Kinetics of 99 Tc speciation in aerobic soils

Technetium-99 is a significant and long-lived component of spent nuclear fuel relevant to long-term assessments of radioactive waste disposal. Whilst 99 Tc behaviour in poorly aerated environments is well known, the long-term bioavailability in aerobic soils following direct deposition or transport to the surface is less well under- stood. This work addresses two questions: (i) to what extent do soil properties control 99 Tc kinetics in aerobic soils and (ii) over what experimental timescales must 99 Tc kinetics be measured to make reliable long-term predictions of impact in the terrestrial environment? Soil microcosms spiked with 99 TcO 4− were incubated for 2.5 years and 99 Tc transformations were periodically monitored by a sequential extraction, which enabled quantification of the reaction kinetics. Reduction in soluble 99 Tc was slow and followed a double exponential kinetic model including a fast component enhanced by low pH, a slow component controlled by pH and organic matter, and a persistently soluble 99 Tc fraction. Complexation with soil humus was key to the progressive immobilisation of 99 Tc. Evidence for slow transfer to an unidentified ‘sink’ was found, with estimated decadal timeframes. Our data suggest that short-term experiments may not reliably predict long-term 99 Tc solubility in soils with low to moderate organic matter contents. radionuclides and our observations suggest that steady state conditions were not attained in the majority of soils, even after 2.5 years incubation. Reduction in soluble 99 Tc over time followed a dual rate, double exponential kinetic model and included (i) a fast component reflecting immediate removal enhanced by low pH; (ii) a slowly removed fraction largely controlled by pH and organic matter; and (iii) a persistently soluble Tc fraction. The slowest rates of sorption were seen in calcareous soils, which clearly highlights potential risk to limestone and chalk aquifers; the rendzina soil types overlying such aquifers would fail to intercept deposited 99 Tc. Soil polyvalent metal oxides only played a marginal role in sequestering 99 Tc. By contrast, a combination of high OC concentrations and low pH produced faster sorption kinetics. The ability of soil humus to form strong complexes with 99 Tc is key to the progressive immobilisation of 99 Tc in soil. Our data suggest that short-term experiments (< 1‒2 years) may not be used reliably to predict long-term solubility and bioaccessibility of 99 Tc, particularly for soils with low to moderate OC contents. Our findings are of practical significance in risk assessment calculations for facilities such as radioactive waste repositories, whether based on generic or site-specific criteria. While our experimental study has not considered the fate and behaviour of 99 Tc in every possible soil type, the range of physico-chemical characteristics examined is wide enough to encompass the environmental circumstances under which many risk assessments will be made.


Introduction
Technetium-99 is an artificial radionuclide produced in nuclear reactors by fission of 235 U and neutron activation of 98 Mo. It has a UK radioactive waste inventory included 3.87 × 10 15 Bq of 99 Tc, equivalent to 6.17 tonnes and representing approximately 5‰ of the total radioactivity within the UK radioactive waste stock (NDA, 2017). As a result of its long half-life and its ability to migrate as a soluble anion ( 99 TcO 4 − ) the long-term environmental impact of 99 Tc after geological disposal of radioactive wastes is of considerable interest. In analyses of the implications of a proposed extension to the SFR repository in Sweden, 99 Tc consistently appears as one of the most important sources of radiation dose to humans exposed in the terrestrial environment, under several different scenarios, tens of thousands of years after repository closure (SKB, 2014).
Technetium-99 is a groundwater contaminant at some nuclear facilities (Hu et al., 2008) and numerous studies have addressed the migration and plant uptake of 99 Tc in the context of groundwater contamination (Ashworth and Shaw, 2005;Murphy and Johnson, 1993). Technetium-99 in groundwater exists predominantly as Tc IV , possibly complexed with soluble or colloidal humic substances (Maes et al., 2004). Even in this form, 99 Tc can be brought to the surface by deep-rooted plants (Murphy and Johnson, 1993). Direct contamination of surface soil with 99 Tc occurred as result of atmospheric nuclear weapons testing up to 1963 (Tagami and Uchida, 1997) and the Chernobyl accident in 1986 (Uchida et al., 1999). Following direct deposition from the atmosphere or transport to the surface environment from the subsurface, the question arises as to the long-term retention and bioavailability of 99 Tc in soils under aerobic conditions. This is important because such contamination may potentially compromise several ecosystem services that soils deliver thereby contributing towards the United Nations Sustainable Development Goals, including provision of food or enhancing biodiversity, among many other key functions (Keesstra et al., 2016).
Despite the expectation that 99 Tc in aerobic soils will exist as 99 Tc VII ( 99 TcO 4 -), several studies have shown that its bioavailability in soilplant systems is reduced with time (e.g. Echevarria et al., 1997). A fouryear lysimeter study by Vandecasteele et al. (1989) showed that 70 % of 99 Tc freshly added to soil as 99 TcO 4 − was rapidly removed by plant uptake, with a half-time of approximately 50 days. The 99 Tc remaining in the (aerobic) plough layer, however, was removed very slowly with an estimated half-time of approximately 30 years, suggesting a kinetically-controlled transformation from bioavailable to recalcitrant form (s). The critical redox potential for reduction of 99 Tc VII to 99 Tc IV is +200 mV at pH 7 (Koch-Steindl and Pröhl, 2001), which is typical of wet but not fully waterlogged soils (Sposito, 2008). Icenhower et al. (2009) have concluded that it is the availability of reducing agents, including organic matter and surface-sorbed Fe(II), which controls the reduction of Tc VII to Tc IV , rather than the bulk redox environment. In assessments of radioactive waste disposal, long-term predictions of 99 Tc availability for plant uptake and human exposure are needed for a range of scenarios. Soils with wide-ranging properties provide a spectrum of physico-chemical characteristics which may render 99 Tc either highly bioavailable or highly immobile, even under the typically aerobic conditions which are conducive to plant growth and crop production. Existing databases of solid-liquid distribution coefficients for 99 Tc provide a starting point for making exposure and risk calculations, but these data are often based on relatively short-term experiments and relatively small numbers of measurements (Gil-García et al., 2009). In this study our aim was to answer two questions: (i) to what extent do easily-measurable soil properties control the kinetics of 99 Tc speciation changes in aerobic soils and (ii) over what experimental timescales must we measure 99 Tc reaction kinetics in order to make reliable long-term predictions of 99 Tc impact in the terrestrial environment? Table 1 Location and general properties of the soils in this study, listed in order of land use. Bdl denotes below the detection limit.

Soil sampling
Twenty topsoil samples (0-15 cm depth) were collected from a number of locations in the UK (Table 1, Fig. S1) covering a broad range of relevant soil characteristics including pH, texture, organic carbon content, land use and parent material. Where possible, soils developed in the same parent material but under different land uses (e.g. arable or grassland versus woodland) were collected. Four or five sub-samples were collected within a ∼5 m 2 grid and combined to form a composite sample of ∼3 kg field moist soil. The field moist soil was homogenised and air dried at room temperature until just dry enough to be sieved to < 4 mm. The soils were not allowed to dry completely so as to maintain microbial activity. After sieving, soils were kept in polythene bags at 4°C to preserve their remaining moisture content without allowing anaerobic conditions to develop.

Soil characterisation
Approximately 30 g of each soil were oven dried (105°C) and ground in an agate ball-mill to produce a fine homogeneous powder for acid digestion and total element analysis. Soil digestion was undertaken by weighing 0.20 g ± 0.01 g of soil into a Savillex™ vial, adding concentrated Primar grade HNO 3 (4 mL) and heating at 80°C overnight to pre-digest the organic matter present in soils. This was followed by addition of concentrated Primar grade HF (2.5 mL), HNO 3 (2 mL) and HClO 4 (1 mL) and heating overnight to 160°C using a stepped heating program to digest silicate and oxide phases. The dry residue was reconstituted by warming with 2.5 mL ultrapure water and 2.5 mL HNO 3 ; the final volume was made up to 50 mL with ultrapure water (18.2 MΩ cm). Three standard reference materials (NIST SRM 2711a Montana Soil, NIST 1646a Estuarine Sediment and IRMM BCR-167 Estuarine Sediment) were digested in duplicate; five reagent blanks were prepared in an identical manner to check the accuracy and precision of the digestion and analysis methods. Elemental recoveries were typically > 80 % for the majority of the certified elements. All the soil digests and reagent blanks were diluted 1-in-5 before analysis by ICP-MS.
Soil pH was determined by shaking 10 g of sieved soil with 0.01 M CaCl 2 at a liquid-to-solid ratio of 2.5 L kg −1 for 30 min. Organic carbon (OC) contents were determined in finely ground material using a FLASH EA1121 CNS analyser. Total free iron oxides (Fe OX , Al OX and Mn OX ) in soils were extracted by shaking ca. 0.25 g finely ground soil with 20 mL of 0.3 M Na-citrate in 1 M NaHCO 3 and 0.07 M Na-dithionite for 24 h in a 20°C water bath followed by centrifugation and filtering to < 0.22 μm. Estimates of amorphous and poorly crystalline oxides (Fe AM , Al AM and Mn AM ) were obtained following extraction in 0.2 M ammonium oxalate and 0.125 M oxalic acid, shaken in darkness for 2 h following a method adapted from Schwertmann (1973). All filtered solutions (< 0.2 μm) were acidified to 2 % HNO 3 and diluted 1-in-100 before analysis by ICP-MS. from High Technology Sources Ltd. This primary stock was diluted with ultrapure water to produce a solution containing 62.6 mg L −1 99 Tc that was used to contaminate soil to be incubated in microcosms (initial background concentrations of 99 Tc in all soils were zero).
Portions of ∼2 kg (on a dry soil basis) of partially air-dried soils were weighed and the moisture contents were readjusted with ultrapure water. No attempt was made to achieve a pre-determined water content; rather, a friable but moist consistency was sought to aid soil mixing, which required different volumes of water to be added to individual soils. The volume of water added to each soil was recorded. The soils were then placed in an incubator at 10°C and allowed to equilibrate for 23 days. This pre-incubation step was to avoid including any short-lived 'flush' of microbial activity when the soils were initially moistened as part of the Tc incubation period; it was felt that this could distort the pattern of 99 Tc dynamics. After this period, a portion of ∼250 g soil was removed and transferred to 0.5 L Duran bottles to be used as control soils (and for further short-term experiments), whilst the remaining soil was amended with the equivalent of 108 μg 99 Tc kg −1 dry soil. To this end, approximately 2.8 mL of the 99 Tc solution were slowly added to each soil while the samples were mechanically stirred (c. 60 rpm) with the aid of a food mixer for 4 min to ensure uniform contamination with 99 Tc.
The Tc-amended soil was distributed equally between three Duran bottles of 1 L capacity as individual microcosms. The bottles had one single hole in the lid to allow gas exchange and to prevent anaerobic conditions developing whilst avoiding excess water loss during incubation. The soil weights for each microcosm (∼500 g dw) were recorded so the dry soil mass could be estimated and the moisture content could be monitored throughout the experiment. The contaminated and control microcosms were incubated in darkness at 10.0 ± 1.0°C for 897 days. The microcosms were regularly shaken by hand to aid aeration. During periodic extractions of the soils, soluble concentrations of Mn and Fe were monitored in all soils as indicators of any redox changes.

Microcosm sampling and sequential extractions
Microcosms were sampled periodically between initial contamination and final sampling at 897 days. The frequency of sampling was high (average of approximately 0.5 day) over the first 5 days (see Section 2.4) then reduced to an average of approximately 50 days for the remainder of the experiment. Physico-chemical transformations of 99 Tc in each soil during incubation were periodically monitored by means of a sequential extraction procedure (Table 2) based on the methods developed by Zhao and McGrath (1994) for sulphate and Shaw et al. (2019) for iodine. At each sampling time, a portion of soil equivalent to 4.0 ± 0.01 g dry soil was transferred from each microcosm to a 50 mL PE centrifuge tube and then subjected to a three-step sequential extraction procedure devised to quantify: (i) soluble, (ii) specifically adsorbed and (iii) organically bound 99 Tc pools in each soil.  Izquierdo, et al. Journal of Hazardous Materials 388 (2020) 121762 First, soluble 99 Tc ( 99 Tc sol ) was extracted with 20 mL 0.01 M KNO 3 (intended to mimic soil pore water composition) for 16 h using an endover-end shaker. This was followed by further equilibration with 20 mL 0.16 M KH 2 PO 4 for 16 h to extract specifically adsorbed 99 Tc ( 99 Tc ads ). At the end of each step the soil suspensions were centrifuged at 3500 rpm for 30 min, the supernatant solutions filtered through 0.22 μm syringe filters and divided into four equal aliquots. Two aliquots, stabilised with 2 % HNO 3 and 1 % TMAH respectively, were used to determine total 99 Tc concentrations within the extracts, whilst a third aliquot was immediately used for speciation analyses (described below). The remaining solution was kept at 4°C for DOC analysis. The third step of the sequential extraction involved heating the soils with 10 % tetra methyl ammonium hydroxide (TMAH) solution at 90°C for 14 h to dissolve organically bound 99 Tc ( 99 Tc org ) linked to humic and fulvic acids. These suspensions were centrifuged at 3500 rpm for 30 min and a 1 mL aliquot was further diluted to give a final concentration of 1 % TMAH for analysis. Only total 99 Tc analyses were undertaken using the TMAH extracts. The residual soil samples were weighed between extractions to account for carryover of the previous extractant and to calculate the total solution volume present at each stage. The difference between the combined sequentially extractable concentrations of 99 Tc and the total 99 Tc added enabled estimation of non-extractable 99 Tc (designated 99 Tc sink ) remaining in the soil.

Assessing short term Tc dynamics
The short term (< 5 days) dynamics of soluble 99 Tc in soils was investigated using sub-samples from the control soils. In order to keep the liquid-to-solid ratio as close as possible to 5 L kg −1 throughout all the extractions, the study of the short term dynamics was undertaken in two independent experiments.
To investigate 99 Tc behaviour during the first 3.5 h after addition to the soils, a moist mass of each control soil equivalent to 6 g dry weight was weighed into 50 mL PE centrifuge tubes in triplicate and 31 mL of 0.01 M KNO 3 were added. An air-filled head space (∼15 mL) was deliberately left to maintain an aerobic environment. An aliquot (2 mL) of 0.01 M KNO 3 solution containing 0.6 μg 99 Tc was added to the soil suspensions which were then shaken end-over-end for 3.5 h at room temperature. Samples of the supernatant (2.0 mL) were then taken after 0.5, 1, 2 and 3.5 h, following centrifugation of suspensions at 3800 rpm for 4 min, and filtered to < 0.22 μm. The filtered samples were then split into two aliquots: 0.5 mL was transferred to a glass vial containing 19.5 mL ultrapure water for DOC analysis whilst the remaining aliquot was transferred to an ICP tube containing 4 mL 1 % TMAH for further analysis. After each sampling, the soil suspensions were disaggregated with a vortex mixer and returned to the shaker immediately.
To study 99 Tc behaviour during the first five days of contact with the soils, an identical procedure as above was followed using another set of sub-samples from control soils. In this experiment the soil suspensions were sampled 6, 21, 46, 75, 100 and 123 h after addition of 99 Tc.

Analyses of solutions 2.5.1. Total concentrations of Tc
Samples were analysed using an iCAP-Q ICP-MS instrument (Thermo Fisher Scientific, Bremen, Germany). When analysing 99 Tc by mass spectrometry, isobaric interference may be encountered from 99 Ru (13 % isotopic abundance) as well as polyatomic interferences from Mo hydride ( 98 Mo 1 H + , 24 % abundance) and a range of K-based polyatomic species such as 40 Ar 41 K 18 O, 41 K 2 17 O and 41 K 40 K 18 O 9 . However, operating the instrument with a collision-cell (Q cell), using He with kinetic energy discrimination, was found to remove spectral interferences successfully, enabling accurate 99 Tc measurements while maintaining a sensitivity of 70-80 kcps ppb −1 . Further details on the analytical method and ICP-MS settings can be found in the Supplementary Information. Samples were introduced from an autosampler (Cetac ASX-520) incorporating an ASXpress™ rapid uptake module through a perfluoroalkoxy (PFA) Microflow PFA-ST nebuliser (Thermo Fisher Scientific, Bremen, Germany). Internal standards were introduced to the sample stream on a separate line via the ASXpress unit and included Ge, Rh (10 μg L −1 ) and Ir (5 μg L −1 ) in 2 % trace analysis grade (Fisher Scientific, UK) HNO 3 and Re and Ir (5 μg L −1 ) in 1 % TMAH. Peak dwell times were 10 mS for most elements with 150 scans per sample. Sample processing was undertaken using Qtegra™ software (Thermo-Fisher Scientific) utilizing external cross-calibration between pulse-counting and analogue detector modes, when required. The Radioactive Standard NIST SRM 4288B (31.55 kBq g −1 , 5 mL of 99 TcO 4 -in 0.001 M KOH) was diluted with ultrapure water to produce 100 mL of standard solution with a concentration of 2493 μg L −1 99 TcO 4 -, which was used for calibration (0.5, 1, 5, 10 and 25 μg L −1 ). Other elements relevant to understanding Tc sorption were also measured, including major cations, such as Fe and Mn, and trace elements.

Solution speciation
Speciation analysis was undertaken by IC-ICP-MS following in-line chromatographic separation using a Dionex ICS-3000 fitted with a 50 mm length PRP-X100 Hamilton anion exchange column. Development of an ion chromatographic method for adequate separation of 99 Tc species within a reasonable transit time was attempted. A number of potential mobile phase compositions and conditions were tested, as reported in the Supplementary Information. Following an optimisation process, the selected mobile phase consisted of isocratic elution with 0.05 M NH 4 ClO 4 solution at 1.5 mL min −1 . The Radioactive Standard NIST SRM 4288B, diluted as described above, was used as a pertechnetate standard for calibration (1, 5 and 20 μg L −1 ). Sample processing was also undertaken using Qtegra™ software (Thermo-Fisher Scientific) by manual integration of peaks. Instrumental drift was addressed by analysing 99 Tc calibration standards every 12 samples.

Statistical analysis
Statistical analyses of data, including linear regression and testing of statistical significance, were undertaken using the RStudio software package (v.1.1.383, RStudio, Inc). Significance testing of model fits is described in Supplementary Material (Tables S1 and S2).

Modelling 99 Tc kinetics
Time-dependent reductions in 99 Tc sol concentrations could be described using a dual rate, double exponential model with an 'offset' representing a persistently soluble fraction (Eq 1): where Tc sol (t) is the soluble 99 Tc concentration at any time, t, after initial contamination of the soils, A and B are the initial 99 Tc concentrations (μg kg −1 ) in solution subject to 'fast' and 'slow' depletion, respectively, and C is the 99 Tc concentration persistently remaining in solution. The first-order rate coefficients k fast and k slow (t −1 ) represent 'fast' and 'slow' rates of depletion of A and B, respectively. This model was fitted to experimental data from each soil using the Solver function in Microsoft Excel®; during fitting, the sum of A+B+C was constrained to a value of 108 μg kg -1 which was the total initial concentration of 99 Tc added to each soil at the beginning of the experiment. We attempted to fit other analytical models, including simple first order and spherical diffusion models, to experimentally measured 99 Tc sol concentrations but Eq. 1 consistently provided the best fits to data from all soils (see Fig. S2). The kinetics of 99 Tc org (TMAH-extractable) accumulation in soil could be described using a dual rate 'ingrowth' model (Eq. 2): where Tc org (t) is the TMAH-extractable 99 Tc concentration at any time, t, after initial contamination of the soils, R 1 and R 2 are fast and slow rates, respectively, of 99 Tc org formation and k 1 and k 2 (t −1 ) are fast and slow rate coefficients, respectively, of 99 Tc org 'loss' after formation.

Soil characterisation
The general properties of the soils used in this study are reported in Table 1. Organic carbon contents (OC) varied from 1.7 to 38.6% strongly reflecting land use and dominant vegetation at the sampling sites. Soil pH ranged between 3.4-8.0, with the lowest values reported for a woodland soil (BY-W) overlying pyrite-rich bedrock and the highest values for calcareous soils (SR-W, SR-A, WS-A) with inorganic C up to 5 %. The concentrations of free oxides spanned four orders of magnitude, ranging from 1180 to 22,700 mg kg −1 across the 20 soils examined. The broad ranges reported above reflect the diverse parent materials and land uses of the sites from which soils were collected, which included arable, grassland, woodland and moorland.

Short term dynamics (< 5 days)
The 99 Tc initially added to the soils as pertechnetate ( 99 TcO 4 − ) remained largely soluble in most soils within the first few hours after addition ( Fig. 1a and Fig. 2). Earlier studies have shown that 99 TcO 4 − is poorly adsorbed on soils or sediments under aerobic conditions during short-term batch experiments (Palmer and Meyer, 1981;Van Loon, 1986). However, evidence for removal of ∼30 % 99 Tc sol was found for peaty acidic soils (DY-M, WK-W) within 0.5 h of addition (Fig. 2). Changes in the soluble concentration of 99 Tc over time indicate that short-term transformations are largely governed by soil properties ( Fig. 1 and 3). For example, in arable soils (2-4 % OC) 99 Tc remained mostly soluble, whilst up to 33 % loss of soluble 99 Tc was observed in grassland soils (5-11 % OC) 5 days after Tc addition. For woodland and moorland soils (5-39 % OC) the loss of 99 Tc sol reached 50 % within 5 days in acidic soils. However, a minor loss of 5-11 % was noted for 2 woodland soils (BH-W and SR-W), which were collected from immature woodlands developed over limestone bedrock (Fig. 3). The above observations indicate that the decline in 99 Tc sol in this period was primarily controlled by both OC and pH.

Long term 99 Tc dynamics (5-897 days)
For all of the soils in the long-term experiment, a proportion of 99 Tc was found to remain in a soluble form throughout the 2.5 yr incubation period ( Fig. 1 a), probably due to pertechnetate being the most stable species under aerobic conditions and over a broad pH range (Shaw et al., 2004). However, a general decline in 99 Tc sol throughout the experiment was observed in all cases, with substantial differences between soils (Fig. 3). After 1 year, removal of 99 Tc sol from solution ranged between 6-73% and this increased to 25-95% following incubation for 2.5 yr. As shown in Fig. 3, only acidic, organic-rich soils (OC > 11 %; e.g. DY-M) appeared to be close to a steady state with respect to 99 Tc sol whilst the remaining soils were still losing soluble 99 Tc at a slow rate after 2.5 years.
Long-term, time-dependent transformations of 99 Tc also appear to be strongly governed by soil properties. Specifically, soil pH and OC were found to play a major role in controlling both the extent and the rate of these transformations (Fig. 2). Regression analyses suggested that the long term dynamics of 99 Tc sol can be principally explained (70-75%) by pH and C org , whilst the role of other potential binding surfaces such as Fe, Mn or Al oxy-hydroxides is marginal, explaining < 7 % of the variation in 99 Tc sol . Thus, 99 Tc largely remained in soluble forms in most arable and grassland soils, with a consistent but slow decline throughout the experiment (Figs. 1a and 3). Total 99 Tc sol loss after 2.5 yr ranged from 22 % for a calcareous arable soil (SR-A) to 95 % for a peat-rich grassland, soil (DY-G), although most values were in the range 40-50%. Abdelouas et al. (2005) reported a 30 % loss of soluble 99 Tc in an aerobic soil with 18 % organic carbon after 26 days, whereas no apparent loss of 99 Tc sol occurred until 72 days when soil organic carbon concentrations were < 6 %.
By contrast, sorption and migration of 99 Tc sol into non-soluble pools were the dominant processes in the majority of the woodland and moorland soils we studied. A substantial drop in 99 Tc sol (15-77% loss of total 99 Tc added) was seen within 49 days in these soils ( Fig. 1-3). This was followed by a slower additional loss of 99 Tc sol over the following months, reaching a maximum of 95 % in soil DY-M. Thus, in organic, acidic soils only 5-20% of the added 99 Tc remained soluble 2.5 years after initial contamination of the soils. However, for calcareous woodland soils approximately 65 % of initially added 99 Tc remained soluble at the end of the experiment.

Changes in the speciation of 99 Tc sol
Speciation analysis following chromatographic separation indicated that there were no statistically significant differences between 99 Tc sol and 99 TcO 4 − sol concentrations across the dataset (Fig. S3). Thus, pertechnetate was the dominant species in the soil solution for all soils throughout the incubation. This observation is of key radioecological relevance given that pertechnetate can be transported at 90 % of groundwater velocity in aquifers (Icenhower et al., 2009) and is the only Tc species that can be taken up by plants in appreciable amounts (Van Loon, 1986); it thus has significant potential to enter the food chain. However, speciation analyses also revealed some evidence for the transformation of pertechnetate in the incubated microcosms, in the form of consistent but unidentified peaks, with a strongly pH-dependent morphology, in the chromatograms of Tc sol . The concentrations of these unknown 99 Tc species in solution remained < 1 μg kg -1 (< 1 % total Tc) throughout the experiment.
For acidic and near-neutral soils, a series of three consecutive, but partially overlapping, sharp peaks ( Fig. 4) with short column retention times (12-50 s) were present from as early as 4 days incubation. The concentrations increased over time to asymptotes of 0.02 -1.1 μg kg −1 at 200 days. The very short transit times through the chromatographic column indicate that the newly formed 99 Tc sol species emerged along with the mobile phase suggesting virtually no retention by the anion exchange resin. This may be attributed to the occurrence of 99 Tc in cationic or neutral forms, such as the sparingly soluble 99 Tc IV O 2 ·nH 2 O aq , following reduction of 99 Tc VII to 99 Tc IV , which is likely in acidic, high organic soils as Eh-pH diagrams suggest (Takeno, 2005). Enhanced solubility of 99 Tc IV O 2 ·nH 2 O aq through formation of polymers or colloids has been reported (Maes et al., 2004). Formation of cationic species such as 99 Tc (IV) O(OH) + , following reduction, could also account for the poor affinity for the resin although Tc stability diagrams (Takeno, 2005) suggest this would only be likely in highly acidic soils. Complexation of reduced 99 Tc with organic ligands is known to enhance solubility (Maes et al., 2004) and the DOC concentrations measured in the KNO 3 extracts (up to 1600 mg kg −1 ) suggest that formation of organic complexes including 99 TcO(OH)-HA and 99 TcO(OH) 2 -HA is possible (Boggs et al., 2011). However it is unlikely that the rapidly emerging unknown peaks reflect these water soluble humic complexes given their negatively charged nature.
Peak broadening and tailing in chromatograms were observed with increasing soil pH (Fig. 4) suggesting co-elution of a number of pHcontrolled species. Technetium extracted from calcareous soils displayed a distinctive broad peak with longer retention times (40-50 sec) suggesting that pH > 7 could lead to formation of Tc species such as carbonate complexes. The presence of carbonate ligands in groundwater has been reported to enhance the formation of soluble 99 Tc IV -carbonate complexes (Wildung et al., 2000) which can be an important transport pathway. 99 Tc IV (CO 3 )(OH) 2 aq is thought to be the main species in bicarbonate media over a large range of chemical conditions and it is predicted to be stable across a wide pH range from 2 to 8 at high P CO2 , whilst 99 Tc IV (CO 3 )(OH) 3 − aq dominates at pH > 8 (Alliot et al., 2009). However, this species is not expected to be dominant in soils with low P CO2 .

Modelling 99 Tc sol kinetics
Removal of 99 Tc from solution was found to conform to a dual rate, double exponential model (Figs. 1a and 3). Pearson correlation coefficients (r) for model fits (see Table S1) and comparison between observed and modelled 99 Tc sol for all soils within the dataset (Fig. S2) showed that the model accurately fitted the experimental data throughout. Even for the weakest model fits, for the arable soils, SR-A (r = 0.543) and WK-A (r = 0.448), the significance of the fits was > 99.99 % (Table S1) indicating that 99 Tc sol declined significantly with time in all soils. Greater scatter of points around the line of unity in Fig.  S2 was seen for higher concentrations of 99 Tc sol -i.e. at shorter incubation times before substantial removal of 99 Tc from soil solution had occurred. Thus, the model fit to the experimental data was generally closer over longer incubation times. Significant correlations between the kinetic parameters (A, B, C, k fast and k slow ) and soil properties were found (Table 3 and Fig. S4-S5). Strong positive relationships between pH and model parameters A and C indicate that soil pH played a key role throughout the experiment, controlling the rapid 'fixation' of 99 Tc at the earliest stage of the incubation and the persistently soluble 99 Tc remaining after 897 days. Lower pH appeared to enhance rapid depletion of 99 Tc sol (as measured by parameter A) at the expense of lower residual 99 Tc sol (indicated by parameter C, Table 3).
It is worth noting that model fits for soils with pH > 6.8 yielded A = 0 (and thus k fast = 0) so that these soils conformed to a single exponential model controlled by slow removal of 99 Tc from solution (k slow ) (Fig. 3).

Depletion period
The 'slow' rate coefficient of 99 Tc sol depletion in Eq. 1 (k slow ) was used to estimate the time required to deplete 99 % of the soluble Tc fraction (Eq. 3): where k slow (t −1 ) is the first-order rate coefficient representing the 'slow' rate of depletion of B. Thus, the parameter T 99 provides an indication of the time taken for Tc sol to approach the persistently soluble concentration, C, which is probably still subject to slow removal from solution but which is not observable in many soils on an experimental timescale of less than a decade or more. T 99 ranged between 1.1 and 27.0 yr and showed a clear relationship with land use: increasing soil OC contents from arable (median T 99 = 6.3 yr) to grassland (median T 99 = 3.9 yr) to woodland (median T 99 = 3.5 yr) shortened the timeframe needed to reduce the 99 Tc sol fraction B to 1 % (Fig. 5). A strong, significant relationship between T 99 and B indicates that the depletion period is also controlled by the size of the pool to be slowly  (Table 3). Soluble 99 Tc in only 6 soils appeared to approach C within or very close to the incubation timeframe (≤2.7 yr after contamination with 99 Tc sol ).

Specifically adsorbed Tc ( 99 Tc ads )
Phosphate competes strongly with anions specifically adsorbed by Fe, Al and Mn oxide surfaces in soils and so methods utilizing competition with phosphate have been used to estimate, for example, plantavailable sulphate and selenite (Stroud et al., 2012). It is reasonable to assume that phosphate will exchange with other oxyanionic species, such as pertechnetate, which are weakly adsorbed in a similar way to selenate and sulphate. Speciation analysis indicated that pertechnetate ( 99 TcO 4 − ads ) was the dominant phosphate-extractable species ( 99 Tc ads ) for all soils throughout the long-term incubation (Fig. S3).
The 99 Tc ads fraction remained relatively small and did not exceed 15 % of the total 99 Tc throughout the incubation period. Soil properties appear to control the maximum concentrations of 99 Tc ads (Figs. 1b and  6 a). For arable and grassland soils, changes in 99 Tc ads over time followed a distinctive trend, gradually rising following addition of 99 Tc, peaking at 200-400 days (up to 9 μg kg −1 ) and subsequently stabilising or declining very slowly for the remainder of the experiment. Acidic, organic-rich woodland and moorland soils showed greater concentrations of 99 Tc ads and more rapid transfer of 99 Tc sol to 99 Tc ads , reaching a maximum concentration of 16 μg kg −1 of 99 Tc ads (15 % of total added) at ca. 200 days before starting to decline thereafter. There were generally weak correlations between the formation of 99 Tc ads and the abundance of potential sorption surfaces, including Al, Mn and Fe oxides, which may reflect the complexity of Fe-99 Tc interactions.

Tc org kinetics
The organic (TMAH-extractable) pool of 99 Tc ( 99 Tc org ) varied widely across all soils and throughout the experiment, ranging from 2 to 70 μg kg −1 which accounted for 2-65 % of the added 99 Tc (Fig. 1c). Up to 25 μg kg −1 99 Tc org (c. 23 %) was measured as early as 4 days after spiking highly organic soils, indicating that Tc can undergo rapid transformation to organic forms in soils. Changes in 99 Tc org over time reflected land use and soil characteristics, with the extent of the transformations strongly influenced by both pH and OC (Fig. 6b). For low to moderately organic, arable and grassland soils, the incorporation of 99 Tc sol into the organic pool slowly increased over time and appeared to reach an asymptote at 400 days (Fig. 1c). Median 99 Tc org concentrations observed were 2-30 μg kg −1 ; thus up to 30 % of the total 99 Tc originally added was organically bound after 897 days. Evidence for faster reaction rates was found for acidic woodland soils in which high organic carbon and low pH enhanced 99 Tc org formation. As shown in Fig. 1c, 99 Tc org increased most rapidly within the first 200-300 days in these soils, after which the rate of increase reduced in accordance with Eq. 2.

Modelling 99 Tc org
A dual rate 'ingrowth' model (Eq. 2) was fitted to data showing accumulation of 99 Tc org over time (Fig. 1c). Pearson correlation coefficients (r) for model fits (Table S2) for all soils within the dataset showed that the model accurately fitted the experimental data throughout. Even for the weakest model fit, grassland soil TK-G (r = 0.606), the significance of the fits was > 99.99 % (Table S2) indicating that 99 Tc org increased significantly with time in all soils. The comparison between observed and modelled 99 Tc org for the full dataset showed slightly greater scatter around the line of unity than was observed for the model describing 99 Tc sol kinetics (Fig. S2). This probably reflects the complexity of 99 Tc transformations through a range of processes with different dependencies on soil characteristics. A weak yet significant correlation between pH and the 'fast' rate coefficient (k 1 ) was observed (Table 3, Fig. S6). Strong and significant relationships between the kinetic parameters describing change in 99 Tc sol and 99 Tc org indicate that both reservoirs are closely connected (Table 3, Fig. S7).

Inaccessible Tc ( 99 Tc sink)
A variable though consistent increase in 99 Tc sink was observed for the majority of soils during the course of the long-term experiment (Fig. 1d); concentrations of 99 Tc sink broadly reflected OC and, to a lesser extent, pH as the data in Fig. 6c suggest. Thus, for arable and grassland soils, 99 Tc sink slowly increased over time and only up to 20 % of the added 99 Tc was immobilised after 897 days. Greater OC contents in woodland soils enhanced the formation of 99 Tc sink (Fig. 1d); an initially sharp increase to 20 μg kg −1 within the first 50 days was followed by a slower increase over time, reaching up to 40 μg kg −1 at 897 days. By the end of the experiment, between 13-40 % of the added 99 Tc sol was present in recalcitrant form(s) across all soils studied, becoming progressively inaccessible to the extraction methods we used. M. Izquierdo, et al. Journal of Hazardous Materials 388 (2020) 121762 4. Discussion

Short term 99 Tc dynamics
The rapid drop in 99 Tc sol following initial contamination reflects a range of processes operating simultaneously. Abiotic mechanisms include electrostatic interactions occurring almost instantly and largely controlled by pH, which is indicated by the strong negative relationship observed between the modelled kinetic parameter A (i.e. the fraction of 99 Tc sol subject to rapid removal from solution) and soil pH. However, in general, anions are poorly sorbed in soils due to low abundance of anion exchange sites. Statistical analysis of our data suggested that Fe/ Mn/Al oxy-hydroxides play a minor role in 99 Tc sol behaviour, which is consistent with the literature. Pertechnetate is known to adsorb only weakly onto mineral surfaces. Kaplan (2003) and Abdelouas et al. (2005) reported no pertechnetate sorption on mineral particles including quartz, clays and calcite in three sterilised soils. Previous studies detected only marginal sorption on Fe, Al and Mn oxides as well as a range of clays among other silicates (Palmer and Meyer, 1981). The 99 Tc ads pool remained < 20 % of the initially added 99 Tc throughout the experiment. Such low 99 Tc ads alongside the rapid decline in 99 Tc sol provides strong evidence that abiotic, surface-charge mediated adsorption on sesquioxides is limited and loss of 99 Tc sol is largely due to other mechanisms.
Our observations indicate that 99 Tc has a strong affinity for OC. Fast 99 Tc sol to 99 Tc org transfers following contamination strongly suggest that Tc rapidly sorbs onto humus. Carboxyl groups on humic surfaces may be protonated, therefore less negatively charged, at low pH thus reducing repulsion of anions and potentially enabling some degree of electrostatic sorption. However, it is highly likely that 99 Tc sorption or incorporation into organic matter is the result of more complex mechanisms largely driven by the rapid reduction of pertechnetate enhanced by organic groups (e.g. hydroquinones), followed by binding to aromatic carbon. Reduction of 99 Tc VII in humic-rich environments has been reported to form 99 Tc IV and 99 TcO 2 ·nH 2 O associated with humic substances in the solid phase (Maes et al., 2004); there is also evidence for 99 Tc IV forming binuclear complex compounds with carboxyl ligands of organic matter (Maes et al., 2004). Different 99 Tc sol to 99 Tc org transfer rates following contamination may reflect differences in the composition and chemistry of humic substances present in soils, which will result in substantially different densities of functional groups and therefore different binding capacities. This has been known for some time but the mechanisms responsible have not been fully elucidated due to the complexity and variability of humic substances (De Paolis and Kukkonen, 1997). The positive correlation of OC with A (rapidly removed fraction of 99 Tc sol ) indicates that OC, in conjunction with low pH, accelerates transfer of soluble 99 Tc sol into the organic reservoir. The  Table 1. close interplay between 99 Tc sol and 99 Tc org is further illustrated by positive relationships between A from Eq. 1 and R 1 and R 2 from Eq. 2, indicating that greater proportions of rapidly removed 99 Tc sol (A) result in faster rates of formation of 99 Tc org (R 1 and R 2 ) (Fig. S7).
The literature suggests that biological reduction of Tc VII to Tc IV in nature is a slow process even under low E h conditions (Icenhower et al., 2009). Therefore, it is reasonable to assume that the 99 Tc behaviour within the first hours following contamination is primarily abiotically driven. However, it is highly likely that other processes with a biotic component, such as enzymatic activities, may also operate. Thus, the presence of reactive enzymes in soils may result in rapid removal of 99 Tc sol and direct 99 Tc sol to 99 Tc org/sink transfers through enzymatic 99 Tc VII reduction and subsequent Tc IV O 2 ·nH 2 O formation. This could contribute to the rapid formation of 99 Tc org and 99 Tc sink observed during the initial stages of the incubation experiment.
It is generally recognised that reduction reactions in soils are enhanced at lower pH and that enzyme activity is greater in soils with large humus concentrations. Thus, the immediate removal of 99 Tc sol appears to be marginal in soils with pH > 6.8. The absence of a fast component in the model (Eq. 1) fits for calcareous soils indicate that 99 Tc sol transfer to sorbed forms is dominated by slow kinetics in these soils. Therefore, high pH in conjunction with relatively low organic carbon (OC < 7.5 %) produced relatively low rates of 99 Tc fixation. This is of key environmental relevance given that rendzina soils overlying karstified calcareous terrains would be inefficient in retaining deposited 99 Tc and would not prevent transport, dispersion and discharges to the underlying aquifer.

Long term 99 Tc dynamics
Electrostatic adsorption onto Fe, Mn and Al oxides apparently played a minor role throughout the experiment. Low pH increased the transfer of 99 Tc sol to 99 Tc ads , which peaked during the first few months of the experiment. This was followed by a slower decline in 99 Tc ads , probably associated with active transfer to the 99 Tc org and 99 Tc sink pools. There are no long-term binding mechanisms for pertechnetate; for 99 Tc to become strongly adsorbed a change in Tc speciation must occur. Pertechnetate can be removed from solution ( 99 Tc sol ) or desorbed from weak sorption sites ( 99 Tc ads ) through reduction to 99 Tc IV and formation of the sparingly soluble 99 Tc IV O 2 ·nH 2 O, the most common 99 Tc IV species that is stable across the pH range encountered in soils. Reduction can occur through a number of mechanisms and, specifically, the ability of microorganisms to reduce Tc has been well documented in the literature. Comparative studies conducted on sterilised and non-sterilised batches of soil have demonstrated the crucial role of bacterial populations in 99 Tc speciation and immobilisation (Abdelouas et al., 2005;Burke et al., 2005). The most extensively documented mechanism is direct enzymatic reduction of 99 TcO 4 − by anaerobic metal-reducing and sulphate-reducing bacteria including Escherichia coli, Geobacter sulfurreducens, Geobacter metallireducens, Shewanella putrefaciens or Desulfovibrio desulfuricans (Abdelouas et al., 2005;Icenhower et al., 2009;Lloyd, 2003;Lloyd et al., 2000;Wildung et al., 2000), with insoluble 99 Tc IV as the final product. Technetium reduction is not necessarily caused by reducing conditions in the growth medium but is the result of a metabolic process (Henrot, 1989) whereby microbially mediated biosorption of 99 Tc on bacterial cells removes 99 Tc from solution (Abdelouas et al., 2005). Another potential biotic pathway for 99 Tc immobilisation involves indirect reduction of  (Henrot, 1989;Lloyd, 2003;Lloyd et al., 2000). Biogenic Fe II has been recognised as particularly efficient in indirectly reducing 99 Tc (Lloyd, 2003). Whilst hematite typically sorbs 99 Tc rather weakly, the reactivity of Fe III towards 99 Tc VII can increase in the presence of specific microorganisms through bioreduction of Fe III and subsequent 99 Tc VII reduction (Druteikienė et al., 2014). Abiotic pathways reported in the literature include interaction with Fe-bearing soil minerals, which can increase 99 Tc sorption under aerobic conditions. Surface mediated reduction of Tc through interaction with abiotic Fe II sorbed onto Fe-bearing minerals in soils or, to a lesser extent, by aluminosilicates has also been reported as an effective scavenging mechanism (Lloyd et al., 2000). The presence of structural Fe II can enhance Tc sorption onto pyrrotine Fe 1-x S (Shen et al., 2002), magnetite (Fe 3 O 4 ) or wustite (FeO) (Druteikienė et al., 2014) through abiotic transfer of electrons and subsequent Tc reduction, even in the presence of oxygen. Many of the above mechanisms have been widely reported in anaerobic environments whilst aerobic cultures of E. coli (Lloyd, 2003) or Desulfovibrio sp. (Henrot, 1989) were not found to induce changes in 99 TcO 4 − solubility or to bioaccumulate 99 Tc. Pertechnetate reduction requires reducing conditions which are common in waterlogged sediments or hydrogeological environments below the phreatic surface, but not frequent or persistent in agricultural soils where aerobic conditions are expected to prevail. However, it is likely that the processes described above operated to some extent in our incubation experiment. Although the microcosms were incubated under aerobic conditions and periodically aerated, the soils were biologically active and contained decomposable organic matter with the capacity to fuel microbial activity. In the presence of moisture (soils remained moist throughout the experiment - Table 1) and soil organic matter as an electron donor, the O 2 supply can become depleted in microsites within the interstices of soil particles in which gas exchange is restricted. This may locally and intermittently induce reducing conditions as O 2 diffusion through a layer of water may be rate-limiting (Icenhower et al., 2009). In addition, local accumulation of CO 2 derived from microbial respiration may induce acidification in these microenvironments, thus enhancing abiotic sorption onto humus and other binding sites. Almost every soil is thought to contain anaerobic microsites (Van Loon, 1986). Abdelouas et al. (2005) stated that degradation of organic matter by indigenous bacteria within soil grains creates localised anaerobic conditions in otherwise aerobic soils that promote reduction and precipitation of Tc. Evidence for 99 Tc binding has also been reported in intermittently wet farmland soils under net oxidising conditions (Icenhower et al., 2009;Tagami and Uchida, 1996). Of all the mechanisms suggested above, reduction via biogenic Fe II seems unlikely as a dominant process in our microcosms, primarily because Fe II has a stronger tendency to react with O 2 than with 99 Tc (Icenhower et al., 2009) and our soils were in a net aerobic state. We also observed consistent Fe sol and Mn sol concentrations over 897 days (Fig. S8) which indicate minimal activity of Fe III -reducing microorganisms. Thus, although no redox potential measurements were taken, there was no evidence for significant changes in the electrochemical status of the 20 soils or of microbially mediated ingrowth of Fe 2+ and Mn 2+ . This suggests that, following exhaustion of O 2 , any further reductive processes were limited to microenvironments and may not have progressed much further down the redox ladder or have been widely spread across the microcosm, whilst the bulk of the soil remained aerated. Most works in the literature have studied 99 Tc reduction in anoxic environments with redox potentials lying within the zone of reductive dissolution of FeOx and MnOx; very few have addressed the potential for 99 Tc reduction in more aerated soils. Abdelouas et al. (2005) stated that, in nitrate-rich environments such as agricultural soil, Tc VII reduction was not possible and no biosorption of 99 Tc on cells of the denitrifying bacterium Pseudomonas aeruginosa occurred during nitrate reduction. However, Maset et al. (2006) observed 99 Tc reduction and removal from solution concurrently with nitrate reduction in soil microcosms with 12 % OC. In addition, Istok et al. (2004) found evidence for bioremoval of 99 Tc VII in tandem with nitrate reduction from nitrate-rich groundwater in the presence of organic C. This supports the notion that (i) biologically mediated 99 Tc VII reduction in aerated microcosms can occur via enzymatic reduction at redox potentials before the onset of Fe reduction and (ii) 99 Tc reduction may be kinetically restrained by the presence of labile organic matter to fuel bacterial activity. It is therefore highly likely that the longer-term decline in 99 Tc sol over time in our microcosms is associated with enzymatic reactions and microbially-mediated transformations enhancing (i) migration into organically-bound pools ( 99 Tc sol to 99 Tc org ) and direct immobilisation ( 99 Tc sol to 99 Tc sink ). Strong, highly significant correlations between soil properties and the fraction of 99 Tc that remains in solution after 897 days i.e. C (Fig. S5) suggest that the rates of these microbially driven reactions are enhanced by low pH and high OC. The slow rate coefficient of 99 Tc sol depletion (k slow acting on fraction B in Eq. 1) would potentially be the rate limiting parameter for slow, longer term 99 Tc sol to 99 Tc org/sink transformations to occur.
Changes in 99 Tc org reflected a complex combination of gains from Tc sol and Tc ads and losses to Tc sink over the course of the (Fig. 1). The soluble and organically bound pools are also closely interlinked in the long term as the significant relationships between model kinetic parameters for changes in 99 Tc sol and 99 Tc org suggest (Fig. S7). Thus, greater proportions of rapidly removed 99 Tc sol (A) were associated with increased 'fast' and 'slow' rates of 99 Tc org formation (R 1 and R 2 ). Low residual 99 Tc sol fractions (C parameter in Eq. 1) were also indicative of greater rates of 99 Tc org formation (R 1 and R 2 ) in the long term (Fig. S7) (Ashworth and Shaw, 2005) and evidence for the formation of Tc species associated with humic substances following pertechnetate reduction has been found using spectroscopic techniques (Maes et al., 2004). Organic matter and its ability to form strong complexes play a critical role in the progressive immobilisation of 99 Tc. 99 Tc org to 99 Tc sink transfers would be driven by complex interactions between 99 Tc IV following pertechnetate reduction and humic substances present in soil including metal-humate complexation and subsequent precipitation (Sekine et al., 1993), oxidic polymer colloid formation followed by interaction with humic substances (Maes et al., 2004) or formation of organic coatings on mineral surfaces (Koch-Steindl and Pröhl, 2001).

Reliability of model predictions
We used an 'offset' double exponential kinetic model to represent the time-dependent decline in soluble forms of 99 Tc in soils. The slow component of the kinetic system (i.e. the slow rate coefficient, k slow ) enabled estimation of the time required to reduce the soluble 99 Tc sol subject to slow removal (B) by 99 % (T 99 ). This addresses the second of the two questions tackled in this study -over what experimental timescales must we measure Tc reaction kinetics in order to make reliable long-term predictions of 99 Tc impact in the terrestrial environment?
A major control on T 99 was found to be the size of the fraction to be removed, with lower values of B (Eq. 1) shortening the period required to reduce 99 Tc sol to 1 % of B. Thus, values of T 99 shorter than the period of the incubation experiment were observed in peaty, acidic and highly reactive soils where immediate fixation processes are dominant, i.e. those soils with a large A fraction in relation to B and C (e.g. DY-G, DY-M). To a lesser extent, other near-neutral to acidic woodland and grassland soils with sufficient OC to prompt fast 99 Tc sol/ads to 99 Tc org transformations and enhance the magnitude of the A fraction have relatively low depletion periods of up to 3.5 yr (Fig. 5). These estimates can be considered to be reasonably reliable given that (i) model fits were very good for these soils, especially for the longer term experimental data and (ii) 99 Tc sol in these soils was low towards the end of the experiment and appeared to be approaching the persistently soluble concentration, C (Fig. 3).
Depletion times shorter than the incubation experiment were also observed for poorly organic, calcareous soils (e.g. SR-A and WK-A) in which 99 Tc sol also appeared to be approaching the persistently soluble concentration. High pH and low OC (i) do not enable fixation (i.e. A = 0) and (ii) lead to poorly reactive soils with limited capacity for long term 99 Tc transformation and binding, with the result that 99 Tc removal from solution was marginal and 99 Tc largely remained as persistently soluble 99 Tc sol over the long term (i.e. C≥80 %, Fig. 5).
By contrast, the experimental data suggest that near neutral, moderately organic soils are far from reaching steady state with respect to soluble 99 Tc within the duration of the experiment. Prediction of timeframes for reduction in 99 Tc solubility may not be accurate for soils actively losing soluble 99 Tc sol beyond the 2.5 yr incubation period, given that changing reaction rates may introduce uncertainty in the estimates. However, our findings provide solid evidence that short-term experiments (less than 1 or 2 years) may not be used reliably to predict the long-term/equilibrium solubility and immobilisation of 99 Tc and, potentially, other radionuclides characterised by slow sorption rates. This is of particular relevance in the context of long-term radiological risk assessments.

Conclusions
The removal of 99 Tc from solution was very slow in relation to other radionuclides and our observations suggest that steady state conditions were not attained in the majority of soils, even after 2.5 years incubation. Reduction in soluble 99 Tc over time followed a dual rate, double exponential kinetic model and included (i) a fast component reflecting immediate removal enhanced by low pH; (ii) a slowly removed fraction largely controlled by pH and organic matter; and (iii) a persistently soluble Tc fraction. The slowest rates of sorption were seen in calcareous soils, which clearly highlights potential risk to limestone and chalk aquifers; the rendzina soil types overlying such aquifers would fail to intercept deposited 99 Tc. Soil polyvalent metal oxides only played a marginal role in sequestering 99 Tc. By contrast, a combination of high OC concentrations and low pH produced faster sorption kinetics. The ability of soil humus to form strong complexes with 99 Tc is key to the progressive immobilisation of 99 Tc in soil. Our data suggest that shortterm experiments (< 1-2 years) may not be used reliably to predict long-term solubility and bioaccessibility of 99 Tc, particularly for soils with low to moderate OC contents. Our findings are of practical significance in risk assessment calculations for facilities such as radioactive waste repositories, whether based on generic or site-specific criteria. While our experimental study has not considered the fate and behaviour of 99 Tc in every possible soil type, the range of physicochemical characteristics examined is wide enough to encompass the environmental circumstances under which many risk assessments will be made.

Author contributions
EB, GS, NC and SY conceived and planned the experiments. EB, GS, MI and SY sampled the soils and set up the incubation experiment. MI and HS carried out the sampling and analyses. MI processed the experimental data and carried out interpretation of results with help from GS and SY. GS performed the model fittings. MI took the lead in writing the manuscript but all authors discussed the results, provided critical feedback and helped shape 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.