Incorporation and Retention of 99-Tc ( IV ) in Magnetite under High pH Conditions

Technetium incorporation into magnetite and its behavior during subsequent oxidation has been investigated at high pH to determine the technetium retention mechanism(s) on formation and oxidative perturbation of magnetite in systems relevant to radioactive waste disposal. Ferrihydrite was exposed to Tc(VII)(aq) containing cement leachates (pH 10.5−13.1), and crystallization of magnetite was induced via addition of Fe(II)aq. A combination of X-ray diffraction (XRD), chemical extraction, and X-ray absorption spectroscopy (XAS) techniques provided direct evidence that Tc(VII) was reduced and incorporated into the magnetite structure. Subsequent air oxidation of the magnetite particles for up to 152 days resulted in only limited remobilization of the incorporated Tc(IV). Analysis of both X-ray absorption near edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) data indicated that the Tc(IV) was predominantly incorporated into the magnetite octahedral site in all systems studied. On reoxidation in air, the incorporated Tc(IV) was recalcitrant to oxidative dissolution with less than 40% remobilization to solution despite significant oxidation of the magnetite to maghemite/goethite: All solid associated Tc remained as Tc(IV). The results of this study provide the first direct evidence for significant Tc(IV) incorporation into the magnetite structure and confirm that magnetite incorporated Tc(IV) is recalcitrant to oxidative dissolution. Immobilization of Tc(VII) by reduction and incorporation into magnetite at high pH and with significant stability upon reoxidation has clear and important implications for limiting technetium migration under conditions where magnetite is formed including in geological disposal of radioactive wastes.


■ INTRODUCTION
Technetium-99 (Tc-99) is a long-lived radioactive fission product (half-life 2.1 × 10 5 years) present in many spent fuel and reprocessing derived radioactive wastes. The environmental mobility of Tc-99 in the subsurface is primarily governed by its oxidation state. Under oxidizing conditions, Tc(VII) is highly soluble (∼11 M) 1 and environmentally mobile, existing as the pertechnetate, TcO 4 − , anion. In contrast, under reducing conditions, Tc(IV) is dominant and generally precipitates as sparingly soluble Tc(IV)-hydroxide phases (10 −8 −10 −9 M) 2−4 or adsorbs to mineral surfaces. 5−8 Due to its high mobility under oxic conditions and its long half-life, Tc-99 is a significant risk driving radionuclide for geological disposal of radioactive wastes, as well as presenting a significant challenge to remediation of radioactively contaminated land 9 in the UK, 10,11 the USA, 12−14 and Russia. 15 Past work has shown that Fe(II) bearing oxides 16−18 and sulfides 19−21 are effective in removing Tc(VII) from solution via reductive adsorption or precipitation. In addition, Fe(II) adsorbed to mineral surfaces such as iron or aluminum (oxyhydr)oxides has also been shown to be a highly effective abiotic reductant for Tc(VII). 17,22,23 Microbially mediated reduction of Tc(VII) has also been documented, with enzymatic and indirect (via reaction with biogenic Fe(II)) pathways possible but with indirect reduction dominating. 6,24−26 Although aqueous Fe(II) is able to reduce Tc(VII), the kinetics of this pathway are relatively slow. 27 In general, the reductive ability of Fe(II) increases from aqueous Fe(II) to structural Fe(II) to adsorbed Fe(II). 9 Reflecting this, transport of Tc-99 in the environment will therefore be significantly controlled by the Tc-99 interactions with Fe(II) bearing solid minerals (e.g., iron oxides, sulfides) through either sorption 8,28 or surface mediated reduction and precipitation. 29,30 Exposure of reduced Tc-99-bearing sediments to oxidizing conditions may induce remobilization of Tc(VII) back into solution, although the extent of remobilization will be dependent upon both the binding mechanism and the oxidant. Sediments labeled with Tc-99 from authorized effluent discharges from Sellafield showed evidence that Tc-99 is retained on sediments over periods of decades even in dynamic saltmarsh environments. 31 Laboratory based studies on remobilization of Tc-99 in shallow sediments indicated that, after Fe(III)-reducing conditions had developed and Tc(VII) was reductively scavenged to sediments, upon oxidation a variable fraction of the sediment associated Tc-99 was recalcitrant to oxidative remobilization. 32−35 Under certain conditions, oxidative exposure to air caused remobilization of >50% of the sediment associated Tc(IV) over several months. 12, 33,36 Further, a study on iron minerals by Wharton et al. 19 reported Tc-99 remaining as Tc(IV) despite oxidation of associated mackinawite (tetragonal FeS) to goethite in air. In these studies, the potential existed for Tc(IV) to be incorporated within the oxidized secondary mineral phases, potentially contributing to its recalcitrance to oxidative remobilization.
Incorporation of technetium into stable mineral phases, such as iron (oxyhydr)oxides, offers a pathway for Tc sequestration with the potential for immobilization. Indeed, evidence for Tc(IV) incorporation into goethite (α-FeOOH) during coprecipitation has been reported, 37,38 and Tc(IV) incorporation to hematite (α-Fe 2 O 3 ) has been predicted from modeling. 39 Both these studies suggest that Tc(IV) can be octahedrally coordinated within the relevant crystal structure and can directly substitute for Fe(III). Products of steel corrosion (magnetite, maghemite, and goethite) have also been shown to be able to partially reduce Tc(VII) to Tc(IV), with the authors concluding the Tc-99 was incorporated into the corroded steel surface. 40 It has also been recently suggested that adsorbed Tc(IV) may, with time, become incorporated into the magnetite (Fe 3 O 4 ) structure. 41 It is clear that the reductive incorporation of Tc-99 as Tc(IV) into magnetite under environmental conditions relevant to contaminated sites and geodisposal of wastes could significantly reduce the mobility of Tc-99. However, the mechanism of Tc-99 incorporation into the structure of magnetite and the amount and mechanism of release during oxidation remains unknown.
Crystallization of ferrihydrite to magnetite proceeds via sorption of Fe(II) onto ferrihydrite followed by a solid-state transformation between ferrihydrite, goethite, and magnetite and with additional growth of already nucleated magnetite particles. 42−44 At low solution Fe(II) concentrations (<0.03 mmol/L), only goethite precipitates, with nucleation of magnetite increasing with increasing Fe(II) concentration and with magnetite crystallization eventually proceeding at the expense of goethite. 42−44 Magnetite oxidizes to maghemite (γ-Fe 2 O 3 ) topotactically through outward diffusion of Fe, with no change in the crystal size or shape and the creation of 2.67 cation vacancies per unit cell. 45 This process initially leads to the formation of a pure Fe(III) surface layer on the magnetite particle which increases in thickness until the entire particle has transformed to maghemite. 45 The identity of the Fe species undergoing outward diffusion during this process is not agreed upon in the literature, although Fe(II) is the more favored. 45−47 Understanding the potential for incorporation of technetium into magnetite during its formation and oxidation and under conditions relevant to radioactive waste disposal is the focus of this study.
Many geodisposal concepts utilize cementitious materials in the design of a deep geological disposal facility (GDF), for example, as part of the engineered barrier system or as structural materials. Leaching of these materials upon resaturation will buffer the pH to hyperalkaline conditions (pH 10.5−13.1), creating a plume of alkaline fluid in the host rock and/or local environment. 48 Furthermore, hyperalkaline conditions can prevail in many contaminated land scenarios, e.g., where cementitious building materials contact subsurface sediments 49,50 and at the underground waste storage tanks at the Hanford Site in Washington State, USA, where there has been considerable attention focused on behavior of Tc-99. 13,14,39,51−53 Additionally, microbial reduction of Fe(III) leading to the formation of magnetite has recently been shown to occur under alkaline conditions, 54 highlighting the importance of biogeochemistry in these processes and lending further significance to the work in this contribution.
The aim of this study was to characterize the mechanism(s) of Tc-99 reaction with magnetite as it crystallizes under alkaline conditions, to determine whether significant amounts of Tc-99 could be incorporated into this environmentally important phase and to explore the effects of oxidative perturbations on Tc-99 speciation and solubility. The system we have explored is crystallization of ferrihydrite as an initial Fe(III) oxyhydroxide product to magnetite by addition of aqueous Fe 2+ . 42 We followed the partitioning of Tc(VII) during this process in three synthetic cement leachates (pH 10.5, 12.5, and 13.1). After a period of aging, the Tc-99 doped magnetite systems were then exposed to oxidizing conditions to test the recalcitrance of any iron oxide associated Tc-99 to reoxidation and thus to explore the stability of the radionuclide. We have combined geochemical data with quantitative powder X-ray diffraction (QXRD), transmission electron microscopy (TEM) and energy dispersive X-ray spectroscopy (EDX), and X-ray absorption spectroscopy (XAS) to thoroughly characterize the fate of Tc-99 during magnetite crystallization and reoxidation at high pH. Throughout, we have focused on the fate of Tc-99: first, during ferrihydrite crystallization to magnetite, we determined the extent of Tc-99 incorporation and defined its atomic scale bonding environment; we then oxidized the Tc-99-labeled magnetite over several months to determine if incorporation offered protection against oxidative remobilization.

■ MATERIALS AND METHODS
Batch experiments were used to follow the Fe(II)-mediated crystallization of ferrihydrite to magnetite 42−44 in three Tc(VII) amended synthetic cement leachates (CL), chosen to simulate a young (YCL, pH 13.1), intermediate (ICL, pH 12.5), and old (OCL, pH 10.5) postclosure groundwater around a GDF. 48,55 Details on preparation of the cement leachates are given in the Supporting Information. Parallel experiments were also set up without Tc(VII) present to allow characterization of the solid products by X-ray diffraction (XRD). The 2-line ferrihydrite used as a starting material was synthesized as per Cornell and Schwertmann, 56 and the Fe(III) content of the resultant paste was determined by dissolution in 1 M HCl and analysis by the ferrozine method. 57 A known quantity of ferrihydrite was equilibrated with the cement leachates at a solid/solution ratio of 3.5 g L −1 for 1 h on an orbital shaker at room temperature in an anaerobic chamber (5% H 2 , balance N 2 , maintained at <1 ppm of O 2 and CO 2 ). After equilibration with the ferrihydrite, the pH was manually readjusted to the initial leachate pH by  (NH 4 TcO 4 ) to give an initial solution concentration of 3 ppm (30.3 μM, 2 kBq ml −1 ). Crystallization to magnetite was then induced by addition of Fe(II) as 0.1 M FeCl 2 , to give stoichiometric magnetite with an Fe(II)/Fe(III) ratio of 1:2. After Fe(II)-addition, the pH was immediately readjusted to the initial leachate pH using KOH. The alkaline magnetite suspension was then aged for 2 days in the anaerobic chamber. Following this, the experiments were moved to a separate glovebag containing CO 2 -free air (Zero grade Air) where they were allowed to oxidize for up to 152 days. A further experiment was set up in the intermediate stage cement leachate (ICL, pH 12.5) as a sorption "control" where the Tc-99 was spiked to preformed magnetite after 1 day. After a further day of reaction, a subsample of the experiment was then transferred to the glovebag containing CO 2 -free air for 152 days for oxidation. All experiments were shaken on an orbital shaker (150 rpm) to allow exchange with the relevant atmosphere (i.e., reducing: H 2 /N 2 mix; oxidizing: CO 2 -free air) but were not purged with the relevant gas. Minor aggregation of particles was observed in all three systems over time. The pH was periodically monitored throughout the experiments, and no further pH adjustment was required after the initial pH was set.
Partitioning of Tc-99 between the solid/solution was determined by analysis of a 1 mL solution aliquot for Tc-99 by liquid scintillation counting (LSC) using a Packard Tricarb 2100TR liquid scintillation analyzer. The distribution of Tc-99 within the magnetite particles and oxidized products was determined by total digestion in 1 M HCl where the relative dissolution rate of Tc-99 versus total Fe in the acidic leachate provided insight into the concentration gradient of Tc-99 through the solid. The mineral sample was allowed to fully dissolve in the acid, with the resultant solution analyzed at selected time points during dissolution. Samples were analyzed for Tc-99 using LSC and Fe by the ferrozine method 57 until no further increase in the concentration of Fe in the acidic solution was observed. Full details of the method are included in the Supporting Information. Where appropriate, selected samples were characterized by XRD using a Bruker D8 diffractometer (λ = Cu K-α1). For TEM, solid samples were characterized using an FEI CM200 Field Emission Gun TEM, operating at 197 kV and fitted with a Gatan Imaging Filter (GIF 200) and an Oxford Instruments Silicon Drift Detector EDX spectrometer running AZTEC software. Tc-99 K-edge XAS spectra were collected on Beamline B18, at the Diamond Light Source 58 at room temperature in fluorescence mode using a 9-element Ge detector. 59 Data reduction and fitting of the X-ray absorption fine structure (EXAFS) spectra were performed using the software packages Athena and Artemis. 60 ■ RESULTS AND DISCUSSION Characterization of the Iron Oxide Products. XRD patterns show that, after 2 days of aging, magnetite was the dominant product formed in each of the leachates (Figure SI-1, Supporting Information) but that in the YCL (pH 13.1) and ICL (pH 12.5) experiments goethite was also formed (QXRD analysis suggested up to 20−25%, Table 1). In contrast, the pH 10.5 OCL product was magnetite only (Figure SI-1, Supporting Information, Table 1). During the oxidation experiments, the color of solids changed from black to brown in all leachates suggesting that oxidation of the magnetite had occurred. After 21 days of oxidation in CO 2 -free air, the XRD patterns for all experiments were the same as the fully reduced samples within error and with no significant increase in the goethite present in the YCL and ICL systems and no measurable ingrowth of goethite into the OCL system (Figure SI-1, Supporting Information, Table 1). However, after 152 days of oxidation, all three samples contained approximately 50% goethite as expected for magnetite oxidation at high pH (Figure SI-1, Supporting Information, Table 1). 61 Oxidative maghematisation was evident from a shift in the magnetite (511) and (440) peaks in the XRD pattern to higher 2θ in the 152 days of oxidation in both the ICL and OCL experiments, in agreement with the QXRD data ( Figure SI-2, Supporting Information). However, the characteristic maghemite (210) and (211) peaks 62 were not observed ( Figure SI-2, Supporting Information), indicating only partial maghematisation has occurred. The length of the unit cell a axis dimension has previously been used as a measure of stoichiometry, i.e., Fe(II)/Fe(III) ratio. 63 Stoichiometric magnetite has an Fe(II)/Fe(III) ratio of 0.5 which decreases to zero in maghemite, while the unit cell length decreases from 8.396 to 8.340 Å. 56,64 In the 152 day samples, a shortening of the unit cell length was observed in all three systems indicating a decrease in the Fe(II)/Fe(III) ratio consistent with oxidation (Table 1). Using the stoichiometric method of Gorski and Scherer, 63 we have calculated the Fe(II)/ Fe(III) ratio and, hence, the degree of maghematization, for our samples (Table 1). After 152 days of oxidation, complete oxidation of the three experiments had not occurred; the lowest Fe(II)/Fe(III) ratio and hence greatest extent of maghematization was observed in the ICL system (0.12 ± 0.03), while the least extent of maghematization was in the YCL system (0.39 ± 0.05). Complete oxidation of maghemite nanoparticles in aqueous systems has been observed to occur over approximately three months 46 suggesting that, in our alkaline systems, magnetite has a degree of recalcitrance to oxidation. Incorporation of trace elements into magnetite (e.g., Co, Ni, Zn, up to 1 wt %) has previously been suggested to stabilize Fe(II) and suppress maghematisation through a decrease in electron mobility. 45 The incomplete maghematization observed in our samples despite almost double the oxidation time compared to the past study 46 could potentially indicate incorporation of Tc(IV) and stabilization of Fe(II) within the magnetite structure in the immediate vicinity of Tc(IV) as a charge compensation mechanism and which may inhibit complete oxidation. TEM images (Figures SI-3 and SI-4, Supporting Information) confirmed the presence of crystalline nanoparticulate magnetite (angular to rounded cubes and rhombs, 5−20 nm in size) and goethite (needles) in the reaction products. Tc-99 was not detected using energy dispersive X-ray spectroscopy (EDX) in any of the samples analyzed, suggesting the Tc-99 was distributed throughout the sample rather than as a discrete phase and/or the Tc-99 concentration was below the limit of detection for this technique.
Technetium removal from solution to the solid during magnetite formation and oxidation is shown in Figure 1.
Regardless of pH, all Tc-99 was removed from solution to below detection limit (ca. 0.5 Bq ml −1 ; 0.8 ppb) within minutes after the addition of Fe(II) and with no release back to solution during aging. Upon oxidation in air, there was an initial rapid remobilization of Tc-99 to solution over the first day with a continued slow release over the first 2 weeks; however, the Tc-99 concentration in solution then remained essentially constant. The total amount of oxidative release of Tc-99 was dependent upon the solution composition. The Tc-99 released in the OCL (pH 10.5) and ICL (pH 12.5) systems was similar with ∼20% of Tc-99 released after 1 day of oxidation rising to ∼30% after 152 days. In the YCL, the release profile was similar, but the total release was higher, at ∼30% after 1 day rising to ∼40% after 152 days. The greater extent of Tc-99 released to solution in the YCL compared to the ICL and OCL may be due to increased oxidation of the magnetite surface at higher pH. 61 In addition, differential aggregation of the particles could potentially influence the Tc(IV) remobilization behavior. However, the particle aggregation on reoxidation was similar for each leachate and seems not to explain the differential remobilization across the different cement leachates and the persistently greater remobilization in the YCL. In contrast, in the adsorption experiment where Tc-99 was adsorbed to preformed magnetite at pH 12.5 in ICL, greater oxidative remobilization was observed compared to the coprecipitated experiments with ∼70% of the Tc-99 oxidatively remobilized to solution after 152 days. In the reoxidation experiment with the adsorbed sample, the remaining ∼30% of Tc-99 was in some way recalcitrant to air oxidation and was retained by the solid phase even though Tc(VII) was exposed to the preformed Fe(II)-bearing mineral. Interestingly, similar behavior has been observed for reduction of Tc-99 onto preformed magnetite, where EXAFS analysis suggested some incorporation of Tc-99 into the magnetite after aging. 41 Technetium Distribution within the Solids. Magnetite dissolves isotropically allowing the distribution profile of trace elements within the solid to be assessed by acid digestion. 65,66 We performed 1 M HCl digestions on the samples from the OCL (pH 10.5) experiment to explore the Tc-99 distribution profile in the solid since these contained negligible goethite implying that all Tc-99 measured would be from magnetite associated Tc-99 ( Figure 2). The dissolution profile of Tc-99 vs Fe in the reduced sample had a gradient initially greater than 1 which then fell to less than 1 when more than 50% of the available Fe was dissolved. This indicated that Tc-99 was not uniformly distributed through the magnetite particles and was relatively concentrated at the surface, an effect previously observed for Mn(II) and Cu(II). 65,66 The OCL oxidized mineral sample from 21 days, which showed no ingrowth of secondary goethite, showed that approximately 10% of the Fe in the sample was dissolved before any Tc-99 was detected in   solution, suggesting the development of an Fe rich/Tc-99 poor rind during oxidation. After this, the Tc-99 and Fe profiles were similar to the magnetite sample with an initial slope greater than 1, and after approximately 50% Fe dissolution, the gradient fell to below 1 reflecting an apparently heterogeneous distribution of Tc-99. From the Tc-99 distribution profile of the reduced OCL magnetite sample, 30% Tc-99 was dissolved in the acid by the time only 10% Fe had dissolved (Figure 2) suggesting that 30% of the Tc-99 released to solution during oxidation (Figure 2) can be explained by release of Tc-99 from the outer 10% of the 1 M HCl leachable magnetite particles.
X-ray Absorption Spectroscopy. X-ray absorption near edge structure (XANES) and EXAFS spectra were collected from the reduced magnetite and two oxidized samples (21 and 152 days) from coprecipitation experiments in each cement leachate, along with the sorption sample in ICL (pH 12.5). It was not possible to collect XAS data for the 21 day oxidized sorption sample as it was found that the Tc-99 concentration was too low for analysis. The edge position of all the XANES spectra aligned to the Tc(IV) reference, 67 and there was no evidence in any samples for the distinctive Tc(VII) pre-edge peak at 21045−21050 eV (Figure 3). A linear combination fit between the oxic and reduced reference spectra resulted in no (0%) contribution from Tc(VII) to each of the experimental spectra (Table SI-1, Supporting Information). This indicated that Tc(VII) was fully reduced to Tc(IV) in each system and that, for reoxidized samples, any Tc-99 that remained associated with the solid was retained as Tc(IV) throughout air oxidation. The distinct double peak observed in the near edge of each of the sample spectra, which is absent from the TcO 2 reference, has previously been attributed to the presence of Tc−Fe bonds. 40 The EXAFS spectra collected from samples in the ICL system (pH 12.5) are plotted in Figure 4, along with the nonlinear least-squares fits to the data. It is clear that the Tc-99 in the sorption sample (Figure 4) is in a very different coordination environment when compared to the coprecipitation samples, with a single wave dominating the k-space in the sorption system. The relatively weak second peak in the Fourier transform at approximately 3.5 Å suggests there are few absorbers/scatterers in the second shell suggesting the Tc-99 was predominantly present on the magnetite surface. Interestingly, all of the ICL coprecipitation samples are strikingly similar including the 21 and 152 day oxidized samples (Figure 4). This confirms that air oxidation had very little effect on the Tc-99 local bonding environment in these samples, even after 152 days. In the coprecipitation systems, again there was similarity across the reduced and oxidized samples and it is also clear that in all the samples the second shell of scattering atoms contributed more strongly to the EXAFS compared to the parallel sorption sample. This was confirmed by intense peaks in the phase-shifted Fourier transform of all three coprecipitation samples in the 2.5−4 Å range indicating a large number of backscattering atoms in these shells and suggesting incorporation of Tc(IV) into the magnetite structure.
Previously, Tc(IV) reduced onto environmental surfaces has been fitted using polymeric Tc chains 22,51,52 or hydrous TcO 2 like phases, 26 and both of these models can be applied to our magnetite sorption data with similar results. In the reduced and oxidized sorption samples, the addition of either Tc-99 or Fe shells beyond the Tc−O shell did not significantly improve the fit, and F-tests 68 confirmed the more complex fits were statistically indistinct from a single Tc−O shell fit (Figure SI-5 and Tables SI-2 to SI-4, Supporting Information). Therefore, the best fit to the sorption sample was an averaged TcO 2 structure, as expected for an amorphous hydrous-TcO 2 precipitate on the magnetite surface. Interestingly, the double peak in the XANES region in all samples suggests the presence of Tc−Fe bonds. 40 The presence of this feature in the magnetite sorption samples may indicate a component of a Tc−Tc chain sorbed to the magnetite surface 22 or indeed some incorporation of Tc(IV) into the magnetite, as postulated previously 41 and supported by the Tc-99 solution data ( Figure  1).
The Tc-99 EXAFS from the coprecipitation samples were fitted assuming a model of Tc(IV) incorporation into the magnetite structure where Tc(IV) was substituted solely into the octahedral site and similar to the model for Ti(IV) substitution into magnetite. 18 In our model, the Tc−O coordination was fixed at 6 while the coordination of the two Tc−Fe shells was allowed to vary. This approach achieved a good fit with interatomic distances similar to those for octahedrally coordinated Fe(III) in pure magnetite 69 (Table  2). The Tc−Fe coordination in both shells was lower than, although within error of, the expected coordination of 6, with 4.7 ± 1.3 and 4.4 ± 1.3 in the Tc−Fe shells. In nanoparticulate systems, surface effects may result in lower coordination numbers; atoms residing in the near-surface of a particle will have a lower coordination in the metal−metal shells than those in the particle core. This effect has previously been observed for U−U coordination in uraninite nanoparticles. 70 We have shown that, in our reduced samples, Tc-99 was relatively concentrated in the near-surface of iron oxide nanoparticles, which suggests surface effects may explain the observed low Tc−Fe coordination. The EXAFS data and associated fits from the reduced and oxidized magnetite in the YCL (pH 13.1) and OCL (pH 10.5) systems are within error of those in the ICL (pH 12.5) system (Figure SI-6 and Table SI  The ionic radii of Tc(IV) and Fe(III) are identical at 0.785 Å, 71 and so, the octahedrally coordinated site in the magnetite structure should accommodate Tc(IV) with little distortion. There is an excess charge associated with Tc(IV) substituting for octahedrally coordinated Fe(III), and this may cause the slightly lengthened Tc−Fe 1 atomic distance (∼3.05 Å) observed from the EXAFS fitting in our samples compared to the Fe−Fe distance in magnetite (2.97 Å). The excess charge associated with the Tc(IV) substitution must be balanced through charge compensation. 39 One suitable mechanism to do this would be for coupled substitution of a lower charged ion, but no credible candidate exists in our system. Alternatively, the solid may compensate the charge through creation of a vacancy site or through reduction of Fe(III) to Fe(II) as in the case of Ti(IV) substitution into magnetite, 72 and this is the most likely mechanism here, although the relatively low concentration of Tc-99 in the samples at 0.1 wt % means that direct evidence for this was not available in our systems. Although there is no indication of the presence of a surface Tc(IV) phase in the EXAFS or the dissolution data, destructive interference between potential Tc−Tc and Tc−Fe signals has been reported 40,73 and we cannot definitively rule out minor contributions from a surface Tc(IV) phase in the reduced coprecipitated sample.
Tc(IV) Retention during Oxidation. Comparison of the spectra from the two oxidized samples and their respective fits to the reduced coprecipitation sample revealed that oxidation of Fe(II) to Fe(III), as confirmed from XRD measurements, had no significant effect on the speciation of the Tc-99 remaining in the solid even though significant maghematization and oxidation to goethite had occurred by 152 days. Indeed, in the 152 day sample, Tc-99 remained as Tc(IV) and could be  a CN denotes coordination number; R + ΔR denotes atomic distance; σ 2 denotes Debye−Waller factor; ΔE 0 denotes the shift in energy from the calculated Fermi level; S0 2 denotes the amplitude factor; X v 2 denotes the reduced Chi square value; R denotes the "goodness of fit" factor; the subscript MS denotes multiple scattering paths. Numbers in parentheses are 1 standard deviation on the last decimal place. b The multiple scattering paths considered were linear paths, and their ΔR and σ 2 parameters were evaluated as multiples of the corresponding single scattering path parameter. c Parameter was fixed. d Parameters were tied in a given fit.  as being predominantly substituted for Fe(III) in octahedral sites within the magnetite/maghemite structure. The presence of significant (∼50%) goethite has little effect on the Tc EXAFS in these systems. Interestingly, the fits to the 21 and 152 day oxidized samples do not perfectly capture features in the EXAFS in the k = 10−12 Å −1 range (Figures 4 and SI-6, Supporting Information) which become more apparent with greater oxidation of the samples. Addition of a Tc−Fe distance that could be indicative of Tc(IV) incorporation into a secondary goethite phase (∼3.2 Å) returned physically unrealistic Debye−Waller Factors, and so, this model was discounted. Changes in the host mineral structure could account for these features, for example, creation of vacancies during maghematization, but such effects were challenging to reproduce within the EXAFS model.

Environmental Science & Technology
Our dissolution data suggest that a Tc-99 poor layer was created in the magnetite upon oxidation, and 1 M HCl extractions suggest ∼10 wt % of the particle volume is depleted in Tc. Formation of trace element depleted surface layers following oxidation has previously been observed for a range of trace-metal incorporated magnetites. 66 Comparison with this past work suggests that this depleted layer forms due to a mix of oxidative Tc-99 diffusion out of the solid into solution and outward diffusion of Fe during dynamic maghematization. 45,66 The Fe(II)/Fe(III) ratio in solids dropped from 0.5 to between 0.12−0.21 in the ICL and OCL oxidation experiments indicating that at least ∼40% of the magnetite had been oxidized to maghemite. This equates to a surface layer with a thickness of ∼15% of the particle radius. This indicates that a model where there are three "zones" in a cross-section through the oxidized particles, an outer maghemite layer with essentially no Tc-99 (∼10%), a maghemite layer containing Tc-99 (∼30%), and a magnetite core containing Tc-99 (∼60%), could explain the Tc distribution in the samples. This three layer model is an oversimplification of the true system since there will not be a discrete boundary between magnetite and maghemite. 46 However, it is clear that, despite significant oxidation of the magnetite, solid associated Tc(IV) remains in the reduced state and resides in regions with complete oxidation of structural Fe(II) to Fe(III) and in a site almost identical to the octahedral site in magnetite ( Figure 4, Table 2). This suggests that a significant fraction of Tc(IV) incorporated into iron oxides may be recalcitrant to oxidation even if the iron oxide itself is oxidized, with the suggestion that irreversible binding of Tc-99 is possible in these systems.
It has recently been demonstrated that secondary armoring of Tc(IV) incorporated into goethite crystals protects the Tc(IV) from oxidation and decreases the release of Tc-99 from within goethite. 37,38 In our systems, it is possible that the maghematization process resulted in armoring of the Tc(IV) incorporated magnetite through creation of an Fe(III)-rich rind via outward diffusion of Fe(III). 45 Such a rind could serve to limit or even prevent further outward diffusion of Tc-99 to solution, which may account for the observation that, on reoxidation of magnetite, Tc-99 release to solution was restricted and was essentially at steady state after the initial rapid release during the first couple of weeks of oxidation. This suggests that Tc(IV) incorporated into magnetite may be stable over an extended time, even in the event of a shift to oxidizing conditions, and implies that reductively scavenged Tc-99 may be essentially irreversibly bound within magnetite over a wide range of geochemical conditions.
Effect of Solution pH. The EXAFS spectra for all of the coprecipitation samples were broadly similar for each of the three leachates (Figure SI-6 and Table SI-5, Supporting Information). This suggests that the behavior of Tc-99 during magnetite crystallization was pH independent, perhaps unsurprising given the predicted speciation of Tc in solution as TcO 4 − over the entire pH range (10.5−13.1). 9 The YCL (pH 13.1) and OCL (pH 10.5) fits were the same as the ICL (pH 12.5) fits; the reduced coprecipitation sample data were again best fit with the model of Tc(IV) incorporation into the magnetite octahedral site, as were the data from the oxidized samples. From the analysis of the XRD patterns, up to 20% of goethite crystallized in the YCL (pH 13.1) and ICL (pH 12.5) reduced systems. This however had no measurable impact on the Tc-99 speciation evidenced by the similarity of the EXAFS spectra and best fits to the pure magnetite OCL (pH 10.5) system. The coordination of Fe(III) in goethite is similar to that of Fe(III) in an octahedral site in magnetite: Fe(III) is octahedrally coordinated by O at ∼2 Å, and there are shells of Fe at ∼3 and ∼3.5 Å with coordination of 2 and 4, respectively; however, there is an additional pair of Fe atoms at ∼3.3 Å. 74 The total Fe−Fe coordination in goethite of 8 matches the total Tc−Fe coordination in our incorporated iron oxides. This necessitated examination of the possibility that Tc-99 resided in the goethite. Despite recent evidence of Tc(IV) incorporation into goethite, 37,38 our data could not be fitted using the published model from this work 37,38 and clearly show a coordination environment where Tc-99 was best modeled by incorporation into magnetite/maghemite. However, we cannot rule out the possibility that a small fraction of Tc-99 is residing in goethite present in the YCL (pH 13.1), ICL (pH 12.5), and 152 day oxidized OCL (pH 10.5) systems.
Implications for Tc-99 Behavior in the Environment. The data presented show that Tc(VII) is reduced to Tc(IV) and incorporated into the magnetite structure during crystallization from a ferrihydrite precursor, with 60−70% of the incorporated Tc(IV) retained as Tc(IV) during relatively robust, long-term reoxidation. Counterintuitively, the oxidation process may in fact aid retention of a sizable proportion of Tc(IV) within the magnetite/maghemite via an armoring process with Fe(III), leaving a significant fraction of the Tc(IV) recalcitrant to oxidation. Incorporation of Tc-99 into magnetite may offer a long-term sequestration mechanism for significant quantities of Tc-99 in alkaline environments (pH >10) where magnetite is present and may help to explain previously observed recalcitrance of Tc-99 during reoxidation of reduced, iron oxide-bearing sediments. It is noteworthy that additional geochemical variables not considered in this study, e.g., groundwater chemistry, may have significant impacts on the ability of iron oxides to sequester Tc-99 and merit further investigation. For example, the presence of silica has been shown to inhibit ferrihydrite crystallization to goethite and hematite, 75 although it does not seem to affect magnetite crystallization at circum-neutral pH. 76 Given that Tc(VII) was reduced and incorporated into magnetite over the entire experimental pH range and that magnetite has been shown to be a significant component in the corrosion of iron and steel under anaerobic conditions, 77 this study highlights the potential for significant retention of Tc-99 in radioactive waste disposal as a consequence of waste from iron corrosion. The mechanism of Tc(VII) reductive immobilization and incorporation to magnetite could therefore   Tables: 5 (Table SI-1 to Table SI The solution was sealed and left on the stir plate for 24 hours, after which it was stored in an anaerobic 28 chamber (5 % H 2 , balance N 2, maintained at < 1 ppm O 2 and CO 2 ).  Figure SI-4), a simple spherical particle may be assumed as a 63 generalization. Therefore, 10 wt. % of the Fe is contained in a surface layer with a thickness of 3.5 % of 64 the radius of the particle. In a 30 nm particle, this equates to a surface layer 0.5 nm thick which is 65 approximately one unit cell. The fits were performed in normalized μ(E) space between -30 and +40 eV. Neither standard was 98 stipulated as required in the fit but the weights of the standards were forced to between 0 and 1, and also 99 to sum to 1. The results of the linear combination fitting are given in Table SI

EXAFS fitting -ICL reduced sorption sample fits 107
The results of fits to the ICL sorption reduced data are presented below. The rationale behind fitting 108 models (a)-(g) are given in Table SI- (1)