Biogeochemical processes and radionuclide behaviour in soil-plant systems

Soil sampling photographs: Sergey Gaschack, Chornobyl Center, Ukraine
Soil sampling photographs: Sergey Gaschack, Chornobyl Center, Ukraine


In its generic post-closure safety assessment, the NDA identifies radionuclides which are expected to contribute most to radiation exposure and risk in the long-term. Iodine-129 and 36Cl are identified as being of general concern. Depending on the waste type being considered 79Se, 99Tc, 135Cs and several uranium isotopes are also identified. Similarly, 129I, 79Se,238U and 99Tc are recognised as key radionuclides in site-specific calculations carried out for proposed repositories in Finland and Sweden. In addition to long physical half-lives (ranging from approximately 2x105 to 4.6x109 yr) the major factor controlling breakthrough to the biosphere is the potential mobilisation of these radionuclides (especially as TcO4-, I-,Se (selenate) and U (UO2) species) in the geosphere. In this work package we considered the behaviour of 129I, 79Se, U-isotopes and 99Tc in soil-plant systems and also took the opportunity to test our existing Cs model on soils from the CEZ. In the case of 129I, 79Se and 238U we capitalised on our recent and on-going research. While 36Cl is a potentially important contributor to dose, Cl behaviour in the soil-plant system is relatively well understood with the possible exception of organic forest soils. We therefore did not consider Cl in this work.

The geosphere and biosphere transport models used by NDA and others to predict the long-term radiological risk from geological repositories have necessarily simplified the representation of many key processes. In particular, adsorption processes are represented using distribution coefficients (Kd) which amalgamate complex physico-chemical processes into a single empirical parameter describing the relative activity concentrations in the solid and liquid phases. Reactive chemistry and transport models are used for more detailed consideration of radionuclide behaviour in the geosphere, but application of these models to the biosphere is restricted to prediction of isotope retardation from Kd values. Models, providing sophisticated representations of physical transport in soil, of radionuclides with complex chemistry (e.g. 79Se), still rely on selection of appropriate Kd values. Values of Kd used in risk calculations are usually derived in batch sorption experiments and are seldom ‘validated’ against field-derived data.

Stable elements and their radioisotopes share the same reaction mechanisms and biogeochemical sinks, subject to small isotopic discrimination effects. It is often assumed that low activity concentrations of radionuclides added to soil will eventually become proportionately distributed within naturally occurring pools of the corresponding stable isotopes. On this basis, it has been suggested that Kd values for indigenous stable analogues, or in the case of U primordial isotopes, may provide reasonable estimates of the solubility of long-lived radionuclides emerging into the biosphere from deep waste repositories. However, periods required for equilibration in soil are unknown and the validity of short-term soil-contact studies remains untested. As 99Tc possesses no stable isotopes, its long-term kinetic incorporation into natural biogeochemical cycles can only be studied at a limited number of contaminated sites around the world. The CEZ, provides a ‘natural laboratory’ within which, uniquely, the long-term behaviour of 129I, 99Tc and U (235U was released by the accident) can be studied and compared with model extrapolations from short-term studies of radioisotope dynamics. The early effects of association of these radionuclides with fuel particles deposited within the CEZ can be taken into account following mapping of the dissolution rates in different soil types across the affected area.

Objective and hypotheses

The primary objective of this work package was to improve our understanding of, and ability to predict, the long-term biogeochemical behaviour of 129I, 79Se, 99Tc, and U isotopes, in soils whilst assessing the validity of models parameterised from short-term laboratory experiments. In most soils 129I, 79Se, 99Tc and U isotopes entering the biologically active zone, from precipitation/irrigation or groundwater sources, either undergo time-dependent transformation to relatively stable organic forms within humus (e.g. Se and Tc) or are occluded within poorly soluble oxides, phosphates and carbonates (e.g. U). The processes, and rate-determining factors, that control this ‘transfer-to-sink’ are poorly understood. For example, iodate (IO3-) is more strongly adsorbed by Fe/Mn oxides than iodide (I-) during transition to organic forms and must under go reduction to a reactive species prior to interaction with humic C in contrast to the oxidation of iodide. Understanding the biogeochemical processes which control the behaviour of these radionuclides in soils will provide a strong platform on which to base defensible models of their future fate and impact which will be applicable to any repository site considered in the UK (or elsewhere). TREE addressed the key question of whether the long-term behaviour of 129I, 79Se, 99Tc and 235U in soils can be adequately predicted from an understanding of short-term biogeochemical dynamics. The hypothesis addressed is that:

The progressive equilibration of 129I, 79Se, 99Tc and U, in soil can be estimated from models parameterised solely from short-term isotopic mixing experiments. The current soil activity concentrations, geochemical fractionation and soil profile distributions of 235U, 99Tc and 129I in abandoned farmland and forest soils close to the Chornobyl NPP were used to test this hypothesis. For Se the comparison with added Se (in practice 77Se as a proxy for 79Se) can only be carried out with native soil Se.

Approach: We undertook controlled incubations to follow changes (over c. 2.5 yr) in speciation/fractionation of isotopic spikes added to well characterised soils; and conducted measurements of current activity concentrations, soil profile distribution and (where analytically possible) the fractionation and speciation of radionuclides in soils collected in the CEZ.

Incubation experiments followed the fixation into humus and/or occluded mineral forms of 129I, 77Se (enriched >99% IA), 99Tc and 238U in soils, and the species transformations associated with this process, in a wide variety of soil types incubated aerobically (c. 80% of field capacity) at 10°C. Soils (c. 30, mainly from the UK but including 10 from the CEZ) were selected from a range of land uses (grassland, woodland, arable, etc.) and included variation in characteristics likely to influence isotope fixation rates (pH, parent material, texture, humus content and Fe/Mn oxide content). Isotopic additions were carefully gauged to enable assay of the added tracer while minimising perturbation of existing pools. Changes in fractionation at 12 time intervals, over c. 2.5 yr in a logarithmic time sequence, were determined by:

(i) equilibration (24 h) at 100% field capacity in 0.01 M Ca(NO3)2 representing soil pore water; (ii) extraction with KH2POto dissolve inorganic sorbed anionic species (e.g. SeO42-, IO3-, TcO4-) bound to Fe/Mn oxides; (iii) extraction with 1 M Mg chloride to solubilise exchangeable UO22+; (iv) determination of isotopically exchangeable  UO22+ - the use 233UO22+ in isotopic dilution was developed at UoN to determine (specifically) labile  UO22+ species in soils; (v) extraction with tetra-methyl ammonium hydroxide (1% TMAH, pH 12) to dissolve colloidal and molecular humic and fulvic acids with accompanying organically carbon-bonded species. While recognising the operational nature of all extraction systems, this nevertheless gives a meaningful basis for fractionation.

All phases were assayed by ICP-MS, anion exchange LC-ICP-MS and SEC-ICP-MS (size exclusion chromatography) to determine speciation and the time-dependent distribution of the spiked isotopes within the native soil pools. Work at UoN with a limited range of arable and grassland soils incubated with 129I, 77Se and 233U have enabled the development of this approach. TREE further developed this methodology and, as an added benefit, provided post-doctoral and post-graduate training in terrestrial geochemistry of radionuclides.

A range of plants were grown in the artificially contaminated soils after c. 2.5 yr incubation by which time soil bioavailability will be changing only slowly. A selection of 12 plant species of contrasting radionuclide uptake patterns, identified via work package 2 were grown in pots containing pre-incubated soils of different characteristics. Plants were grown to seedling stage only as we have shown that this is adequate to establish phylogenetic patterns. It was not our intention to produce new sets of soil-to-plant transfer coefficients; it is accepted that the limited growth stage of the test plants precludes this. These data are intended as a proof of concept link between the geochemical characterisation and plant uptake. The studies also provided material for analyses for the development of phylogenetic and ionomic models of plant transfer in work package 2 for a wide range of elements relevant to radiological assessment. Linking the geochemical characterisation and plant uptake enabled us to test the extent to which predicting uptake in conjunction with bioavailability reduces uncertainty.

CEZ soils, which received a discrete and significant input of radionuclides, including 235U, 129I and 99Tc in 1986 were used to investigate the approach to steady state conditions after 27 – 29 yr contact. These soils provided a unique resource with which to test our hypotheses. We assayed 129I, 235U and 99Tc in the study soils, using accelerator mass spectrometry (AMS) and ICPMS as appropriate. In addition, we determined the extractable fractions listed above and isotopically-exchangeable fractions of contaminant nuclides and native analogues where a suitable tracer isotope exists (e.g. 233U) and where these measurements are analytically feasible.

Kinetic soil lability models were developed to represent the time-dependent distribution between available and inaccessible forms with rate coefficients parameterised using the measured distributions and related to readily measureable soil characteristics. The short-to-medium term observations (days – months) of isotopic transformations in soils were used to parameterise these rate coefficients and, in turn, relate them to relevant soil characteristics (pH, humus, Fe, Mn, Al oxides etc.). The resulting models of radionuclide dynamics will have the capacity to assess themobility and availability of radionuclides following soil contamination and to identify the soil characteristics which influence the rate of fixation into less labile forms. The approach is similar to that we have previously applied for Cd and Zn. The key development in this work is that the models we develop for 129I, 99Tc and 235U from short-term incubation studies were evaluated by comparison with observations of current radionuclide fractionation in soils from the CEZ. This provided a unique opportunity for testing short-term laboratory-parameterised models using field soils subject to long-term radionuclide contact.

We also rigorously assessed uncertainty in model predictions arising from the simplifying assumptions implicit in the formulation and parameterisation of bioavailabilty models. Crucially, we determined uncertainty in: (i) predictions of long-term isotopic availability from models parameterised using short-term (0-2.5 yr) laboratory incubations and (ii) the transition from site and timescale-specific models to generic simulations using model parameters predicted from accessible aerobic soil characteristics (e.g. pH, LOI, DOC). To achieve this, a series of comparisons between observation and prediction were made using models parameterised using progressively fewer site and timescale specific observations. This involved models parameterised from short-term incubation experiments applied to the historically contaminated field soils through to models parameterised using the full combination of laboratory- and field-contaminated soils. We recognise that in the latter case the model evaluation becomes increasingly confounded with model development as site-specific observations of field contamination are included in the parameterisation. However, cross validation procedures, in which subsets of the observations were systematically excluded, were applied to address this limitation. Although this work mainly considered the experimental data we collected during TREE it was extended to consider existing datasets and models for radiocaesium; these extend over >25 years and provide an excellent means of testing the viability of extending short-term model parameterisation to long-term prediction.

The core activities of this WP will be supplemented by two PhD studentships which will: (i) determine mechanisms and reaction rates of the conversion of 129I, 79Se and 99Tc to ‘fixed’ humic- and oxide-bound forms in soils and aquatic systems; (ii) evaluate the bioavailability to plants of radionuclides in contaminated soil using the ‘Diffusive Gradients in Thin films’ technique. The latter will utilise WP1 incubated soils and contaminated lysimeters of project partner HPA. We will work with project partners: UIAR who have data on Tc uptake by vegetation from artificially contaminated soils; SCK·CEN who have interests in repository assessments; UMB who are initiating a programme in the CEZ considering speciation of radionuclides other than those being considered here.