Distribution of studtite and metastudtite generated on the surface of U3O8: application of Raman imaging technique to uranium compound

ABSTRACT Studtite and metastudtite are uranyl peroxides formed on nuclear fuel in water through the reaction with H2O2 produced by the radiolysis of water. However, it is unclear how the two types of uranyl peroxides are generated and distributed on the surface of nuclear fuel. Here, we used Raman imaging technique to exemplify distribution data of the two uranyl peroxides formed on the surface of a U3O8 pellet through immersion in a H2O2 aqueous solution. As a result, we observed that studtite and metastudtite are heterogeneously distributed on the U3O8 surface. No clear correlation between the distributions of studtite and metastudtite was observed, suggesting that the two uranyl peroxides are independently generated on the surface of U3O8. We anticipate that this Raman imaging technique could reveal how these uranyl peroxides are generated and distributed on the surface of the nuclear fuel debris in the Fukushima-Daiichi nuclear power plants.


Introduction
, which is the dehydrated form of studtite, are solid phases of uranyl peroxides and have been found in spent nuclear fuel exposed to water [1] and UO 2 exposed to water under a radiation field [2]. The uranyl peroxides are formed through the reaction of UO 2 with H 2 O 2 generated by the radiolysis of water [3,4], although the mechanism that determine which uranyl peroxide phase is generated remains unclear. Because the nuclear fuel debris in the Fukushima-Daiichi nuclear power plants are in contact with coolant water under a high radiation field, the uranyl peroxides are likely ones of the dominant alteration phases of the fuel debris that could determine the properties of the fuel debris.
Because of the significance of the uranyl peroxides as alteration phases on nuclear fuel [5], there have been many studies to characterize uranyl peroxides properties such as: thermal stability [6][7][8], stability under irradiation [9], electrochemical reactivity [10], solubility [11], and crystal structures [12,13]. However, previous studies have not paid much attention to how the two uranyl peroxides, studtite and metastudtite, are generated and distributed on the surface of nuclear fuels. Clarens et al. reported the formation of studtite from the reaction of UO 2 surface with H 2 O 2 [4], whereas Sattonnay et al. found metastudtite as the dominant alteration phase on UO 2 irradiated by He ion beam in water [2]. On the other hand, Hanson et al. observed the formation of studtite on spent nuclear fuel and at the same time metastudtite as suspended particles in the samples of their immersion experiments [1]. Therefore, to reveal the generation and distribution mechanism of the two uranyl peroxides on nuclear fuels, it is necessary to distinguish the two uranyl peroxides and observe their distributions on nuclear fuels.
Raman spectroscopy is a useful method used to identify the two uranyl peroxides [7,14]. However, Raman imaging technique, which can show the distributions of certain chemical species in a microscopic area [15,16], has not yet been applied to the study of the uranyl peroxides. In the present study, we used the Raman imaging technique and successfully demonstrated the distribution data of the two types of uranyl peroxides formed in microscopic areas on the surface of a U 3 O 8 pellet. U 3 O 8 was selected as the starting material because it has been unclear whether the uranyl peroxides were generated from U 3 O 8 , although U 3 O 8 is a key component generated from UO 2 in oxidized conditions [17]. From the distribution data, we discussed some aspects of the generation and distribution of the two uranyl peroxides on U 3 O 8 . We anticipate that this Raman imaging technique will be applicable even to highly inhomogeneous nuclear materials, such as the nuclear fuel debris in the Fukushima-Daiichi nuclear power plants. CONTACT

Preparation of U 3 O 8 sample
U 3 O 8 sample stored at Tohoku university was heated at 800°C under air for 4 hours and confirmed as α-U 3 O 8 structure by XRD ( Figure S1).

Synthesis of studtite and metastudtite
In order to analyze Raman spectra measured on the surface of the U 3 O 8 pellet with Raman spectra of studtite and metastudtite, we synthesized studtite and metastudtite and obtained their Raman spectra. Studtite was prepared from UO 2 powder by reaction with H 2 O 2 [6]. A suspension of the UO 2 powder (c.a. 430 mg) in 1 × 10 −3 mol dm −3 aqueous HCl solution (10 cm 3 ) was prepared in a plastic tube and then 30 wt% H 2 O 2 solution (0.73 cm 3 ) was added to the suspension. The amount of the added H 2 O 2 corresponds to the 2 times excess of the stoichiometric amount for the studtite formation. The suspension was then slowly shaken with a tube rotator at 20 rpm. After 1 to 3 days of reaction, the supernatant containing fine yellow powder was transferred to a glass flask. To precipitate, again 10 cm 3 of the HCl solution and 0.73 cm 3 of the H 2 O 2 solution were added. This procedure was repeated 5 times to complete the reaction. Then, the fine yellow powder was obtained from the suspension and washed with 10 cm 3 of pure water 3 times by centrifuging and decantation. The washed powder was dried in a vacuum desiccator. All the synthesis procedures were performed at room temperature. Metastudtite was obtained by heating the studtite at 90°C for 10 minutes. The synthesis of studtite and metastudtite was confirmed by XRD ( Figure S2 and Figure S3) and TG-DTA ( Figure  S4) analysis.

Raman imaging
A confocal Raman microscopy setup (NRS-4500, JASCO) was used to obtain Raman imaging data on the surface of the U 3 O 8 pellet. A 532 nm laser of 0.5 mW was introduced on a sample stage through a 100x objective lens. The spatial resolution is XY < 1 μm and Z < 1.5 μm. By moving motor stages with 1 μm step in a 10 μm × 10 μm square area for a two-dimensional scan, 121 Raman spectra were collected for the square area. The XY scan was conducted three times in different areas optionally selected, and 363 (= 121 spectra × 3 areas) Raman spectra were obtained. The distribution data of studtite and metastudtite generated on the surface of the U 3 O 8 pellet were obtained using their strong bands in the 780-900 cm −1 region. For the synthesized studtite and metastudtite, their Raman spectra were measured from a single point on the surface of their particles using a 20x objective lens.  (Figure 1(b)), the bands due to U 3 O 8 are still observed, but two intense bands appear at ~820 cm −1 and ~865 cm −1 , showing the generation of chemical species on the surface of U 3 O 8 by the immersion. Figure 1(c, d) shows Raman spectra of the synthesized studtite and metastudtite, and both uranyl peroxides show the two intense bands at ~820 cm −1 and ~865 cm −1 , with small bands in the region lower than 400 cm −1 . The positions and the relative intensity of the ~820 cm −1 and ~865 cm −1 bands of studtite and metastudtite are very similar to those observed in the spectrum of Figure 2(b). Therefore, the two intense bands observed at ~820 cm −1 and ~865 cm −1 in the immerged U 3 O 8 pellet (Figure 2(b)) are attributed to the uranyl peroxides. Figure 2 shows an enlarged view of the four Raman spectra of Figure 1 in the region of 780-900 cm −1 . From the enlarged figure, it is evident that the bands at ~820 cm −1 and ~865 cm −1 observed for the immersed U 3 O 8 (Figure 2(b)) are broader than those observed for studtite and metastudtite (Figure 2(c, d)). Furthermore, the band positions of studtite and metastudtite (Figure 2 (c, d)) are slightly different between each other, and as indicated by the dashed lines, the broader bands in Figure  2(b) consist of the bands of studtite and metastudtite. Therefore, both uranyl peroxides, i.e. studtite and metastudtite, are generated on the surface of U 3 O 8 through the immersion in the H 2 O 2 aqueous solution.

Results and discussion
Although Figure 2(b) clearly indicates that both studtite and metastudtite are formed on the surface of the U 3 O 8 pellet after the immersion, any information about the distributions of the studtite and metastudtite generated on the U 3 O 8 surface cannot be obtained, because Figure 2(b) is an averaged spectrum of 363 Raman spectra observed on the U 3 O 8 surface. To illustrate the distributions of studtite and metastudtite generated on the U 3 O 8 surface, we approximate each Raman spectrum of the immersed U 3 O 8 as the linear combination of the two Raman spectra of studtite and metastudtite as follows: where a and b are coefficients corresponding to the Raman intensity of studtite and metastudtite generated on the surface of U 3 O 8 , respectively. The two coefficients were obtained by fitting with the spectra of the synthesized studtite and metastudtite in Figure 2(c, d) in the least-square method in the 780-900 cm −1 region. Figure  3 shows two-dimensional plots of the two coefficients, a and b, for three 10 μm × 10 μm square regions (Region 1, Region2, and Region 3) on the surface of the immersed U 3 O 8 pellet. The magnitude of the coefficients was expressed by the color scheme shown in the right side of the figure. The coefficients, a and b, were normalized by using the largest coefficient among all coefficients (N max ), and the normalized values, in this case, are a/N max and b/N max . Since the Raman intensity is proportional to the quantity of studtite and metastudtite on a measurement point, Figure 3 represents distributions of the quantity of studtite and metastudtite generated on the U 3 O 8 surface. The optical images of the three regions are shown in the bottom of Figure 3 and show smooth surfaces of the pellet in the microregions. The measurements for the three regions were sequentially performed under the same experimental condition. The relative relationship between the a and b values indicates the relative quantity of studtite and metastudtite generated on the U 3 O 8 surface. Since the two spectra of Figure 2(c, d) retain the relative intensity of the synthesized studtite and metastudtite for the fitting, when a = b, the relative quantity of studtite and metastudtite generated on U 3 O 8 surface (q studtite,U3O8 /q metastudtite,U3O8 ) is the same as the relative quantity of the studtite and metastudtite per the same area size of the standard samples (q studtite,synthesized /q metastudtite,synthesized ). Moreover, when a > b, in comparison with the ratio of studtite and metastudtite of the standard samples, the proportion of studtite is higher than that of metastudtite on U 3 O 8 , and conversely when a < b, in comparison with the ratio of studtite and metastudtite of the standard samples, the proportion of metastudtite is higher than that of studtite on U 3 O 8 . Figure 4 shows examples of the spectral fittings at the three positions in Region 1: (a) (X, Y) = (1, 10), (b) (X, Y) = (9, 5), (c) (X, Y) = (4, 7). In Figure 4, the black solid curves are Raman spectra measured at each position, the red solid curves are fitting spectra, and the blue and green dashed curves are the Raman spectral components of studtite and metastudtite. Each Raman spectrum is well reproduced, indicating that the distributions of studtite  and metastudtite generated on the U 3 O 8 surface are accurately obtained by this analysis.
The distributions of studtite and metastudtite in Figure 3 show that the U 3 O 8 surface is covered with the two uranyl peroxides, i.e. studtite and metastudtite, and they are heterogeneously distributed over the U 3 O 8 surface. The heterogeneous distributions include a minimum unit area smaller than 1 μm × 1 μm because there are some pixels that have very different intensities from neighboring pixels. The heterogeneous distribution patterns of the two uranyl peroxides cannot be observed from the optical images at the bottom of Figure 3, demonstrating high performance of Raman imaging technique for selective detection of studtite and metastudtite in such a microscopic area. Furthermore, it is apparent that the overall intensity of studtite is larger than that of metastudtite, indicating that studtite is preferentially distributed compared to metastudtite. The ratio of the averaged coefficients, <a>/<b> = 0.34/0.23 = ~1.5, indicates that the yield of studtite is ~1.5 times larger than that of metastudtite in the present experimental condition. The thickness of the layer of the uranyl peroxides formed on the U 3 O 8 surface is estimated to be thinner than ~3 μm from depth profiles of the uranyl peroxides ( Figure S5), and because of the thin layer of the uranyl peroxides, the U 3 O 8 signals below the layer of the uranyl peroxides are clearly observed in Figure 1(b).
Finally, in order to investigate correlation between the distributions of studtite and metastudtite on the U 3 O 8 surface, the two coefficients (a and b) obtained in each measurement position are plotted in Figure 5. No clear correlation between the two coefficients was observed. Additionally, the correlation coefficient, r ab , was calculated with the following equation, Þ ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi where n is the data number (= 363), a and b are the coefficients defined in Equation (1), and brackets, < >, stand for average. A near-zero value (−0.1) was obtained for the correlation coefficient, supporting the observation that there is no apparent correlation between the two coefficients. No clear correlation between the two coefficients indicates that the two uranyl peroxides, studtite and metastudtite, are independently generated on the U 3 O 8 surface. It should be noted that the metastudtite observed on the U 3 O 8 surface is not generated through the thermal decomposition of studtite on the surface because if the observed metastudtite was generated from studtite, the quantity of metastudtite increases as that of studtite decreases, and this tendency is not observed in Figure 5. Furthermore, we checked whether metastudtite is generated on the U 3 O 8 surface during the drying process of the immersed U 3 O 8 , where the immersed U 3 O 8 was dried in a vacuum desiccator for 3 months at room temperature. However, no increase in the amount of metastudtite was observed due to the drying process, supporting the conclusion that studtite and metastudtite are individually generated on the surface of U 3 O 8 .

Conclusion
The distributions of studtite and metastudtite generated on the surface of U 3 O 8 through the immersion of U 3 O 8 in H 2 O 2 aqueous solution were successfully observed using Raman imaging technique. It was revealed that the two types of uranyl peroxides cover the U 3 O 8 surface and were heterogeneously distributed over the surface, while studtite was generated at ~1.5 times higher yield compared to metastudtite. Any correlations between the distributions of studtite and metastudtite were not observed, suggesting that the two uranyl peroxides are independently generated on the surface of U 3 O 8 . The present study demonstrates that Raman imaging technique could potentially be used to obtain information about the formation and distribution mechanism of the alteration phases on the nuclear fuel debris in the Fukushima-Daiichi nuclear power plants.