Ghada
El Jamal
*a,
Thomas
Gouder
b,
Rachel
Eloirdi
b and
Mats
Jonsson
a
aSchool of Engineering Sciences in Chemistry, Biotechnology and Health (CBH), Department of Chemistry, Applied Physical Chemistry, KTH Royal Institute of Technology, SE-100 44 Stockholm, Sweden. E-mail: ghadaej@kth.se
bEuropean Commission, Joint Research Centre, Directorate for Nuclear Safety and Security, Postfach 2340, DE-76215 Karlsruhe, Germany
First published on 16th March 2021
Thin UO2 films exposed to water plasma under UHV conditions have been shown to be interesting models for radiation induced oxidative dissolution of spent nuclear fuel. This is partly attributed to the fact that several of the reactive oxidizing and reducing species in a water plasma are also identified as products of radiolysis of water. Exposure of UO2 films to water plasma has previously been shown to lead to oxidation from U(IV) to U(V) and (VI). In this work we have studied the dynamics of water plasma induced redox changes in UO2 films by monitoring UO2 films using X-Ray photoelectron Photoemission (XPS) and Ultra-Violet Photoemission (UPS) spectroscopy as a function of exposure time. The surface composition in terms of oxidation states obtained from U4f7/2 peak deconvolution could be retraced along the exposure time, and compared to the valence band. The spectral analysis showed that U(IV) is initially oxidized to U(V) which is subsequently oxidized to U(VI). For extended exposure times it was shown that U(VI) is slowly reduced back to U(V). UPS data show that, unlike the U(V) formed on the surface upon oxidation of U(IV), the U(V) formed upon reduction of U(VI) is localized in the bulk of the film. It also displays a different reactivity than the initially formed U(V). The experiments can be reproduced using a simple kinetic model describing the redox processes involved.
Research on oxidative dissolution of UO2 has been conducted quite extensively using various material types as powder, pellets, crystals, films of UO2 or simulated spent fuels.7,11–17 Such studies include interactions between aqueous radiolysis products and UO2 surfaces, conducted in chemical and electrochemical experiments.1,18–22 Some studies were made by continuously exposing uranium oxide to low steady-state concentrations of oxidants produced by irradiating suspensions of uranium oxide or pellets immersed in aqueous solution.8 Other studies focused on the leaching of pellets (pure or doped with non-radioactive isotopes of fission products)23–27 and even of thin films.18,28,29 The impact of dose rate, pH, temperature, oxidant or reductant concentration/pressure, bicarbonate and other groundwater constituents, have been assessed.14,19,25,26,30–32
During the last decade, considerable progress has been made in the use of surface spectroscopy to study thin films, including uranium oxide films under UHV conditions.18,28,33 Very recently, X-ray Photoelectron Spectroscopy (XPS) and UV Photoelectron Spectroscopy (UPS) were used to study changes in uranium oxide thin films exposed to O2, H2 and H2O plasmas.34 The plasmas were used to mimic the impact of various water radiolysis products, thereby providing a novel model system for radiation induced oxidation of uranium oxide. This model system is particularly useful for studies of the mechanism of oxidation of UO2 (and other oxidation states) as it does not involve dissolution of oxidized uranium oxide or deposition of secondary phases due to solubility limitations. The oxides previously under study were UO2, U2O5 and UO3. Interestingly, upon exposure to H2O-plasma at 20 °C, UPS showed that the final state of the surface layer of the oxide was U(VI) regardless of starting material. However, at 400 °C, XPS reveal differences depending on the starting material.
UO2 has a fluorite lattice structure, which can accommodate oxygen at high temperature to form non-stoichiometric oxides UO2+x.32 For x < 0.5 the fluorite structure is preserved, while for x > 0.5 a layered structure forms. For example, U3O8 is a thermodynamically stable phase consisting of a mixture of U5+ and U6+.35 U2O5 was previously assumed to be characterized by a mixed valence state.36 However, recent XPS characterization of thin U2O5 films produced by reducing UO3 with H2 plasma shows that U(V) is the only oxidation state.37 UO3 exists in seven different crystallographic phases and an amorphous phase.32 Recent studies on radiation induced oxidation of UO2-pellets reveal that considerable amounts of U(V) can be found on the surface of the pellet after exposure to alpha-irradiated aqueous solutions.38 This implies that the UO2-surface has been one-electron oxidized and that complete oxidation of UO2 is most probably attributed to two consecutive one-electron oxidation steps, very much in line with previous mechanistic studies.10,11 To shed more light on the detailed mechanism of radiation induced oxidation of UO2 we have studied ECR water plasma-induced transformation of thin UO2 films produced by sputter deposition as a function of exposure time. The water plasma is largely composed of the same constituents as are formed in radiolysis of water and the experiment therefore serves as a model for radiation induced oxidation of UO2.39 The majority of the experiments were carried out at 400 °C to allow fast diffusion and formation of homogeneous compounds under equilibrium conditions. For the same reason thin films of 20 nm thickness were used.
Surface speciation was done by XPS and UPS, following U 4f, O 1s and valence band regions. The oxidation state of uranium was determined by deconvolution of the U 4f7/2 core level peak into its U(IV), U(V) and U(VI) components. Surface characterization is also based on the binding energy shift and peak broadening of U 4f5/2 and O 1s, on the U 4f5/2 satellites, and on the valence band region with the O 2p valence band and the U 5f level.
A Pyrolytic Boron Nitride (PBN) heater installed below the sample holder maintained the sample temperature at 400 °C during the exposure to water plasma.
![]() | ||
Fig. 1 Core level X-ray photoemission spectra of U 4f recorded for unexposed UO2 film, and 9 other UO2 films exposed to water plasma (WP) for different length of time. |
The U 4f/O 1s surface area ratio swiftly decreases, reflecting incorporation of oxygen atoms into the lattice of the oxide and pronounced oxidation of the UO2 film. All these observations confirm progressive quantitative oxidation of UO2 by the water plasma. The oxidation of UO2 involves incorporation of O2− ions into vacant interstitial locations in the UO2 cubic fluorite structure accompanied by the charge-balancing conversion of adjacent U(IV) cations to U(V).1
In the time interval of 9–15 minutes of exposure to the water plasma, the U 4f main peaks shift to slightly higher BE. The U 4f/O 1s intensity ratio passes through a minimum corresponding to that of UO3. In addition, the FWHM reaches a minimum. This stage presents the maximal oxidation of the film (to almost pure UO3). The small FWHM proves that only one compound is present. Two oxidation states with their shifted U 4f peaks would produce a broader composite peak. The maximum surface oxidation of UO2 is reached after ca. 10 minutes of exposure to the water plasma. After an exposure longer than 10–15 minutes, the U 4f/O 1s intensity ratio increases indicating that the oxide is being slowly reduced.
The U 4f/O 1s intensity ratio increase is accompanied by a FWHM increase, indicating superposition of U 4f peaks with different binding energies. After 60 minutes of water plasma exposure, the two oxidation states become clearly distinct as shown in Fig. 1. The BE difference is about 0.8 eV which is close to the BE difference between U(VI) and U(V) reference lines. After 90 minutes exposure to water plasma the peak assigned to U(V) becomes predominant, indicating that further reduction of U(VI) has taken place.
Fig. 3 shows the O 1s region of the starting UO2 film and the films exposed to water plasma for 2, 10 and 60 minutes. The reference line of U2O5 was added for comparison. As can be seen, it is almost overlapping with the spectrum of the film exposed to water plasma for 2 minutes. The exact values of binding energy peak position and FWHM are displayed as functions of exposure time with reference peaks of UO2, U2O5 and UO3 from the literature.37 Upon exposure to water plasma, the FWHM of the O 1s peak (after 2 minutes) first increases to the value corresponding to U2O5 and then (after 10 minutes) increases to the value corresponding to UO3. After 60 minutes, the FWHM has decreased again to a value between UO3 and U2O5. This is consistent with UO3 formation and the intensity increase of the U(V) satellite peak in the U 4f scans of the same film as will be shown below. Evolution of the peak binding energy is less informative. The shift is mainly due to changing work function and Fermi-energy in the different oxides. This is a rigid shift affecting all photoemission peaks (see e.g. the valence region in Fig. 5) of both U and O. In all cases, oxygen is at the state O2− in contrast to uranium, whose oxidation state varies from +4 to +6. So, there is no reason to expect significant chemical BE shifts (used for speciation in photoemission) for the O 1s.
A close plot of the satellite region after different water plasma exposure times is displayed in Fig. 4. The satellite for the unexposed UO2 film is characteristic for U(IV). After only 1 minute of water plasma exposure, the U(V) satellite peak appears and partly replaces the U(IV) satellite. The early stage of oxidation of UO2 by water plasma (from 1 to 6 minutes) produces a mixture of U(IV) and U(V). For slightly longer exposures (from 9 to 15 minutes), the satellite lines of U(IV) and U(V) gets weaker with increasing exposure time. This is accompanied by the appearance of the U(VI) satellite peak. For even longer exposures to water plasma (30 to 90 minutes), the U(VI) satellite peaks gradually become weaker and after 60 minutes of plasma exposure, the U(V) satellite peak becomes the dominating satellite peak.
![]() | ||
Fig. 4 U 4f5/2 satellite region recorded for unexposed UO2 film, and initial fresh UO2 films exposed to water plasma (WP) for different times. |
For intermediate exposure times (9 and 30 minutes), the satellite peaks are much less pronounced than in pure, crystalline oxides. Obviously, their intensity not only depends on the oxidation state but also on additional factors, in particular defect concentration affecting the lifetime induced broadening. Ion sputtering of a surface is known for inducing structural disorder and amorphisation.42–44 It has been previously observed that uranium oxide surfaces sputtered with Ar ions, amorphise the surface and introduces defects.43 Also a previous study on UO2 film showed that its oxidation (into UO2+x) leads to less defined U 4f satellite intensity peaks.45 Conversely, sample annealing, which restores the crystal order,46 intensifies the satellite peaks.46 In the present case by oxidation, the plasma alters the initial fluorite structured UO2 into layer structured UO3. This process destroys crystal order. The longer water plasma exposure times (60 to 90 minutes) at 400 °C work as annealing and allow the film to order again. As a direct consequence, the satellites become more pronounced. Hence, satellite intensities are useful, but not unambiguous criteria for oxidation states. It is therefore important to combine the analysis of satellite peaks with other sources of information, such as valence region spectra, which give an independent indication on the oxidation state via the intensity and width of the U 5f emission (see below).
The black curve shows the spectrum of the unexposed UO2 film. The peak from 2 to 8 eV is due to the O 2p valence band, with two prominent features at 4.5 and 7 eV, related to the band structure.37 The peak at 1.2 eV is attributed to the U 5f emission. In UO2 this peak corresponds to the U 4f2 level, which is a localized (non-bonding) state. After water plasma exposure the U 5f peak shifts to lower BE, together with the O 2p valence band signal. This is again due to a rigid BE shift related to work function and Fermi-energy change, as mentioned for the O 1s line. The U 5f peak narrows (already after 2 minutes) and its intensity has drastically decreased after 10 and 60 minutes of exposure. For further discussion, we show two representations of the spectra, one normalized on the O 2p intensity (Fig. 5a) and one on the U 5f intensity (Fig. 5b).
The films exposed to water plasma had almost the same peak width for all three exposure times. This width corresponds to that of pure 5f1 of U2O5, which is narrower than that of the 5f2 line of UO2. This is due to the different electron configurations (5f2 and 5f1) leading to different final state multiplet structures for UO2 and U2O5. Photoemission from the 5f2 initial state of UO2 leads to the 5f1 final state with a doublet structure (5f5/2 and 5f7/2) due to spin–orbit interaction. Photoemission from the 5f1 initial state of U2O5 leads to the 5f0 final state, with a singlet structure (no spin–orbit interaction). Therefore, the 5f emission in U2O5 is narrower than in UO2. Also, the intensity of 5f1 is lower than that of 5f2. One would expect a reduction by a factor of about 5, but it is hard to obtain a quantitative analysis from the figure, because there is no intensity reference: the 5f line decreases with exposure while the O 2p grows because of O incorporation and change in the type of bonding of the U–O bond (affecting the photoemission cross-section of the valence band). The overall effect is a strong decrease of the U 5f/valence band intensity ratio. For longer surface exposures (>2 minutes) the U 5f decreases in intensity but keeps the same width, (5f1). We conclude that U4+ disappears very fast (mostly after 2 minutes exposure) and the oxidized films are a mixture of the two oxidation states: U(V) and U(VI). The change in intensity is due to a change in the proportion of each valence state: U(V) has a 5f1 signal while U(VI) has a 5f0 initial state configuration, i.e. no 5f line at all.
As mentioned above, the water plasma exposures were done at 400 °C to ensure fast diffusion of oxygen through the film and homogeneity of compounds. However, UPS measurements reveal that the top layers are not homogeneous. This is shown in Fig. 5, where the same energy region (0 to 12 eV) as the valence region monitored with XPS has been monitored with the significantly more surface sensitive techniques UPS (information depth of 1 monolayer instead of 6 monolayers). This is due to the mean free path of UV photons which is much smaller than that of X-ray photons due to the relatively small optical depth (the scattering cross section of UV photons is much larger than that of X-ray photons).47 UPS shows the same trends in intensity and position of the U 5f and the valence band as XPS. The different 5f/VB intensity ratios for the same water plasma exposure times are due to the different cross-sections in UPS and XPS.48 However, the complete disappearance of the 5f after 60 (90) minutes exposure cannot be explained by cross-section effects but is due to the complete oxidation of the outermost surface layer. This difference in oxidation is not due to a lack of diffusion (kinetics) but to the difference in oxide stability at the surface and in the bulk. Oxygen is well known to accumulate to the surface of incompletely oxidized compounds because it lowers their surface energy.
Oxidation states | Binding energy of U 4f7/2 (eV) | Binding energy of U 4f5/2 (eV) | Spin orbit splitting of U 4f main lines (eV) | FWHM of U 4f7/2 (eV) | Satellite peaks separation from U 4f5/2 (eV) |
---|---|---|---|---|---|
U(IV) | 380.1 | 390.9 | 10.7 | 1.5 | 6.7 |
U(V) | 380.4 | 391.3 | 11.2 | 1.46 | 8.1 |
U(VI) | 381.1 | 391.8 | 10.7 | 1.2 | 4.4 and 9.9 |
Fig. 6 shows results from peak deconvolution for the UO2 films after 2, 10 and 60 minutes water plasma exposure, respectively. The XPS scans were better fitted using all three reference peaks, even when the concentration of one of the valence states was very low. A larger number of parameters (peaks) always provides a better fit, still we tentatively apply it to have all three oxidation states in our consideration (even if valence spectra point to the preponderance of only two, (V) and (VI), in most cases).
![]() | ||
Fig. 6 Deconvolution of U 4f7/2 main peak of UO2 films exposed to water plasma for (a) 2 min, (b) 10 min and (c) 90 min. The deduced fractions of uranium oxidation states are plotted below each case. |
The deconvolution of the U 4f7/2 peak after 2 minutes of water plasma exposure, Fig. 6a, implies that U(IV), U(V) and U(VI) are concomitant. When the exposure time is extended to 10 minutes, Fig. 6b, the dominant oxidation state according to the deconvolution is U(VI) (∼80%) with slight contributions of U(IV) and U(V). This is consistent with the shape of the satellite peaks of the U 4f scan of the same film plotted in Fig. 4 (even though the satellite point to a smaller (IV) and (VI) contribution after 2 minutes). The satellites after 10 minutes of water plasma exposure are matching with U(VI) satellite lines, rising at 4.4 eV and 9.9 eV higher binding energy than the U 4f5/2 main line.
The fitting result somewhat deviates from the XPS valence band spectrum, which shows principally U(V) – however curve fitting also indicates a majority of U(V) (70% versus 30% U(IV)). So there is qualitative agreement. In Fig. 6c, the deconvolution of the peak shows that the fraction of U(V) has increased significantly after 90 minutes of water plasma exposure when compared to the film exposed for 10 minutes.
The results of all the spectral analyses presented above are internally very consistent and show that when exposing a UO2 film to a water plasma under UHV conditions, U(IV) is rapidly converted to U(V) and subsequently to U(VI). For extended exposures to water plasma, the fraction of U(VI) slowly decreases and the fraction of U(V) increases with the same rate. To verify that the slow reduction of U(VI) is indeed caused by the exposure to the water plasma and not just by the prolonged heating of the film, a control experiment was performed. In this experiment a UO3 film was exposed to water plasma for 10 minutes at 400 °C. Thereafter, the water plasma was switched off and the film was kept at 400 °C under UHV conditions for 50 minutes. During this time, the U(VI) fraction decreased by ca. 10% which is considerably less than the reduction observed upon exposure to the water plasma for the same time period. This verifies that the slow reduction of U(VI) to U(V) observed in the plasma exposure experiments should mainly be attributed to the plasma exposure itself and not to thermal decomposition of UO3.
![]() | ||
Fig. 7 The fraction of uranium oxidation states deduced from the simulation and the U 4f7/2 peaks deconvolution results, plotted as function of exposure time to water plasma. |
To derive a mechanism and possibly to account for the dynamics of this process, we must first look at the properties of the water plasma. The water plasma contains H2, H˙, HO˙ and O and recombination products thereof, including O2. Hence, both reducing and oxidizing species are present. It has recently been shown that oxidation of UO2 by molecular O2 is much less pronounced than the oxidation by water plasma under comparable conditions. Consequently, O2 can be ruled out as the main oxidant here. However, a pure oxygen plasma where the main constituent is atomic oxygen, is considerably more efficient than the water plasma in oxidizing the UO2 film. The oxidation of the UO2 film exposed to water plasma can therefore, at least partly, be attributed to atomic oxygen but also to hydroxyl radicals. The actions of the oxidants are partly balanced by the actions of the reducing species and the magnitude of oxidation is also related to the concentration of oxidants in the plasma. For a tentative one-electron oxidation step we could write the following general reactions:
U(IV) + PlasmaOx → U(V) | (1) |
U(V) + PlasmaRed → U(IV) | (2) |
It has previously been shown that U(VI) as well as U(V) exposed to H2-plasma are quantitatively reduced to U(IV) within 10 minutes at 400 °C and that U(IV) is quantitatively oxidized to U(VI) by an O2-plasma under the same conditions. Considerably shorter exposures at lower temperature reveal that U(V) is formed as an intermediate between U(IV) and U(VI). U(V) exposed to water plasma is partly converted to U(VI) as shown by XPS. At the surface, the conversion is quantitative as shown by UPS. From these observations we can conclude that hydrogen atoms are capable of reducing U(VI) to U(V) and U(IV) and U(V) to U(IV) and that atomic oxygen can oxidize U(IV) to U(V) and U(VI) and U(V) to U(VI). As atomic oxygen as well as hydrogen atoms are also constituents of the water plasma, it is clear that a water plasma has the potential to drive oxidation of U(IV) all the way up to U(VI) and reduction of U(VI) all the way down to U(IV). The dynamics of these processes would then depend on the relative rate constants of the reactions involved and on the plasma composition (which is temperature dependent). The simplest mechanistic approach would be to assume that we can apply the general scheme presented above (i.e., that plasma oxidants can drive a one-electron oxidation step and that this is counteracted by plasma reductants reducing the one-electron oxidized product back to the original oxidation state). For the system studied here we would get the following reaction scheme:
U(IV) + PlasmaOx → U(V) | (3) |
U(V) + PlasmaRed → U(IV) | (4) |
U(V) + PlasmaOx → U(VI) | (5) |
U(VI) + PlasmaRed → U(V) | (6) |
Admittedly, the kinetics for adsorption and desorption of the plasma products on the surface should also be included in a complete mechanism. The surface is being continuously exposed to a flux of the plasma and there is no accumulation of plasma products. For this reason, the rate of each step can be expressed as being proportional to the flux of the active plasma constituent multiplied with the surface fraction of uranium in the oxidation state of relevance. In our system we start with pure U(IV) and if we apply the mechanism above, we would reach a state where the fractions of the three different oxidation states become constant (i.e. steady-state). Depending on the relative fluxes of the plasma products and the relative rate constants for each step, the final state could either contain all oxidation states or only U(VI).
As it is quite clear from the deconvolution results in Fig. 7 (symbols) and from the spectral analyses presented above that we do not reach this steady-state situation, we must conclude that the mechanism presented above is not sufficiently describing the system. What we can see in Fig. 7 is that U(VI) is slowly being reduced to U(V) and that this process does not appear to be balanced by oxidation of U(V) back to U(VI). It is interesting to note that UPS-data (in Fig. 5) show that the surface of the film is initially oxidized from U(IV) to U(V) and then to U(VI) but no slow reduction of U(VI) to U(V) is observed for longer exposure times. This implies that the slowly produced U(V) is formed deeper in the film and therefore displays a different reactivity. This can be accounted for by introducing an additional reaction to the overall mechanism. In Fig. 7 (lines), the results of a numerical simulation (using the software MAKSIMA49) taking this mechanism into account is also presented. As the exact plasma composition is not known at this point, the mechanism only includes the five redox reactions stated above (3 to 6). As a consequence, the contributions from adsorption and desorption and the possible diffusion from the film surface to deeper layers of the film are all assumed to be included in the rate coefficients used for the five redox reactions.
As can be seen, the numerical simulation reproduces the experimental (XPS) results fairly well. In these simulations the two oxidation steps have the same first order rate constant (i.e. the oxidant concentration multiplied with the second order rate constant for oxidation, kox[Ox]) and the two reduction steps also have the same first order rate constant (kred[Red]). The simulation results are obtained with a ratio of 8.6 between the first order rate constants for oxidation and reduction. The first order rate constant for the reaction leading to formation of the U(V) species inside the film (i.e. U(V)b) is a factor of 10 lower than for the other reduction steps. The overall dynamics is fairly insensitive to the ratio between the first order rate constants for oxidation and reduction (in the reversible part) as long as it is higher than 8.6, provided that the rate of oxidation is not changed. However, a higher ratio than 8.6 (i.e. slower reversible reduction) leads to higher maximum U(VI)-concentration than what is found experimentally. This verifies that the water plasma is overall oxidative in the uranium oxide system. The overall mechanism is depicted in Scheme 1.
This work has been partially supported by the ENEN+ project that has received funding from the Euratom research and training Work Programme (2016-2017-1#755576).
We thank Frank Huber for excellent technical assistance.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1dt00486g |
This journal is © The Royal Society of Chemistry 2021 |