Bioavailable actinide fluxes to the Irish Sea from Sellafield-labelled sediments

Nuclear discharges to the oceans have given rise to significant accumulations of radionuclides in sediments which can later remobilise back into the water column. A continuing supply of radionuclides to aquatic organisms and the human food chain can therefore exist, despite the absence of ongoing nuclear discharges. Radionuclide remobilisation from sediment is consequently a critical component of the modelled radiation dose to the public. However, radionuclide remobilisation fluxes from contaminated marine sediments have never been quantitatively determined in-situ to provide a valid assessment of the issue. Here, we combine recent advances in the Diffusive Gradients in Thin Films (DGT) sampling technique with ultrasensitive measurement by accelerator mass spectrometry (AMS) to calculate the remobilisation fluxes of plutonium, americium and uranium isotopes from the Esk Estuary sediments (UK), which have accumulated historic discharges from the Sellafield nuclear reprocessing facility. Isotopic evidence indicates the local biota are accumulating remobilised plutonium and demonstrates the DGT technique as a valid bioavailability proxy, which more accurately reflects the elemental fractionation of the actinides in the biota than traditional bulk water sampling. These results provide a fundamental evaluation of the re-incorporation of bioavailable actinides into the biosphere from sediment reservoirs. We therefore anticipate this work will provide a tool and point of reference to improve radiation dose modelling and contribute insight for other environmental projects, such as the near-surface and deep disposal of nuclear waste.


Introduction
Nuclear activities have routinely released radionuclides to the oceans which accumulate within marine organisms. This presents a radiotoxicity and biotoxicity hazard to a range of biota, including human consumers. Radionuclides in the seawater, depending on their speciation and particle reactivity, can also accumulate over time in sediment bodies via adsorption and other removal processes. Significant radionuclide accumulations have contaminated sediments near to sites of authorised or accidental nuclear discharges, including Fukushima (Japan) (Mouri et al., 2014;Taniguchi et al., 2019;Takata et al., 2015), La Hague (France) (Cundy et al., 2002) and Sellafield (UK) (Morris et al., 2000;McDonald et al., 2001;Hunt et al., 2013). Such accumulation may be reversible over longer time periods however, and sediment-associated radionuclides have been observed to remobilise back into the surrounding water in various settings (Morris et al., 2000;Hunt et al., 2013;Mitchell et al., 2001;Caborn et al., 2016). Several mechanisms have been proposed for this, including chemical mobilisation processes and bioturbation (Kershaw et al., 1983). Therefore, a reduction or cessation in nuclear discharges does not guarantee the removal of a significant supply of radionuclides to local waters, where they can again become available for uptake by biota. This effect has been observed near to the UK's Sellafield facility for example, where plutonium concentrations in seaweed and shellfish changed very little in the years following a very steep reduction in nuclear discharges to the Irish Sea (Ryan et al., 1999). Radiation dose modelling must therefore account for radionuclide remobilisation in order to be fully comprehensive. This is critical around sites such as Sellafield, where the sediments of the Cumbrian coastal region contain significant radionuclide loadings (>20 Bq g − 1 for some actinides). This presents long-term radiological risk implications, including from the in-growth of 241 Am from accumulated 241 Pu (Day and Cross, 1981).
Despite this, no time-integrated and in-situ method currently exists which can determine the remobilisation fluxes of actinides from marine sediments. The need for such a technique is highlighted by large temporal and spatial inconsistencies in sediment-water partition coefficients (k d ) for radionuclides (Agency, 2004). These k d values have been used to indicate radionuclide remobilisation in the Cumbrian coastal region (Hunt et al., 2013;Mitchell et al., 2001) alongside ex-situ partitioning and desorption experiments (Takata et al., 2015), but cannot validly replicate in-situ geochemical conditions and their variability.
The Diffusive Gradients in Thin Films (DGT) technique is a wellestablished passive sampling method and bioavailability proxy which has previously been employed to calculate remobilisation fluxes for various metals from sediment (Harper et al., 1998;Wu et al., 2015;Norgbey et al., 2020;Zhang et al., 1995;Gao et al., 2016;Zhang et al., 2001). When deployed in sediments, DGT samplers function as a sink for their target analyte(s), mimicking the dynamic action of a plant root as they induce a flux of labile (bioavailable) species in-situ from porewater to their resin-gel (Zhang et al., 2001;Davison, 2016). Recently developed resin-gels (Chaplin et al., 2021(Chaplin et al., , 2022 have made possible the measurement of actinides in the marine environment by DGT. In this work, we combined deployments of DGT samplers with ultrasensitive measurement by accelerator mass spectrometry (AMS) (Christl et al., 2013), to calculate remobilisation fluxes of plutonium, americium and uranium isotopes from the Esk Estuary sediments (UK) to the Irish sea ( Fig. 1). Isotopes ( 239-241 Pu, 241 Am and 236 U) captured by the DGT samplers and in local biota and seawater were measured at the compact, actinide-optimised ETH-Zürich AMS facility TANDY (Christl et al., 2013), allowing for the detection of isotopes at the sub-femtogram level. Analysis of isotopic signatures throughout the DGT-sampled sediment profiles (Figs. 2 and 3), local biota ( Fig. 3) and seawater ( Fig. 3) allowed us to constrain the origins of the contamination throughout the Esk Estuary and assess the ability and validity of the DGT technique as a proxy for actinide bioavailability determination.

Site selection
We selected the Ravenglass Estuary for the deployment of DGT samplers due to its well documented and significant sedimentary inventory of legacy Sellafield-derived actinides. Two sampling campaigns were undertaken. Campaign one was undertaken between 24th August 2020 and 7th September 2020, and campaign two was undertaken between 24th May 2021 and 7th June 2021. During campaign one, DGT Research® R-SP sediment probe housings containing the IIP-Y 3+ and KMS-1 configurations (Chaplin et al., 2021) were deployed, at sites 1-4 (Methods Fig. 1, Ordnance Survey (OS) grid references and elevations are given in Methods Table 1). During campaign two, custom in-house "DGT-55" sediment probe housings containing the IIP-Y 3+ and KMS-1 configurations were deployed at sites 1 and 5, sampling deeper up to 55 cm depth. Sites 1-4 are ungrazed marshland, with gradation shown between the lower to upper marsh environments from the River Esk main channel. The Upper Railway Marsh (site 1) is a mixed high marsh sward with sea aster, sea arrow grass, sea rush (Juncos maritimus), salicornia (Salicornia europaea), sea plantain and sea purslane (Sesuvium portulacastrum). The Lower Railway Marsh (site 2) is characterised by sea aster (Tripolium pannonicum), sea plantain (Plantago maritima) and sea arrowgrass (Triglochin maritima). The Brighouse Marsh (Upper -site 3, and Lowersite 4) is a consistent mixed high marsh sward including sea aster, sea lavender (Limonium vulgare) and sea plantain in the upper region, with additional sea purslane and sea aster more common in the lower region. Site 5 is an intertidal sandflat near to the mouth of the Esk Estuary which exposes a lagoon at low spring tide.

DGT sampler manufacture
Diffusive gels were purchased from DGT Research® (Lancaster, UK). The KMS-1 resin was synthesised according to the hydrothermal method reported in point C of the Supporting Information of Manos et al. (2008) IIP-Y 3+ was synthesised in a modified procedure, following that of Chauvin et al. (2006) but with the ligand L 1 instead synthesised according to Scheme 1 of Chaplin et al. (2021) The KMS-1 and IIP-Y 3+ DGT resin-gels were synthesised as described by Chaplin et al. (2021) The internal resin distribution in the synthesised resin-gel was verified to be homogenous before being assembled into the DGT sampler housing. DGT R-SP sediment probe housings were purchased from DGT Research Ltd (Lancaster, UK) and used to house the KMS-1 and IIP-Y 3+ configurations during the first sampling campaign. Our custom DGT-55 sediment probes were designed and manufactured in the workshop of Lausanne University Hospital's Institute of Radiation Physics (CHU-V-IRA) and were used to house the KMS-1 and IIP-Y 3+ configurations during campaign two. 105 cm 2 surface area DGT samplers as described by Cusnir et al. (2015) were also used to house KMS-1 and IIP-Y 3+ sampling configurations for seawater deployments. These were encased  Table 1 for coordinates and elevation.  Chaplin et al. in protective stainless-steel holders and deployed facing upwards at roughly a 45 • angle, secured to poles in the seabed. Prior to sampling, DGT samplers were stored and transported in sealed plastic bags with several drops of 0.3 M NaCl inside the bags to maintain moisture for a total period of no more than 3 weeks.

DGT and porewater peeper sampling
DGT samplers were unsealed from their packaging in the field. Care was taken not to step within at least 50 cm of the point of insertion into the sediment, allowing the natural sediment porosity and pre-existing porewater pathways to remain undisturbed. During campaign one, DGT Research® (Lancaster, UK) R-SP sediment probes were deployed into sediment at least 30 cm apart from each other. During campaign two, custom in-house DGT-55 sediment probes were inserted into sediment at least 30 cm apart from each other and at least 2 m from the exact deployment site of the samplers during campaign one. 55 cm long porewater peepers, with 20 mL compartments at 2.5 cm intervals and sealed with 0.45 µm polyethersulphone filter membrane, were deployed adjacent to DGT samplers during campaign two. This allowed for the porewater concentration to be determined by diffusion at equilibrium insitu, which is particularly important for accurate readings of U (Toole et al., 1984). All DGT samplers and porewater peepers were left in-situ for 14 days during both campaigns. This depleted the analyte from the porewater, inducing the desorption of labile sediment-adsorbed species into the porewater to maintain the equilibrium defined by the k d . An extended deployment period (> 1-3 days) in sediment therefore ensures the flux of the analyte to the DGT sampler is exclusively diffusion-controlled by depletion from the sediment. Additionally, this minimises the impact caused by the initial disturbance of inserting the sampler by allowing the sediment profile to re-establish via redox buffering mechanisms (Davison, 2016).

Sediment core sampling
Sediment cores were extracted at the same time as the DGT Research® R-SP DGT sediment probe deployments on 24th August 2020, by hammering a length of PVC pipe into the marshland around 50 cm from the DGT samplers at each site. Cores were sectioned into slices of around 2 cm thickness, freeze-dried, ground, and the powder was measured in HPGe gamma spectrometers to obtain 137 Cs and 241 Am bulk concentration data. Suitable aliquots (5-300 mg) of the powdered core fractions were then immersed in 20 mL of concentrated HNO 3 and dispersed using ultrasound. After at least 24 h of leaching at a temperature close to reflux, the powder was filtered using a 0.45 µM syringe filter and prepared for Pu separations (as described in the following sections) to determine the 239+240 Pu bulk concentrations by alpha spectrometry, using 450 mm 2 passivated implanted planar silicon detectors and Alpha-Analyst (Canberra, France) software.

Sediment core age model
Sediment cores were dated according to the depth of the activity concentration maxima of 137 Cs, 239+240 Pu and 241 Am in the core profile. The years of maximum discharge from the Sellafield facility were 1975 for 137 Cs, 1973 for 239+240 Pu and 1974 for 241 Am (retrieved from Gray et al. 1995). The surface of the sediment core was considered as the date of retrieval (24 August 2020). The depth of the maximum activity concentration of the radioanalyte was considered as 01 January for the year of maximum discharge from Sellafield. A lag/ transit time of 1-3 months has previously been observed between the discharge of some radionuclides ( 106 Ru, 144 Ce, 95 Zr and 95 Nb) from Sellafield and their deposition in Esk Estuary sediments (Hetherington and Jefferies, 1974). This was therefore considered negligible at the resolution of our core profiles and not included in the model calculations.
The sediment accumulation rates (cm.yr − 1 ) based on 137 Cs, 239+240 Pu and 241 Am were calculated by dividing the distance between the surface and the depth of maximum activity concentration by the amount of time between the retrieval date and the year of maximum discharge, assuming a constant rate of deposition (Methods Table 2). The final accumulation rate of each core was calculated as the average of the three accumulation rates calculated for 137 Cs, 239+240 Pu and 241 Am. This model produced an accumulation rate of 0.605 cm.yr − 1 at site 1 (which is consistent with previous work at this marsh (Morris et al., 2000;Oh et al., 2009;Marsdem et al., 2006)), 0.336 cm.yr − 1 at site 3 and 0.308 cm.yr − 1 at site 4. The activity concentration maxima were not necessarily observed at the lower reach of the core retrieved at site 2, so this profile could not be objectively dated.

Biota sampling
Samples of marsh vegetation, seaweed and periwinkles were retrieved on 7th September 2020 between sites 1-5. 'Marsh vegatation' includes sea purslane (Sesuvium porulacastrum), salicornia (Salicornia Europea) and sea plantain (Plantago maritima). The common periwinkle (Littorina littorae) was the only live mollusc species found on site, and two different species of seaweed were collected: common bladderwrack (Fucus vesiculosis) and Irish moss (Chondrus crispus). All marsh vegetation and seaweed samples were retrieved whole from the root, except for separate additional samples of the endmost ≤2.5 cm of Fucus vesiculosis fronds, at the furthest point from the root. Samples were collected and stored in sealed airtight plastic bags and frozen in a cooler on site prior to transportation to the laboratory.

Radiochemical procedures: DGT resin-gels
DGT sampler housings were disassembled, and the filter membranes and diffusive gels were discarded. The resin-gel in DGT Research R-SP sediment probes was cut into 5 cm x 1.8 cm or 3 cm x 1.8 cm sections using a scalpel. 5×4 cm resin-gels in DGT-55 sediment probes were retrieved whole. Resin-gels or resin-gel segments were gently washed with a stream of deionised water and placed in a new glass beaker which was pre-washed with 0.1M HCl. 10 mL of 8 M HNO 3 plus five drops of H 2 O 2 were added to elute Pu and U from the KMS-1 resin-gels, and 10 mL of 3 M HCl was added to elute Am from IIP-Y 3+ resin-gels. Appropriate tracers were added to the eluent at this point; around 1 pg each of 233 U and 242 Pu for KMS-1 resin-gel samples and 243 Am for IIP-Y 3+ resingel samples. Resin-gels were left in the eluent overnight and separated into a clean beaker using a 0.45 μM syringe filter, with three deionised water beaker washouts of approximately 5 mL also passed through the syringe filter. All filtered eluents were evaporated to dryness (the 'sample residue'). The sample residue was then put through radiochemical separations (see the 'Plutonium separations' section next for KMS-1 resin gel samples, or the 'Americium separations' section next for IIP-Y 3+ resin gel samples).

Radiochemical procedures: biota samples
Marsh vegetation and seaweed samples were washed thoroughly with deionised water. Externally precipitated actinides were then eluted from the exterior of the leaves by immersing marsh vegetation and seaweed in 1 L 0.1 M citric acid / 0.1 M sodium citrate buffer for one hour. This solution was then decanted, and the plants were thoroughly rinsed with deionised water before desiccation at 80 • C for 24 hours and then calcination overnight at 550 • C. Ashes were dissolved in 20 ml concentrated HNO 3 in Teflon® canisters, traced with ~1 pg 233 U, 242 Pu and 243 Am each, and wet ashed in a pressurised microwave reactor (Milestone UltraClave IV) at 180 • C under 50 bars pressure. The sample was decanted and evaporated to dryness (the 'sample residue'). The sample residue was prepared for sequential extraction of Pu, then U, then Am (see following sections).

Plutonium separations
Pu separations were always performed prior to U separations, as described by Chaplin et al. (2021) for sequential Pu and U extraction from KMS-1 resin-gel samples. The sample residue was redissolved in 10 mL 8 M HNO 3 . 50 μg Fe(III) dissolved in 1 mL 0.1 M HCl was added, and the sample was gently agitated. 1 mL 100 mg.ml − 1 ascorbic acid solution was added to reduce all Fe(III) to Fe(II), consequently co-reducing all Pu in the sample to Pu(III). 40-60 mg NaNO 2 was added, and the sample was heated on a hot plate at 150 • C for 10 min to fix all Pu in the sample to Pu(IV). The sample was allowed to cool, and Pu(IV) was extracted on a Bio-Rad Poly-Prep column containing 2 mL of Bio-Rad AG 1-X4 resin, preconditioned with 10 mL 8 M HNO 3 . The column was washed with 3×5 mL 8M HNO 3 beaker washouts and subsequently with 15 mL of 9 M HCl to elute any Th present in the sample. The waste fraction was retained for subsequent separation of U (and then Am, for biota samples). Pu was eluted from the column with 15 mL of H 3 NO⋅HCl (5% aq.). H 3 NO⋅HCl in the eluate was destroyed in an acid fume cupboard by dropwise addition of HNO 3 (concentrated) and the remaining eluate solution was evaporated gently at 150 • C. The eluate residue was prepared for AMS target manufacture (see 'AMS target preparation' section).

Uranium separations
The waste fraction from Pu separations was evaporated to dryness and redissolved in 3 mL of 8 M HNO 3 . 100 mg of UTEVA (Triskem, France) resin was packed into a 1 mL micropipette cartridge column and conditioned with 3×1.5 mL 8 M HNO 3 . The samples were passed though the column and 3×1.5 mL 8 M HNO 3 beaker washouts were then transferred through the column. The columns were further washed with 3×1.5 mL 6 M HCl to remove residual Th, and U was eluted with 10 mL of 0.01M HCl. The waste fraction of biota samples was retained for subsequent separation of Am. The eluate was evaporated gently at 150 • C and prepared for AMS target manufacture (see 'AMS target preparation' section).

Americium separations
The sample residue was redissolved in deionised water at pH 3.0 ± 0.1 (adjusted to pH 3.0 ± 0.1 with dropwise addition of 0.1 M HCl). A 2 mL cartridge was packed with IIP-Y 3+ resin (Chaplin et al., 2021;Chauvin et al., 2006) and pre-washed with 5 mL 0.1 M C 2 H 2 O 4 . The cartridge was then conditioned with 20 mL of deionised H 2 O (pH 3.0 ± 0.1). The loading solution was collected into a syringe and passed through the cartridge using a syringe pusher at 1 mL.min − 1 . The column was washed with 3 × 5 mL H 2 O (pH 3.0 ± 0.1) beaker washouts and Am (III) was eluted with 20 mL of 3 M HCl. The eluate was evaporated gently at 150 • C and prepared for AMS target manufacture (see 'AMS target preparation' section).

AMS target preparation and measure
Residues of the evaporated eluates were dissolved in 10 ml 0.1 M HCl. The sample was heated gently at 50 • C and stirred at 150 rpm. 1 ml Fe(III) 2 mg.ml − 1 solution was added and left to equilibrate in the stirred solution for at least 15 minutes. Fe-hydroxides were then precipitated at pH 7 by dropwise addition of NH 4 OH (30%) and left to equilibrate in the stirred solution for at least 15 minutes. The actinides in the sample were co-precipitated with the Fe-hydroxides. The precipitate was decanted and centrifuged, then re-suspended in Milli-Q water and re-centrifuged to wash out any remaining NH 3 prior to baking. The precipitate was transferred into a 2.5 mL quartz crucible (held within an open-top brass sample wheel to withstand high heating) and desiccated at 80 • C until dry. The powder was baked at 650 • C for at least 4 hours to convert Fe into the oxide form. 2-3 mg Nb powder was added to the baked sample powder and mixed with a spattle. The powder was then compressed into a modified NEC cathode as per ETH Zürich's Laboratory of Ion Beam Physics (ETHZ-LIP)'s internal procedure (Christl et al., 2013). U measurements were normalised to the ETHZ-LIP in-house ZUTRI standard (Christl et al., 2013) and Pu measurements were normalised to the ETHZ-LIP in-house CNA standard (Christl et al., 2013). Am measurements were normalised to a standard produced from certified NIST-traceable 241 Am and 243 Am sources provided by the Radiometrology group of CHUV-IRA. Most measurements were performed using EHTZ-LIP's TANDY cAMS system (Christl et al., 2013), with some U measurements performed on the recently developed MILEA system. For note, if a U target was planned for measure by MILEA, it was co-precipitated with 1.25 ml Fe(III) 2mg.ml − 1 solution and mixed with 6 mg of Nb.

Radioactive decay corrections 241
Pu (t 1/2 = 14.4 years) masses were decay-corrected from the date of measurement to the reference date of sampling (or the midpoint of 14day DGT sampler deployments). In all samples except IIP-Y 3+ DGT samplers for Am measure (which do not capture Pu), the consequent ingrowth of 241 Am from the decay of measured 241 Pu present in the same sample was also accounted for. Other measured isotopes ( 236 U, 239 Pu, 240 Pu, 241 Am) have much longer half-lives and decay correction was therefore not performed as it does not significantly alter the results.

Calculation of c DGT
The average analyte concentration at the sediment-porewater interface according to the DGT sampler, c DGT (g.mL − 1 ), was calculated using Eq. (1): A is the quantity of the analyte eluted from the resin-gel (in this work calculated by AMS, with the atom ratio of the analyte isotope to the tracer isotope measured and then converted to g); t is the exposition time (seconds); S is the surface area (cm 2 ) of the sampler section exposed to the sediment; Δg is the thickness (cm) of the material diffusion layer, and D is the diffusion coefficient of the analyte (cm 2 .s − 1 ). D was taken from the values reported for U, Pu and Am in seawater in Table 1 of Chaplin et al. (2021) and calibrated to the average seawater temperature during the deployment period (Methods Table 3) using the Stokes-Einstein correction (Eq. (4) of Chaplin et al. (2021)). Daily average seawater temperatures for Whitehaven harbour for each day of both sampling campaigns were retrieved from seatemperature.info on 16 June 2021. These were used to calculate the average temperature throughout the deployment period to calibrate D. We confirmed using porewater concentration (c pw ) data produced using adjacent peepers, that the rate of solute resupply (Davison, 2016) (R, = c DGT /c pw ) was <<0.1 (Supporting Information Table 2). This confirmed that the supply of the analyte to the DGT sampler was diffusion-controlled. The 14-day deployment period therefore completely depleted the sediment in the vicinity of the DGT sampler, to the point where supply to the DGT sampler was diffusion-controlled only. In this case, we normalised c DGT to 24 h, as is appropriate to represent the solution concentration and the maximum induced flux to the resin-gel (Davison, 2016). The normalised c DGT was used to calculate the remobilisation flux (F).

Calculation of the remobilisation flux
The remobilisation flux, F (fg.hr − 1 .cm − 2 ) was calculated using Eq. (2): The D used in this work for sediment and seawater DGT deployments are presented in Methods Table 3, below.
In this work, Where δ g is the thickness of agarose cross-linked polyacrylamide (APA) diffusive gel used during deployments (0.078 cm or 0.039 cm), δ f is the thickness of the membrane filter (0.014 cm), and δ dbl is the diffusive boundary layer. For DGT samplers deployed in seawater, this was taken as 0.049 cm from Chaplin et al. (2021)). For DGT samplers deployed in sediment, δ dbl = 0 cm.

Remobilisation flux variations and controls
Remobilisation fluxes of Pu, Am and U in the Esk Estuary significantly vary both temporally (Fig 1a,b) and spatially (Fig. 1a, f and b-e). This underlines the need for a time-integrated method for their valid determination. Pu and Am remobilisation fluxes between 24th May and 7th June 2021 were significantly higher in fine-grained sediment (silt and clay) of intertidal marshland (average 239+240+241 Pu and 241 Am fluxes of 1.9 ± 0.5 and 0.120 ± 0.015 fg⋅h − 1 ⋅cm − 2 , respectively, Fig. 1a) than in sand in the lower estuary channel (0.5 ± 0.03 and 0.010 ± 0.003 fg⋅h − 1 ⋅cm − 2 , respectively, Fig. 1f). Higher fluxes may be expected for sandy sediments compared to equally contaminated silts and clays, if considering the main controlling factor is the lower k d of sand versus k d of silt and clay for Pu and Am. The results here therefore reflect that the finer sediments hold a larger actinide inventory than the coarser sand, and therefore supply more Pu and Am to the interstitial waters despite having a higher and therefore more retentive k d .
There were consistent temporal variations in the Pu fluxes (measured by DGT samplers configured with the KMS-1 resin-gel) and the Am fluxes (measured by separate DGT samplers configured with the IIP-Y 3+ resin-gel) between two deployment periods, nine months apart. At site 1, the 239+240+241 Pu and 241 Am fluxes in the top 20 cm of sediment, respectively averaged 10 ± 2 and 1.09 ± 0.09 fg⋅h − 1 ⋅cm − 2 between 24th August and 7th September 2020 (Fig. 1b), compared to 1.3 ± 0.5 and 0.150 ± 0.02 fg⋅h − 1 ⋅cm − 2 between 24th May to 7th June 2021 (Fig. 1a). We additionally determined the flux of anthropogenic 236 U to the same KMS-1-configured DGT samplers that measured the Pu fluxes. We found the opposite trend with the 236 U fluxes compared to Pu and Am between the two sampling campaigns; the 236 U fluxes were lower in the top 20 cm of sediment at Site 1 between 24th August and 7th September 2020 (0.07 ± 0.005 fg⋅h − 1 ⋅cm − 2 ) than between 24th May to 7th June 2021 (0.116 ± 0.007 fg⋅h − 1 ⋅cm − 2 ). As U is a highly conservative radionuclide, compared to non-conservative Pu and highly nonconservative Am, this suggests that more infiltration of seawater into the marsh sediments during this period. Further evidence for this is provided by the lower average 240 Pu/ 239 Pu signatures of the DGT resingel segments deployed between 24th May to 7th June 2021 (0.197 ± 0.015, Fig. 2a) compared to those deployed between 24th August and 7th September 2020 (0.224 ± 0.018, Fig. 2b). This suggests a dilution of higher sediment 240 Pu/ 239 Pu signature 3a) with the lower 240 Pu/ 239 Pu signature of the seawater, which is closer to global fallout (Fig. 3), Furthermore, the 236 U/ 239 Pu ratio also demonstrated this trend, indicating a much higher relative presence of 236 U between 24th May to 7th June 2021 (Fig. 2a, b). This points to the influence of the hydraulic regime as a driving factor behind the remobilisation fluxes at site 1. The organic content of the sediments also likely has an influence on the magnitude of the fluxes. A steep drop in the Pu fluxes at the 12.5 cm depth midpoint of Fig. 1a corresponds with a layer from an adjacent core retrieved during the same sampling period, which has decreased Si, K, Ca, Fe and Rb and increased S (Supporting Information Fig. 1, marked as (c org ). This indicates an organic-rich layer, as further evidenced by a lighter horizon on the X-radiograph. The association of Pu with organic humic and fulvic acids has previously been shown to decrease its diffusion coefficient (Cusnir et al., 2014), possibly explaining the drop observed in the fluxes in Fig. 1a. We consider other influences on the fluxes to be minimal; for example, the fluxes generally do not show major variations throughout the profiles and especially at depth, whilst factors such as potential aging effects may be expected increase the sediment-water k d of metal cations and therefore significantly reduce the fluxes. Furthermore, previous work has demonstrated that U aged in soil in-situ for >2000 years behaves similarly to artificially introduced depleted U ( Straub et al., 2022).

Constraining the actinide sources using isotopic reports
We analysed several isotopic fingerprints in the sediments (Figs. 2 and 3), seawater (Fig. 3) and local biota (Fig. 3) of the Esk Estuary. 240 Pu/ 239 Pu and 241 Pu/ 239 Pu atom ratios reflect the degree of Pu-burn up of the source material and therefore indicate its origin, whilst the 236 U/ 239 Pu and 241 Am/ 239 Pu atom ratios are moreover influenced by geochemical elemental fractionation. 240 Pu/ 239 Pu in the top 10 cm of the marsh sediments between Sites 1-4 averaged 0.215 ± 0.016 (n = 13; Figs. 2a-e, 3a). This is significantly higher than the seawater at site 5 (lower marine estuary, 0.196 ± 0.015 (n = 3)) and site 2 (middle estuary with strong mixing of seawater and freshwater, 0.191 ± 0.015 (n = 3)). The seawater and adjacent seawater DGT probes correspond closer to the global fallout value of 0.18 ± 0.01 (Fig. 3a). These values fall significantly below the IAEA-381 reference Irish Seawater from 1993, which more closely matches the signature of the sediments (0.23 ± 0.02) (Corcho Alvarado et al., 2022). Additionally, this is higher than the IAEA-135 Irish Sea sediment reference material from 1992 (0.205 ± 0.002) (Corcho Alvarado et al., 2022). This indicates that the marsh sediments have accumulated more Pu from reprocessing discharges (which has a higher signature than that of global fallout) from the seawater over the preceding decades. Additionally, modern Irish Seawater has become diluted with the residual global fallout Table 3 Average seawater temperatures and corresponding D used to calculate c DGT .

Campaign
Average seawater temperature (Whitehaven) The bulk sediment profiles in the sediment cores at Sites 1-4 demonstrate co-incident maxima in the 239+240 Pu, 241 Am, and 137 Cs profiles with depth ( Fig. 2a-e), representing accumulation with sedimentation over time as a function of the seawater concentration. With 210 Pb-dating problematic around the Cumbrian coastline due to other non-nuclear anthropogenic discharges (McCartney et al., 1990), we used the 239+240 Pu, 241 Am and 137 Cs maxima to date the cores (see Methods for age model). The presence of well-defined subsurface activity maxima in bulk sediment, which can be correlated with the Sellafield discharge history, shows a lack of large-scale mixing or bioturbation in these marsh sediments. The period from the 1950s to early 1960s (when the primary purpose of the Sellafield facility was the production of weapons-grade Pu, which requires lower Pu-burn up and therefore has 240 Pu/ 239 Pu of <0.07 (Warneke et al., 2002;Ketterer and Szechenyi, 2008)) correlates with a significant drop in the 240 Pu/ 239 Pu ratio to 0.148 ± 0.010, indicated by the DGT probe at 35-40 cm depth below Site 1 (Fig. 2a). Ratios remain above 0.14 in these deeper samples however, indicating an attenuation of the weapons-grade signature, either due to remobilisation throughout the sediment column and/ or post-depositional dilution with later sources with higher 240 Pu/ 239 Pu, from (and after) the period of peak Sellafield discharges in the 1970s. Furthermore, 240 Pu/ 239 Pu ratios at depths corresponding to before the onset of the nuclear age demonstrate remobilisation of Pu downwards   Table 3). 241 Am/ 239 Pu global fallout ratio calculated from the 241 Am/ 239+240 Pu ratio reported by Froidevaux et al. (2010), assuming global fallout 240 Pu/ 239 Pu ratio of 0.18 ± 0.01. Box and whiskers plotted for n ≥ 4. Outliers marked as crosses. ±2σ uncertainties. throughout the sediment column, through the infiltration of precipitation and/or tidal pumping of seawater in the marsh. This outlines once again the influence of the local hydraulic regime in the remobilisation of Pu in intertidal marshland.
The 241 Am/ 239 Pu isotopic ratios varied significantly between sites, demonstrating elemental fractionation of the two chemically distinct actinides as a function of the local geochemical conditions. An anomalous peak in the 241 Am/ 239 Pu ratio to around 1 indicated by the DGT sampler at 10-15 cm below the surface at site 1 (Fig. 2a) likely reflects an increase in k d for Pu at that depth and/or a decrease in k d for Am. This correlates with an increase in the c DGT ( 241 Am) and c DGT ( 236 U) concentrations, respectively produced by the separate IIP-Y 3+ -configured DGT sampler and KMS-1-configured DGT sampler, alongside a decrease in c DGT (Pu) produced by the same KMS-1-configured DGT sampler (Fig. 1a). This effect was not observed nine months previously at the same depth beneath site 1 (Figs. 1b, 2b), pointing to time-specific geochemical conditions that impacted the Pu flux. However, the 241 Am/ 239 Pu ratio also peaked to around 1 at 14-17 cm below Site 4, but alongside a sharp drop in the k d (c DGT ) (the bulk sediment concentration/ c DGT ) of 241 Am (Fig. 2e). The 241 Am/ 239 Pu ratio is well correlated and inversely proportional to the k d (c DGT ) of 241 Am at the Brighouse marsh (sites 3 and 4, Fig. 2d, e), but is not closely correlated at the Railway Marsh (Sites 1 and 2, Fig. 2a-c). The sharp drop in k d is outside of the characteristic range of Am for clays, and better represents that of sands. Finer marsh sediments are known to deposit on top of coarser sand in this area (Clifton and Hamilton, 1982), therefore it is most probable that the lower sections of the DGT sampler were in increasingly coarse sand 11-19 cm beneath Site 4, given its proximity to the sands of the channel. These relationships between the k d (c DGT ) and the fluxes demonstrate that k d is not necessarily the main or sole controlling factor on the remobilisation fluxes of Am, and that a more holistic technique such as DGT is required to assess these.

Bioaccumulation of actinides from identified sources
We hypothesised that marsh vegetation would accumulate actinides from the root zone (up to 10 cm beneath the surface of the sediments) and thus its isotopic signature should match those of the near-surface sediments, whilst the isotopic signatures of the seaweeds should rather match those of the seawater. In fact, we observed that the higher 240 Pu/ 239 Pu of the sediments closely matches that of all the biota that we sampled in the estuary (Fig. 3a). This demonstrates the sediments have been the principal supply of Pu to local organisms, indicating the remobilisation of sediment-associated Pu into the local estuarine environment.
The bulk seawater samples from Sites 2 and 5 had a much lower 241 Am/ 239 Pu signature than the biota (Fig. 3c). There is also a significant difference between the 236 U/ 239 Pu signatures of the bulk seawater and biota (Fig. 3d). However, the DGT samplers placed in-situ for two weeks at the same sites better reflected the 241 Am/ 239 Pu and 236 U/ 239 Pu fractionation in the periwinkles, seaweed and marsh vegetation compared to the bulk seawater (Fig. 3c, d). Effectively, DGT more closely reproduced the elemental fractionation in the biota, demonstrating its ability to mimic the diffusive uptake of bioavailable species through living cell membranes and thus provide a bioavailability proxy for Pu, U and Am isotopes.

Conclusion
We have demonstrated that DGT presents strong advantages to provide a site-specific and time-weighted appraisal of actinide remobilisation and bioavailability. The variability in the remobilisation fluxes underline the necessity to have an actinide-specific and time-integrated technique at dispositionespecially a technique which provides a bioavailability assessment to gauge the radiological impact to organisms. Furthermore, the combination of DGT and AMS allows these data to be produced without requiring the processing of larger and significantly more radioactive samples of sediment and seawater, allowing the assessment of remobilisation fluxes at sub-femtogram levels. We therefore see scope to employ this methodology to improve dose modelling during wider studies of radionuclide remobilisation using DGT. Furthermore, the insights provided here regarding the potential of actinide re-incorporation into the biosphere from sediment will contribute to other environmental projects which must take this into account, such as the near-surface and deep disposal of decommissioning and nuclear reactor/reprocessing wastes.

Authors contributions
J.D.C. developed and manufactured DGT samplers, took part in the planning of field deployments, undertook sampling on-site, performed radiochemical processing of samples, processed and interpreted data, coded and constructed graphics, wrote the manuscript and compiled edits to the manuscript. M.C. performed AMS measurements and data reduction, took part in the planning of field deployments, processed and interpreted data, contributed to the study conception and contributed to the manuscript. A.B.C. and P.E.W. took part in the planning of field deployments, undertook sampling on-site, processed and interpreted data, contributed to the study conception, and contributed to the manuscript. D.G.R. performed gamma spectrometry on sectioned sediment cores, interpreted experimental data and contributed to the manuscript. F.B. supervised and administrated the PhD studentship, took part in the planning of field deployments, interpreted experimental data, contributed to the study conception, and contributed to the manuscript. P.F. conceived the study and the development of DGT samplers, supervised and administrated the PhD studentship, acquired funding, took part in the planning of field deployments, interpreted experimental data and contributed to the manuscript.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request.