On the Change in UO 2 Redox Reactivity as a Function of H 2 O 2 Exposure

To assess the long-term leaching behaviour of UO 2 , the main constituent of spent nuclear fuel, the oxidative dissolution of UO 2 pellets was studied at high H 2 O 2 exposures ranging from 0.33 mol m -2 to 1.36 mol m -2 . The experiments were performed in aqueous media containing 10 mM HCO 3- where the pellets were exposed to H 2 O 2 three consecutive times. The results indicate that the dissolution yield (amount of dissolved uranium per consumed H 2 O 2 ) at high H 2 O 2 exposures is significantly lower compared to previous studies of both pellets and powders and decreases for each H 2 O 2 addition for a given pellet. This implies a change in redox reactivity, which is attributed to irreversible alteration of the pellet surface. Surface characterization after the exposure to H 2 O 2 , by SEM, XRD and Raman spectroscopy show, that the surface of all pellets is significantly oxidized.


Introduction
After its use in a reactor, nuclear fuel is planned to be stored in deep geological repositories in many countries 1 . A deep repository must be designed to prevent groundwater intrusion in order to remain intact for long time spans 2 . Predictive modelling as well as studies of natural analogues are crucial for its safety assessment [3][4][5] . Under the reducing groundwater conditions prevailing at deep repository sites, the UO 2 matrix, which makes up most of the spent nuclear fuel, has very low solubility 6 . At the time when water intrusion can be expected to occur, the predominant radiation from the spent nuclear fuel will be alpha-radiation 7,8 . The radiation will induce radiolysis of water in close proximity to the fuel. This can generate oxidizing conditions at the fuel surface under which UO 2 is considerably more soluble 6 thereby facilitating the migration of U and other radionuclides in the environment 9 . Among the radiolysis products, H 2 O 2 has been demonstrated to be the main oxidant responsible for uranium dissolution in systems exposed to alpharadiation 10 .
(1) 2  It is important to note that the hydroxyl radicals formed in reaction (1) are stabilized by adsorption to the UO 2 surface, adsorption being a prerequisite for this reaction to occur spontaneously. Under certain conditions, reaction 5 can be rate limiting but in the presence of the common groundwater constituent HCO 3 -, dissolution of U(VI) is facilitated by the formation of soluble complexes [14][15][16] .
To quantify the competition between oxidation of UO 2 by H 2 O 2 and catalytic decomposition of H 2 O 2 on the UO 2 surface, the term dissolution yield has been introduced 17 . It is defined as the ratio between the amount of U(VI) dissolved from a pellet or a given amount of powder and the amount of H 2 O 2 consumed on the same uranium is dissolved is identical to the surface area on which H 2 O 2 is consumed, the dissolution yield can be regarded as independent of the UO 2 surface area that was exposed to the solution. This entity offers a straightforward way to compare experiments on specimens with different surface areas or where surface areas are difficult to determine experimentally. It was recently shown that the dissolution yield for UO 2 powder ranges from 44 % to 100 % depending on the initial H 2 O 2 concentration 18 . For UO 2 pellets the dissolution yield was reported to be around 15 % [19][20][21] . The range of dissolution yields observed for UO 2 powder was rationalized from the mechanism for catalytic decomposition of H 2 O 2 . Hydrogen peroxide acts as radical scavenger for surface bound hydroxyl radicals (reaction 2) and the higher the concentration of H 2 O 2 , the larger is the fraction of H 2 O 2 consumed through catalytic decomposition. In fact, the keycompeting reactions involving the surface-bound hydroxyl radical ( ) can be described as follows (6-7) 18 : From this mechanism a kinetic definition of the dissolution yield accounting for the H 2 O 2 -concentration dependence can be derived (8): It should be noted that the U(V) species formed in reaction (6) must be oxidized one step further before dissolution takes place.
It was recently observed, that the reaction between H 2 O 2 and UO 2 powder does not follow strict first order kinetics 18,22 . Instead, it turned out that the rate for H 2 O 2 consumption at a given H 2 O 2 concentration depends on the initial H 2 O 2 concentration. In other words, the overall kinetics changes with turnover of H 2 O 2 . This change in surface reactivity can only be attributed to a surface alteration process, possibly due to passivation of reactive sites by oxidized uranium or surface bound hydroxyl radicals. In the experiments mentioned above, the surface area to solution volume ratios were between 2700 m -1 to 10800 m -1 for powder experiments 18 while they were around 22 m -1 for pellets 19,20 . This means that the UO 2 surfaces are exposed to very different amounts of H 2 O 2 in powder experiments as compared to pellet experiments. In the powder experiments the H 2 O 2 exposure ranges from 1.8 x 10 -5 mol m -2 to 7.4 x 10 -4 mol m -2 while in the pellet experiments the H 2 O 2 exposure is around 0.08 mol m -2 . In view of the time spans relevant for a deep repository, the very low exposures of the powder experiments correspond to an extremely short leaching time and the results may therefore not be relevant for a long-term safety assessment. To explore the possible change in surface reactivity of spent nuclear fuel, considerably higher exposures are needed.
In this work, we have studied the oxidative dissolution of UO 2 pellets at H 2 O 2 exposures ranging from 0.33 mol m -2 to 1.36 mol m -2 . Based on the experimental results of this and previous studies 18,20,22 , the effect of H 2 O 2 exposure on the UO 2 surface reactivity is analysed and discussed.

Material and Methods
Pellet Fabrication and Characterisation.
Caution! Although natural uranium was used in this study and the radioactivity of the material is low due to its long half-life, safety precautions regarding the work with radioactive materials should be followed. The work with radioactive materials should only be conducted by trained staff and take place in appropriate facilities. Precursor powders for the pellets were produced by coprecipitation, where ammonium di-uranate (ADU) is precipitated from an aqueous solution containing 16.5 M ammonia and 2 M uranyl-nitrate. The [UO 2 (NO 3 ) 2 6H 2 O] ≥ 99 % was supplied by Merck. After precipitation the powder was washed four times with high purity water to remove NH 4 OH residuals. During the last washing step water was replaced with ethanol and the powder was left to dry. It was then calcined in air at 600 °C for 5 h for denitration, de-hydration, and transformation to U 3 O 8 . In a second step, the U 3 O 8 powder was reduced to stoichiometric UO 2 during thermal treatment for 5 h at 600 °C in a tube furnace flushed with a 4 % H 2 -96 % Ar mixture (HYTEC). Afterwards, the calcined powder was compacted (for pressures see supplementary information S1) to disk shaped green bodies (10 mm diameter) and pellets were sintered at 1700 °C for 10 h in a 4 % H 2 in Ar atmosphere to stoichiometric UO 2 according to 23 . After sintering, the pellets were polished in several steps, the last step with a 0.04 μm colloidal SiO 2 paste, and then thermally treated at 1350 °C for 180 min in reducing atmosphere (HYTEC). Total densities (Table 1) of the pellets were determined by a modified Archimedes method as described in 24 , where paraffin is used to cover all surfaces including open porosity and to avoid an uptake of water into the ceramic during the measurement. Additionally, the porosity of the pellets was determined from SEM observations via image analyses (ImageJ). A comparison of the two measurement methods shows coherent results, i.e. the porosity observed at the surface is representative for the volume of the pellets (SEM images see supplementary information S2). Grain sizes were determined by measuring and averaging the length of the long and short axes of at least 100 grains per pellet. The average grain size of all samples was 12 ± 1 μm and the average weight of the pellets is 0.97 ± 0.03 g. The surface areas of the pellets were determined geometrically and amounts (1.53 ± 0.05) x 10 -4 m 2 .
Please do not adjust margins Please do not adjust margins Changes to the microstructure during the exposure to H 2 O 2 where analysed using a Quanta 200 FEG SEM (Thermo Fisher / FEI). Images were recorded before and after dissolution using both secondary electron (SE) and backscatter electron (BSE) detectors at an acceleration voltage of 20 kV in low vacuum mode.
XRD patterns of the pellets were recorded on a Bruker D4 Endeavor diffractometer, 40 kV and 40 mA, in Bragg−Brentano geometry. The diffractometer is equipped with a copper X-ray tube and a primary nickel filter producing graphite monochromized CuKα1 radiation (λ = 1.54187 Å). A linear silicon strip LynxEye detector (Bruker -AXS) was used. The patterns were recorded in the range of 2θ = 5 − 130° with a 0.02° increment and a scan speed of 2 s / step. The aperture of the fixed divergence slit was set to 0.2 mm and the receiving slit to 8.0 mm respectively. To determine the reaction products that form on the surface of the pellets during the exposure to H 2 O 2 , non-polarized Raman spectra were recorded after three consecutive exposures to H 2 O 2 on a Horiba LabRAM HR spectrometer using a Peltier cooled multichannel CCD detector. An objective lens with a 50 × magnification was linked to the spectrometer, allowing the analysis of samples as small as 2 μm in diameter. The incident radiation was produced by a He-Ne laser at a power of 17 mW (λ = 632.8 nm). The focal length of the spectrometer was 800 mm and a 1800 gr mm -1 grating was used. The spectral resolution was approximately 1 cm -1 with a slit of 100 μm. All spectra were recorded in the range between 200 − 2000 cm -1 . For each pellet three spectra were recorded at three different locations.
Dissolution Experiments. Prior to the dissolution experiments each pellet was washed in de-aerated 10 mM bicarbonate solution to remove pre-oxidized uranium from the surface, as pre-oxidized uranium might appear during production and handling of the pellet. The washing was carried out with the following procedure: First, each pellet was immersed in de-aerated bicarbonate solution for 10 min. In a second step, the bicarbonate solution was replaced several times after 10 min, 12 h and again after 10 min. After washing, the bicarbonate solution was replaced once more and H 2 O 2 was added immediately to a concentration of 2.25 mM and a total sample volume of 40 25 whereas U(VI) was measured at 653 nm using the Arsenazo III method 26 . The samples were purged with Ar (≥ 99.999 %, Air liquide) throughout the experiments.
Chemicals used in all experiments were of reagent grade or higher unless otherwise stated. Purified water (18.2 MΩ.cm, Merck MilliQ) was used throughout. Between individual H 2 O 2 exposures each pellet was rinsed with water and sonicated to remove reaction products (i.e. U(VI)) from the previous experiment. It was then repeatedly washed with 10 mM de-aerated bicarbonate solution using the same procedure as stated above.
After the last experiment the pellet was washed with de-aerated MilliQ water and then left to dry in an Ar atmosphere.

Results and Discussion
Dissolution Experiments. Figure 1 shows the results of dissolution experiments for pellet P5 during three consecutive H 2 O 2 exposures. For pellets P1 to P4 the trends are similar (see supplementary information S3). From the figure it becomes clear that the amount of uranium released decreases in every consecutive exposure to H 2 O 2 . In contrast, the consumption of H 2 O 2 is almost identical for each exposure. It should be noted that the data in Figure 1 is not corrected for the background consumption of H 2 O 2 on the glass surfaces of the reaction vessels. Data corrected for the background consumption can be found in Table 2 along with the corresponding dissolution yields. It is clear that the dissolution yields are quite low compared to previous studies and that for a given pellet, it decreases for every new exposure to H 2 O 2 . As mentioned above, previously reported dissolution yields for UO 2 pellets are around 15 % 19-21 while for powders the dissolution yield is even higher 18 . The results of the consecutive H 2 O 2 exposures presented here provide some new insights since the systematic decrease in dissolution yield strongly indicates that the pellet surface is being altered, even in the presence of 10 mM bicarbonate and thorough washing with HCO 3solution between the exposures. This behaviour has, to the best of our knowledge, not been reported previously.
It was previously suggested that the dissolution yield depends on the initial H 2 O 2 concentration and that this could be explained by the mechanism for catalytic H 2 O 2 decomposition 18 . The results presented here do not support the previous conclusion. Instead, the dissolution yield appears to depend on the total exposure to H 2 O 2 expressed in mol m -2 of UO 2 surface.
As already pointed out, there is a significant difference in the solid surface area to solution volume ratio between pellet experiments and powder experiments. In the current experiments, the surface area to volume ratio is around 3.5 -4.0 m -1 while in powder experiments the corresponding ratio is often around 5000 m -1 or higher. This means that a pellet is exposed to about three orders of magnitude more H 2 O 2 per surface area of material compared to the powder at the same initial H 2 O 2 concentration. To analyse the possible correlation between the dissolution yield and H 2 O 2 -exposure per surface area, we have plotted the dissolution yields reported above as function of total H 2 O 2 -exposure per surface area (Fig. 2). For the second and third exposures, the accumulated H 2 O 2 exposure is used. The results indicate a clear trend where the dissolution yield decreases with increasing H 2 O 2 -exposure per surface area. Again, this implies that the surface is continuously being altered in what seems to be an irreversible way. Although the overall reactivity of H 2 O 2 towards the pellets is the same in all three exposures, the change in dissolution yield is a direct consequence of a change in the redox reactivity by as much as a factor of 3 to 4. This is in line with the change in kinetics observed in the powder experiments discussed above 18,22 . Since the dissolution yields are considerably higher for powders, a change in the redox reactivity also has a significant impact on the overall reactivity.
Compared to the results from pellet experiments (Fig. 2), powder experiments represent the other extreme in terms of solid surface area to solution volume ratio. In recently published powder experiments 18 the same general trend was observed, despite the quite obvious H 2 O 2 concentration dependence for high surface area experiments. In order to connect the low and high solid surface area to solution volume experiments, we plot the results from powder experiments 18 and results from the pellet experiments in Please do not adjust margins Please do not adjust margins this study in the same graph (Fig. 3a). In addition, the result of a previously published pellet experiment is included 20 .
As can be seen, the data from the previously published pellet experiment performed at a slightly higher surface to volume ratio than in the present study connects the data from this work with the data for powder experiments. Given the wide range in H 2 O 2exposure per surface area, we have also made the same plot with logarithmic axes (Fig. 3b).
Again, we observe a more or less continuous trend that clearly shows how the pellet becomes increasingly resistant to oxidative dissolution induced by H 2 O 2 . It is interesting to compare the present results to the expected exposure conditions in a deep repository. The rate of H 2 O 2 consumption on a 1000 year old fuel with a burnup of 55 MWd / kgU has been calculated to be 1.21 x 10 -10 mol m -2 s -1 according to 27 . Consequently, a total H 2 O 2 exposure of 1.36 mol m -2 , as in the present work, corresponds to 356 years of exposure for the 1000 year old fuel. This is a fairly short time span compared to the time span that has to be considered for a deep geological repository. For the same fuel at 100 000 years age the time span to reach the same H 2 O 2 exposure is slightly above 14 000 years.
For lab experiments aiming at elucidating the kinetics of spent nuclear fuel dissolution, the results presented above demonstrate the importance of the history of the specimens used in leaching experiments.
It should be noted that under the conditions used in the experiments presented here (10 mM HCO 3 -), studtite formation is thermodynamically favourable 28 . However, when considering the equilibrium constants for uranyl carbonate-and uranyl peroxo carbonate complexes 29,30 , the solubility under the present conditions is still significant and therefore Studtite precipitation is not to be expected. Studtite formation is usually confirmed in systems where UO 2 has been exposed to H 2 O 2 in the absence of HCO 3 -31, 32 . In a previous study 33 , UO 2 powder, a UO 2 pellet and a SIMFUEL pellet (UO 2 doped with 11 nonradioactive isotopes of fission products to mimic real spent nuclear fuel) were exposed to high concentrations of H 2 O 2 in aqueous solutions not containing HCO 3 -. High H 2 O 2 concentrations were used to favour studtite formation. Post exposure investigation of the pellet surfaces with Raman spectroscopy confirmed the formation of studtite/metastudtite on UO 2 powder and on the UO 2 pellet. Interestingly, studtite/metastudtite formation was not observed on SIMFUEL. For SIMFUEL, the dissolution yield is very low and the reactivity of H 2 O 2 is dominated by catalytic decomposition.
It has been reported that the presence of HCO 3efficiently removes oxidized uranium from the surface in oxidative dissolution experiments 34,35 . However, post exposure surface characterization showing an unaltered surface, have only been reported for HCO 3containing systems where the oxidant is molecular oxygen. H 2 O 2 containing systems still remain to be explored.
To shed some light on the nature of the solid phase alteration caused by the exposure of UO 2 to H 2 O 2 in this study, surface characterization of the pellets was performed using SEM, XRD, and Raman spectroscopy.
Scanning Electron Microscopy. BSE-SEM images of pellet P3 before (A) and after (B) three consecutive exposures to H 2 O 2 are shown in Figure 4. A significant change in the surface morphology is visible after the exposure to H 2 O 2 , mainly at the grain boundaries, but also at the grain surfaces. Already before the experiments, the grain boundaries are slightly etched due to the thermal treatment. When looking at the grains themselves, small square shaped structures appear after thermal etching. We ascribe them to a surface relaxation feature that appears at elevated temperatures. After the exposure to H 2 O 2 these features become more pronounced, indicating dissolution, even from the grains themselves. After the dissolution experiment, some grains are liberated due to etching of the complete grain boundary.  A closer view of P3 after dissolution is provided in Figure 5 to reveal the details of individual grains. A significant surface roughness of the grain surfaces appears after the dissolution experiments (SE-image, Fig. 5A), which coincides with a new contrast appearing in the BSE image (Fig. 5B). The increased contrast and surface roughness may be due to differences in reactivity of the surface, leading to the preferential dissolution of parts of the surface. These irregularities may be a result of orientational or stoichiometrical effects. The latter will be discussed in the Raman spectroscopy section. It becomes obvious that dissolution mainly occurs on certain grain boundaries while other grain boundaries remain intact. Also grain boundaries from just under the pellet surface dissolve, opening up small crack-like structures throughout different grains. Additionally, individual grains are loosened up from the matrix during the exposure to H 2 O 2 leaving holes behind.

X-ray Diffractometry.
XRD patterns recorded prior to the dissolution experiments (Fig. 6a) confirm the stoichiometry of UO 2 in a single cubic phase 36 . The lattice parameters for samples P1 and P5 were calculated to be 5.477 ± 0.002 Å and 5.479 ± 0.003 Å based on the Debye-Scherrer method, which is in good agreement with literature 23 . XRD results from after the exposure to H 2 O 2 (Fig. 6b) reveal, that hyperstoichiometric UO 2 was formed which remains on the surface during the experiments; i.e. a smaller secondary peak appears at higher 2Θ angles next to the primary UO 2 peak. Broadening of this secondary peak can be due to both, a decrease in crystallinity based on the formation of interstitial oxygen defects as well as a range of different oxidation sates. Assuming that different states of oxidized uranium form during the reaction with H 2 O 2 , the range of hyperstoichiometry can be narrowed down to x ≤ 0.25 in UO 2+x . The highest oxidation state therefore matches the U 4 O 9 phase as indexed in purple 37 . It should be noted that no signs of studtite formation can be observed.
Raman Spectroscopy. The Raman spectrum of a sintered pellet which is a twin of the pellets which were used in the H 2 O 2 experiments, revealed the stoichiometric UO 2 bands at 445 cm -1 and 1150 cm -1 38 . Also a small band at 918 cm -1 was observed 39 . Spectra of samples P1 and P5 after the third exposure to H 2 O 2 are shown in Figure 7.  New bands appear for all samples after they were exposed to H 2 O 2, . These bands can be ascribed to various oxidation states of hyperstoichiometric UO 2 as shown in Table 3. Their intensities vary for each pellet and location where they were recorded. Even oxidation states up to U 3 O 8 were found. As expected, no signs of studtite formation were observed using this technique either. It should be pointed out that studtite formation was previously observed using Raman on UO 2 specimens exposed to radiation in solutions without HCO 3 -41, 42 As compared to the hyper-stoichiometry found by XRD, the Raman results reveal higher oxidation states. Since the XRD is insensitive to distortions in the anion sub lattice, transitional phenomena regarding distortions to the oxygen sub lattice remain unnoticed with XRD when cubic UO 2 is gradually oxidized to tetragonal U 3 O 7 . Based on the attenuation length of photons penetrating UO 2 , the 8 keV x-rays penetrate the surface of the pellet much deeper as compared to the 1.95 eV Raman laser. Therefore one can assume, that the oxidation gradually decreases from the surface towards the center of the pellet as higher oxidation states than U 4 O 9 were only measured by Raman spectroscopy.
Low frequency vibrations (below 343 cm -1 ), which only occur very locally, could not be identified in detail. However, we assume that they are an indication of distortion to the cationic sub lattice.

Conclusions
The experimental results presented in this work show that the redox reactivity of UO 2 decreases quite dramatically with exposure to H 2 O 2 in HCO 3containing aqueous solutions. Surface alteration to which this change in redox reactivity is attributed appears to be permanent and cannot be reverted merely by exposure to bicarbonate. The nature of this alteration is an increase in oxidation state; however, the degree of oxidation is not straightforward to assess. Surface characterization techniques indicate that an oxidation state gradient evolves upon high exposure to H 2 O 2 .
These finding demonstrate the importance of keeping track of and reporting the H 2 O 2 exposure history when performing lab experiments on oxidative dissolution of UO 2 specimens in general and in particular when extrapolating these results to repository conditions.
In case of groundwater intrusion into a deep repository for spent nuclear fuel, the observed change in redox reactivity would lead to a fairly drastic inhibition of radiation induced fuel matrix dissolution and thereby also a limit the radionuclide release. In fact, by numerically integrating the curve presented in Fig. 3 we can estimate the fraction of dissolved fuel matrix before reaching complete inhibition of oxidative matrix dissolution to less than 0.1 % of the total inventory. This would have a tremendous impact on the safety assessment of a repository for spent nuclear fuel.

Conflicts of interest
There are no conflicts to declare.