of matrices.

Materials
The radium calibration and spike solutions used in this study were prepared from a standard reference material, SRM 4966A, that was provided by the National Institute of Standards and Technology (NIST, Gaithersburg, Germany: [ 226 Ra] = 7.867 ± 0.101 ng/g or 287.6 ± 3.7 Bq/g).Other metal calibration solutions were prepared from single or multi-elements stock solutions purchased from PlasmaCAL or Inorganic Ventures with concentrations ranging from 10 to 1000 mg/L.The prepacked strontium (Sr, 50-100 µm)-specific resin used for the preconcentration of radium was purchased from Eichrom Technologies (Lisle, IL, USA).The AG50W-X8 resin columns, also purchased from Eichrom Technologies (Lisle, IL, USA), were prepared in the laboratory with 1 g of the resin (100-200 mesh) placed in a plastic cartridge.
All acids (nitric acid, HNO 3 , and hydrochloric acid, HCl) used were purchased from Fischer Scientific (Ottawa, ON, Canada).In addition, 18.2-ΩM-grade water was obtained using a Milli-Q purification system from EMD-Millipore (Darmstadt, Germany).A 25 mm syringe filter with a 0.45 µm cellulose acetate membrane was acquired from VWR International North America.Ultra-high-purity gaseous argon and helium for the ICP-MS were purchased from Air Liquide Canada Inc. (Montréal, QC, Canada).

Sample Preparation
Oysters were obtained at a local grocery store in New Brunswick, Canada.The soft tissues obtained from the shucked oysters were dried in an oven at 80 • C for 48 h, and the shells were ground using a ceramic mortar.First, we proceeded with the mineralization of a large sample size of each matrix, soft tissue and shell in order to provide enough material for all testing required to validate and optimize the proposed method: 50 g of dry soft tissues (~36 oysters) and 16 g of shell were mineralized separately with 150 mL of a mix of nitric acid and hydrochloric acid (2 HNO 3 :1 HCl) at a temperature of 80 • C for 24 h.The acid digestion was based on method 200.2 of EPA, USA [28].At the time of digestion, Milli-Q water was added to both samples to reach a final volume of 500 mL.The digestates were then filtered using a 0.45 µm cellulose acetate membrane filter.Afterwards, a 50 mL sub-sample from each matrix was preserved for trace metal determination including the ambient 226 Ra level, while the remaining 450 mL was spiked with 226 Ra at a 2 pg/L concentration.Finally, the resulting spiked digestates were evaporated to dryness and reconstituted to their initial volume of 450 mL using 0.1 M HNO 3 .
For the two matrix digestates, loadings of varying volumes were tested on the AG50W-X8 resin and the Sr-specific resin [21].The digestate was first loaded onto the AG50W-X8 resin at a controlled flow rate of 1 mL min −1 using a peristaltic pump (Elemental Scientific Inc., Omaha, NE, USA), followed by 15 mL of 2.5 M HCl.The HCl solution allowed the major interfering elements, such as Na, Ca, Mg and Sr, to be eluted and discarded.Afterward, the AG50W-X8 and Sr-specific resin were connected in series to pass 10 mL of a 4 M HNO 3 solution to elute Ra from the cationic exchange resin, separating it from Ba and the remaining Sr. Finally, 10 mL of a 3 M HNO 3 solution was run only on the Sr-specific resin to ensure the complete recovery of Ra.Both 10 mL of the 4 M and 3 M HNO 3 solutions were collected, evaporated to dryness, and reconstituted in 4 mL of a 0.1 M HNO 3 solution for ICP-MS analyses.Prior to the first use and between each use, both resins were prewashed and conditioned.Cation exchange resins were conditioned in a 3-step protocol with 10 mL of 4 M HNO 3 , 5 mL of Milli-Q H 2 O and 10 mL of 1.7 M HCl successively.Sr-specific resins were rinsed with 10 mL of Milli-Q H 2 O.
In our study, varying volumes of the digestates of both the soft tissues and shell samples were tested using the preconcentration method.Volumes from 10 to 120 mL for soft tissues and from 1 to 50 mL for shell samples were, respectively, tested . 226Ra concentrations and the saturation of the cationic exchange resin were determined using ICP-MS.

Analysis
Measurements were performed using the iCAP-Q ICP-MS (Thermo Scientific, Bremen, Germany) interfaced with the CETAC Technologies ASX-520 autosampler (Omaha, NE, USA). 226Ra concentrations were measured in kinetic energy discrimination (KED) mode with indium (In) at 1 µg/L being used as an internal standard [21]. 226Ra spike recoveries were calculated using the following formula: 226 Ra spike recovery (%) = 226 Ra total − 226 Ra ambient V reconstituted solution In total, 18 elements (Li, Na, Mg, Al, K, Ca, V, Mn, Fe, Co, Ni, Cu, Zn, As, Rb, Sr, Ba, and U) were quantified in both samples before and after the preconcentration step.Different dilution factors were applied to measure these elements in the oyster shell and soft tissue digestates.Amounts of divalent and trivalent cations were used to determine the saturation of the cationic exchange resin considering its capacity of 2 125 micro-equivalents per gram [29].Then, 5 µg/L of an In solution at 2% HNO 3 was used as the internal standard for the analysis of the target metals.

Method Validation and Application
To validate our analytical procedure, two reference materials purchased from the International Atomic Energy Agency (Vienna, Austria) were analyzed: IAEA-470 (oyster) and IAEA-A-12 (animal bones).IAEA-A-12 is a natural composite material whose main components are organic and mineral compounds, such as phosphates and carbonates [30].The latter might not be fully representative of oyster shell, whose principal mineral is calcite, but it remains the closest available reference material from IAEA with a recommended value for 226 Ra of 142 fg/g (5.2 mBq/g).While IAEA-470 is not certified for Ra, it is for several other elements (Na, Mg, Ca, Zn, V, Mn, Fe, Cu, As, Rb, Sr and Pb), as it was used to assess the mineralization efficiency of our method for the oyster soft tissues.In total, 3 g of IAEA-470 and 1.6 g of IAEA-A-12 were mineralized in 15 mL of aqua regia solution at 80 • C for 24 h, diluted to 50 mL, evaporated to dryness and reconstituted in HNO 3 0.1 M. Optimized volumes of 48 mL and 2.5 mL (see Results and Discussion) of the soft tissue and shell digestates, respectively, were then preconcentrated on the resin and analyzed using ICP-MS.The same procedure was applied to 5 g of oyster soft tissues and 1.6 g of shell from two different regions in Atlantic Canada.The natural variability in the 226 Ra in the oyster shell was investigated by analyzing 3 sub-samples of the same shell.

Mineralization Efficiency
The acidic mineralization efficiency of the oyster soft tissues, O-SoftT hereafter, was evaluated by measuring the concentrations of 12 certified elements (Na, Mg, Ca, Zn, V, Mn, Fe, Cu, As, Rb, Sr and Pb) from three digestion replicates of the IAEA-470 reference material.The percentage recoveries ranged between 80% and 113%, with a mean value of 95 ± 9% (n = 12, Table 1); this suggests that the method used in this study yields the complete mineralization of the soft tissues.In the oyster shell, O-Shell hereafter, the calcium level was used as a proxy for the mineralization efficiency.Assuming O-Shell is mostly composed of calcite (100% of CaCO 3 ), a theoretical concentration of Ca could be calculated in the digestate knowing the initial weight sample of the mineralized shell.With a recovery percentage of 99 ± 1% (n = 3), the mineralization of the shell using the mix of nitric and hydrochloric acid is also considered complete.

Separation Efficiency
The matrices originating from the O-SoftT and O-Shell digestates differed significantly in terms of composition (Table 2).High levels of Na, K, Mg, Ca and Zn were monitored in the O-SoftT digestate, while an extremely high concentration of Ca was measured in the O-Shell digestate.As expected, both matrices presented high salt concentrations, which can cause a significant reduction in the ICP-MS sensitivity [31].The use of an internal standard can counterbalance the instrumental drift [20] and overcome the matrix deviation in the ICP-MS signal.The presence of other elements, even at lower concentrations (i.e., Ba and Sr), can generate polyatomic interferences at mass 226, thus jeopardizing the accuracy of the ICP-MS direct analysis [32].Separation and preconcentration steps are therefore required.Once the preconcentration procedure was applied on both digestate samples, all major salts were fully eliminated (Table 2).As a result, the reconstituted 0.1 M HNO 3 solution matrix was simpler and did not affect the instrumental response; except for a classical instrumental drift, the variation in the internal standard from one sample to another remained limited through the whole analysis run (less than 10%).Minor constituents-except for Al-including potential interfering elements (Pb, Ba, Sr. ..) [20] were also efficiently discarded during the preconcentration procedure.Although 25% of the initial Ba content remained in the final solution, the subsequent concentrations of Ba (~90 µg/L) and Sr (<1.0 µg/L) were low enough to prevent any significant polyatomic interferences in the 226 Ra ICP-MS-measured signals [20].

Ra Spike Recovery
If major salts and potential interfering elements were successfully separated from 226 Ra during the preconcentration procedure, the quantification of radium recovery was needed to validate the proposed method.For both matrices, varying volumes of the spiked digestate were tested.Volumes ranging from 10 to 120 mL of O-SoftT digestate were treated on the two selected resins, as described in the Method section.For all tested volumes of O-SoftT, 226 Ra spike recoveries were complete (105 ± 2%); meanwhile, the saturation of AG50W-X8 resin was only reached for the assay using 120 mL of digestate (Figure 1).In this assay, the Ra spike was fully recovered (108 ± 6%) even though the resin capacity was overstepped by 23%.This fact confirmed the stronger affinity of AG50W-X8 resin for Ra compared to other divalent and trivalent cations [33].
The saturation of resins is ultimately driven by the mass of mineralized O-SoftT and the digestate volume preconcentrated.As the capacity of 1 g of AG50W-X8 resin is reached with 120 mL out of the 500 mL of digestate obtained from a 50 g sample, we can estimate that the same amount of resin can tolerate a maximum of 12 g of dry O-SoftT prior to saturation.For a typical digestion of 5 g of dry O-SoftT (~6 oysters) in 50 mL, the whole solution, after being evaporated and reconstituted in 0.1 M HNO 3 , could be preconcentrated on AG50W-X8 and Sr-specific resins without any significant Ra loss.From a practical standpoint, as other elements might need to be analyzed from the same digestate, a lower volume might be available for Ra preconcentration.Using 48 mL of the digestate and a final volume of 4 mL, a high preconcentration factor of 12 is achieved without worrying about over-saturating the cationic exchange resin.The analytical sensitivity could be virtually improved by a factor of 2.4 by digesting 12 g of dry O-SoftT.However, as the composition of O-SoftT may vary slightly from one sample to another, we do recommend measuring the level of divalent and trivalent cations in the digestate to ensure that the resin capacity will not be exceeded.Pooling several oysters (up to 12 g of dry soft tissues) in a single sample offers the advantage of integrating the inter-individual variability in living organisms with Ra exposure and a better figure of the true aquatic ecosystem contamination [34].
For O-Shell samples, volumes of 1 to 50 mL of digestate were tested (Figure 1b).Due to the extremely high concentration of Ca (>10 g/L), AG50W-X8 resin was saturated with a volume of O-Shell digested solution as small as 4 mL. 226Ra spike recovery was then proven uncomplete (90 ± 7%) and declined even more rapidly once larger volumes were loaded onto the resins. 226Ra spike was fully recovered with an aliquot of at most 2.5 mL (105 ± 15%), while the AG50W-X8 resin remained unsaturated (72 ± 1%).Out of 16 g of sample digested into 500 mL, 2.5 mL could be treated without impacting Ra recovery, meaning that 1 g of AG50W-X8 resin can sustain a maximum load of 80 mg of shell to ensure a full Ra recovery.Compared to O-SoftT, the preconcentration factor of Ra in the O-Shell samples was significantly reduced, leading to a much higher LOQ.For the rest of our analyses, a 1.6 g sample size was selected, digested in 50 mL, and 2.5 mL of the digestate reconstituted solution was run through the ion exchange columns.However, using a lower mass and the full volume of digestate for preconcentration makes possible the determination of any variation in the Ra concentration in O-Shell.Like trees, oyster shells have rings corresponding to their seasonal growth: a slow growth and a dense age ring is observed in cold winters, while a faster growth occurs in warm summers [35].Oysters could be therefore transposed from their natural, uncontaminated habitat to a potentially contaminated one for a full season.A comparison of the radium content in the newer ring and the older ones should allow the determination of the Ra excess accumulated in oyster shells in relation to exposure in the new ecosystem.The saturation of resins is ultimately driven by the mass of mineralized O-SoftT and the digestate volume preconcentrated.As the capacity of 1 g of AG50W-X8 resin is reached with 120 mL out of the 500 mL of digestate obtained from a 50 g sample, we can estimate that the same amount of resin can tolerate a maximum of 12 g of dry O-SoftT prior to saturation.For a typical digestion of 5 g of dry O-SoftT (~6 oysters) in 50 mL, the whole solution, after being evaporated and reconstituted in 0.1 M HNO3, could be preconcentrated on AG50W-X8 and Sr-specific resins without any significant Ra loss.From a practical standpoint, as other elements might need to be analyzed from the same digestate, a Based on 80 analytical blank signals and 10 times their standard deviation, the iCAP-Q instrumental LOQ was estimated at 0.9 pg/L for 226 Ra.However, this LOQ could be improved using a better introduction system, such as the APEX Q 20 or a more sensitive instrument [21].Procedural blanks were run through the resins, preconcentrated and analyzed using ICP-MS.Recorded signals were not statistically different from their analytical blank counterparts (paired t test, p < 0.05).The contribution of all chemical reagents used in our preconcentration procedure (including the resins and the acidic solutions) to the 226 Ra overall measured signal remained extremely limited.Using a fictive mass and the appropriate volume (1.6 g and 2.5 mL for O-Shell vs. 5 g and 48 mL for dry O-SoftT), procedural blank LOQs were converted into mass concentrations (Table 3).Due to the lower mass mineralized and the lower volume of solution preconcentrated on resins, the 226 Ra LOQ for O-Shell (23 fg/g, n = 7) is 76 folds higher than the LOQ for dry soft tissues (0.3 fg/g, n = 9).Table 3. 226 Ra concentration measured in O-Shell and O-SoftT from two New Brunswick oyster farms, as well as the corresponding methodological limit of quantification (LOQ).

Oyster Samples
Oysters from two farms located in New Brunswick (Canada) were processed to determine their natural 226 Ra levels (Table 3).The 226 Ra concentration in O-SoftT ranged from 3.7 ± 0.6 to 20.9 ± 0.4 fg/g, well above the LOQ determined earlier (0.3 fg/g), but still below the 63 ± 11 fg/g reported in a bay of the Gulf of Mexico [10].These results demonstrated the potential of our analytical procedure to measure 226 Ra in oyster soft tissues even at low, natural levels.
In O-Shell, the 226 Ra concentrations were above the LOQ (23 fg/g), only for one of the two samples (Table 3).To improve our overall analytical sensitivity, we decided to run three aliquots of the same O-Shell digestate on three different sets of columns.Eluates from each set of columns were combined, evaporated to dryness, and reconstituted in 3 mL of 0.1 M HNO 3 .Such procedural modification increased theoretically the Ra preconcentration factor by three and decreased the LOQ accordingly from 23 to 7.7 fg/g.A new measurement of 226 Ra in the shell from oyster 2 was carried out and a concentration of 14.4 fg/g was determined.The relative standard deviation associated with the 226 Ra ICP-MS measurement was less than 8%.Two other subsamples from the oyster 2 shell were digested and analyzed following the improved procedure.An average of 15.1 ± 2.1 fg/g was measured for the three subsamples.If the overall relative standard deviation increased to 14% (n = 3), it included all the steps from the analytical procedure, as well as the potential heterogeneity in 226 Ra shell distribution.Even considering such uncertainty, our analytical method successfully discriminated 226 Ra levels in shell from the two oyster samples (t-paired test, p < 0.05).Both concentrations were, however, lower than the ones (117 ± 21 fg/g) reported in the Gulf of Mexico [11].These results demonstrated the ability of our analytical procedure to measure 226 Ra in O-Shell even at natural levels.

Method Validation and Figures of Merit
To further validate the proposed improved method for O-Shell, six replicates of IAEA-A-12 were mineralized.For each sample, three aliquots of 2.5 mL were preconcentrated on three cationic exchange columns sets.HNO 3 eluting solutions were combined, evaporated to dryness, reconstituted in 0.1 M HNO 3 and analyzed using ICP-MS; the average concentration measured for 226 Ra was 149.0 ± 3.4 fg/g, within the 95% confidence interval for the recommended value (142.2 fg/g: 120.4 to 183.3 fg/g [36]).A 105 ± 3% 226 Ra recovery was determined for this reference material, confirming the accuracy of our analytical procedure.The results were also highly reproducible for the six tested samples of IAEA-A-12 (less than 3% of standard deviation for 226 Ra).
IAEA-470 was analyzed in duplicate for 226 Ra.An average of 2.5 ± 0.2 fg/g was measured for the Korean O-SoftT.With no certified nor recommended value for 226 Ra in this reference material, the analysis accuracy could not be further assessed.However, the measured concentration could still be used as an indicative one by other radioanalytical laboratories or/and serve as the first step of an intercalibration exercise.
In terms of separation efficiency and consistency, three sets of columns were used more than 110 times, running a 15 pg/L standard solution every three to five O-Shell or O-SoftT samples.The 226 Ra recovery from these solutions was always above 90%, with a mean value of 101 ± 9% (n = 330; Figure 2).Despite the application of a digestate-enriched matrix with organic matter for O-SoftT and calcium for O-Shell, 226 Ra separation using the AG50W-X8 and Sr resins remained efficient and consistent over time.The regeneration capacity of both resins is therefore demonstrated with real samples, which constitutes a significant benefit to consider in any preconcentration method.In terms of analytical capacities, Ra preconcentration on columns remains the limiting step.In our laboratory, five samples could be run through a set of columns within a single day.After 3 days, using three sets of columns, 45 samples including QA/QC (blanks, reference materials, standard solution for column performance) could be collected.A fourth day is necessary for sample evaporation and reconstitution in HNO3 0.  In terms of analytical capacities, Ra preconcentration on columns remains the limiting step.In our laboratory, five samples could be run through a set of columns within a single day.After 3 days, using three sets of columns, 45 samples including QA/QC (blanks, reference materials, standard solution for column performance) could be collected.A fourth day is necessary for sample evaporation and reconstitution in HNO 3 0.1 M, and a final day is necessary for 226 Ra analysis using ICP-MS.This results in the analysis of 36 O-SoftT or 12 O-Shell samples typically taking a week with the improved method.These numbers could be increased by using additional sets of columns or by automatizing the preconcentration method [37].

Conclusions and Perspectives
A new method of 226 Ra analysis in oyster soft tissues and shell was developed, optimized, and validated.In this study, 5 g (dry weight) of O-SoftT or 1.6 g of O-Shell were mineralized in a 15 mL mix of nitric and hydrochloric acid (2 HNO 3 :1 HCl).Once the strong acidic matrix was eliminated via simple evaporation and replaced by 0.1 M HNO 3 , 100% of the O-SoftT digestate or 5% of the O-Shell analog were successfully preconcentrated on 1 g of AG50W-X8 resin.To improve the method's sensitivity, the same shell digestate could be run on three different sets of columns and then combined prior to analysis.Using procedural blanks, limits of quantification of 0.3 fg/g (11 µBq/g) and 7.7 fg/g (260 µBq/g) were determined for dry O-SoftT and O-Shell, respectively.Full recoveries of 226 Ra were achieved from the spiked digestate or certified reference material, IAEA-A12, confirming the accuracy of the analytical method.The same set of resin columns could be used more than 100 times without any significant loss of performance.Using three sets of columns, 36 O-SoftT samples or 12 O-Shell samples could be processed within a single week.The analytical performances achieved here in terms of both capacity and sensitivity make this method suitable for most monitoring programs devoted to 226 Ra using bivalves.For this purpose, the pooling of multiple oyster soft tissues may provide a better representation of the population's average concentration, while minimizing the number of samples to be analyzed.The mass of shell in the sample could be reduced (and the corresponding digestate volume preconcentrated on resins increased) to focus on specific growth rings.In both cases, the levels of divalent and trivalent cations should be measured to determine the optimal volume of digestate to be preconcentrated on a 1 g AG50W-X8 resin with no risk of saturation.Otherwise, a higher mass of AG50W-X8 resin could be used, but this would also require a full re-optimization of the whole separation procedure.The preconcentration efficiency could also be monitored in this case for each sample using an isotopic tracer, such as 228 Ra.Note that the preconcentration procedure proposed in this work could be coupled to any radiometric instrument to measure all radium isotopes at relevant environmental levels.Finally, the method designed here for oysters could be easily expanded to other sentinel organisms and matrices; the sample mass and digestate volume run on columns should be simply adjusted to ensure that resin oversaturation is not reached.

Separations 2023 , 13 Figure 1 .
Figure 1.Radium recovery and saturation of AG50W-X8 resin for each volume of tested digestates of both (a) O-SoftT and (b) O-Shell.The error bars represent one standard deviation of the results based on ICP-MS instrumental measurement uncertainty.

Figure 1 .
Figure 1.Radium recovery and saturation of AG50W-X8 resin for each volume of tested digestates of both (a) O-SoftT and (b) O-Shell.The error bars represent one standard deviation of the results based on ICP-MS instrumental measurement uncertainty.

Figure 2 .
Figure 2. Chart control for radium recovery of a 15 pg/L standard solution in three sets of columns.The separation and preconcentration capacities of the resins were monitored every three to five samples (110 samples were run in total for each set of columns).The grey dashed line represents the mean value, while the solid grey and blue lines indicate the boundaries of two and three standard deviation intervals, respectively.The error bars represent one standard deviation of226 Ra recovery based on ICP-MS instrumental measurement uncertainty.

Figure 2 .
Figure 2. Chart control for radium recovery of a 15 pg/L standard solution in three sets of columns.The separation and preconcentration capacities of the resins were monitored every three to five samples (110 samples were run in total for each set of columns).The grey dashed line represents the mean value, while the solid grey and blue lines indicate the boundaries of two and three standard deviation intervals, respectively.The error bars represent one standard deviation of226 Ra recovery based on ICP-MS instrumental measurement uncertainty.

Table 2 .
Metal concentration in digestates of O-SoftT and O-Shell prior to their preconcentration in AG50W-X8 and Sr-specific resins, and % of elimination after resin separation.