Monitoring the reduction of UO 3 thin film by hydrogen atoms using valence-level spectroscopy: correlating the U5f 1 signal to surface hydroxyls

The reaction of a UO 3 thin film with atomic hydrogen was studied by He(II) ultraviolet photoelectron spectroscopy (UPS) in the temperature range 190–300 K. UO 3 reduction was instantaneously observed once it contacted H atoms at 10 –7 torr. The reduction was manifested by the presence of U5f 1 electrons in He(II) UPS at approximately 1.5 eV below the Fermi level. Based on the peak characteristics, the valence band shape (composed largely of O2p orbitals in addition to some contribution from U6d and U5f orbitals), and X-ray photoelectron spectroscopy (XPS) U4f lines, the reduction of U 6+ in UO 3 only results in the formation of U 5+ cations and was largely limited to those on the surface. Associated with the reduction was the formation of surface hydroxyls (-OH species) due to the transfer of a proton of the H atom (H.) to surface oxygen ions, while the electron of H. is transferred to a U5f orbital. The pseudo-first-order rate constant of the initial rate of reduction at 10 –7 torr and 190 K was found to be approximately 0.01 s –1. Qualitative analysis of the valence band before and after reduction indicates that O2p hybridization with U6d and U5f orbitals leads to well-distinguished features that are characteristic of UO 3, U 2 O 5, and UO 2. These features, which were quantitatively reversed during the redox process, furthers the assessment of the stoichiometry of a given binary uranium oxide.

The study of the oxidation and reduction of uranium oxide surfaces is of fundamental and technological importance. Our knowledge of the redox properties of the surface of actinide oxides considerably lags behind that of early transition metal oxides (Cossard et al., 2021). This is in part due to the lack of model surface—of the uranium oxides series, only bulk UO 2 single crystals are available (Idriss, 2010), although epitaxy of UO 2, α-U 3 O 8, and α-UO 3 on oxides can be made (Mudiyanselage et al., 2019; Enriquez et al., 2020)—and due in part to the complexity of the system of one of the richest known oxides (Cotton and Wilkinson, 1988). For example, UO 3 has seven known polymorphs and a pure U 5+ cation oxide system, while U 2 O 5, has only recently been identified (Gouder et al., 2018).

On the technological aspect, the oxidation of the surface of UO 2 to UO 3 or other mixed oxides containing U 6+ is particularly important because the latter is soluble in water and is therefore of environmental concern for geological repositories of spent nuclear fuel (Valsami-Jones and Ragnarsdöttir, 1997). Moreover, because of continued radioactivity, molecular and atomic hydrogens are produced by the radiolysis of groundwater and may interact with the surface, potentially leading to the reduction of the oxidized species back to U 4+ or U 5+ cations. This would counteract the radiation-driven oxidative dissolution of the UO 2 fuel matrix.

The ease of oxidation and reduction of uranium oxides—partly due to its relativistic effect and the fact that the 5f orbitals partly contribute into the valence band (Su et al., 2016)—makes it a reactive oxide for redox catalytic (or surface) reactions. Indeed, it was found that U 3 O 8 has high activity for water–gas shift reaction (Dong et al., 2015), that uranium antimonate catalysts also have high activity for the ammoxidation of propylene to acrylonitrile (Foster et al., 2020), that uranium oxide solid solution with cerium oxide shows potential for thermochemical water splitting (Al-Shankiti et al., 2013; Scaranto and Idriss, 2015), and that β-UO 3 drives the oxidative coupling of acetylene (Madhavaram and Idriss, 2002) to furan and aldol condensation of acetaldehyde (Madhavaram and Idriss, 2004).

There is another reason to study the reduction of UO 3 by UPS. Recently, we have observed its reduction under the UV photons of He(II) light (40.8 eV) when ice is formed on top (El Jamal et al., 2023). The reduction was manifested by the presence of electrons in the U5f band at ≈ 1.5 eV below the Fermi level (E F). Associated with this reduction was the presence of surface hydroxyls (3σ) at ≈ 10.8 eV below E F and a change in the O2p band shape due to a change in hybridization with U5f orbitals. While this can be studied computationally, the shape and binding energy position of the O2p/U5f/U6d band changes significantly with the U eff value (mostly the Hubbart parameter) used in density functional theory (DFT) + U studies (Singh et al., 2018) and by the degree of Hartree Fock (HF) mixing with DFT for the hybrid methods (Prodan et al., 2007; Roy et al., 2008). Both methods may therefore benefit from the detailed valence study of uranium oxides.

UO 2+x (x < ca. 0.1) thin film was synthesized by reactive direct-current (DC) sputtering from a uranium target; the sputter gases used were Ar and O 2. The UO 2 film was deposited on a gold foil (99.99% purity) that had been cleaned by Ar ion sputtering, then annealed to 200°C for 10 min. The plasma in the diode source was maintained by injecting 25–50 eV energy electrons; the O 2 pressure was 2 × 10 −6, and the Ar pressure was 10 –3 torr. After film formation and to produce stoichiometric UO 2.0, the initial film was exposed to atomic hydrogen produced by an electron cyclotron resonance (ECR) plasma source at 400°C for 10 min. This eliminated surplus oxygen, leaving UO 2.0. The UO 3 film was prepared by exposing the UO 2 films to atomic O under the same conditions (ECR exposure at 400°C for 10 min). More details on the characterization of the films can be found in El Jamal et al. (2021); the procedure is described in detail in Gouder et al. (2018) and shown in Scheme 1. The content and purity of the films were determined by XPS. The positions of the U4f main lines, their FWHM, and the characteristic satellite peaks are a direct probe of the oxidation states of the films (Al-Salik et al., 2014).

Thin film deposition, plasma treatment, gas exposures, and data acquisition were carried out in situ, and all chambers (growth and spectroscopy) were interconnected. The background pressure was lower than 3 × 10 −7 Pa. Low temperatures were reached by cooling the copper sample holder with liquid nitrogen.

The ion-free H 2 gas beam was produced by a H atom beam source (HABS), which is a thermal gas cracker. It was heated by an e-beam, which has the advantage of avoiding ion-induced damage to the substrate. The operational parameters are the flow rate of H 2 and the heating power which controls the intensity of the source. The narrow beam of the HABS allows high atomic H flux rates at the sample position while keeping the H 2 background pressure of the chamber lower than the plasma sources. The adsorption was made at low temperature (typically lower or approximately 200 K).

UPS spectra were taken with He(II) (40.81 eV) UV light which was produced by a high-intensity windowless discharge lamp. XPS spectra were recorded using monochromatized Al K α (1,486.6 eV) radiation produced by a SPECS μ-focus source.

Figures 1A, B present the XPS and UPS of relevant uranium oxides thin films prepared as references. These were presented in El Jamal et al. (2023) and given here to highlight the main differences in the oxidation states of uranium cations. The known oxidation states of U cations are +4, +5, and +6, and these can be distinguished by their binding energy, their shapes (such as their full-width half maximum—FWHM), and by the presence of their satellites. Here, U 4+ cations have a U4f 7/2 line at 380.3 eV and U4f 5/2 line at 391.1 eV; associated with both are their shake-up satellites at a binding energy of approximately 6.9 eV above each line (denoted S 1 and S 1 *). The binding energy values are with a margin of ±0.2 eV, depending on the purity of the samples (the unavoidable presence of surface hydroxyls) that may slightly shift the Fermi level energy. No attempt was made to further calibrate it; the film thickness is in the order of 20 nm and is conductive. U 6+ cations have a U4f 7/2 line at approximately 381 eV and a U4f 5/2 line at 392 eV binding energies. They are sharper than those of U 4+ cations (narrower FWHM). These lines have two pairs of satellites (at approximately 4.0 and 9.6 eV above each line, denoted S 3, S 4, S 3 *, and S 4 *). More details about the binding energy positions can be found in Ilton et al. (2007) and about the satellites in Van den Berghe et al. (2000), Ilton and Bagus (2008), and Tobin and Yu (2011). For simplicity, XPS of U 5+ cations of U 2 O 5 are not presented in the figure and are discussed below. UPS He(II) spectra of UO 2, U 2 O 5, and UO 3 are presented in Figure 1B. The lines provide information related to the oxidation states and degree of hybridization between the U5f and the 6d orbitals with O2p orbitals (Magnuson et al., 2006; Tobin and Yu, 2011; Kvashnina et al., 2018). The difference in hybridization is seen in the binding energy region between 4 and 8 eV and in the filling of the U5f orbitals at approximately 1.5 eV below the Fermi level. The weak signal with a binding energy of approximately 10 eV is that of surface hydroxyls. U 4+ cations in UO 2 are distinguished by their pronounced U5f line with a large FWHM (approximately 1.5 eV) due to the presence of two electrons in the 5f orbitals. U 5+ cations have their U5f signal at the same binding energy position as that of U 4+ cations but are much less pronounced with a narrower FWHM (approximately 1.0 eV). U 6+ cations, having no electrons left in the 5f orbitals, do not give a signal in this region. It should be noted that the oxidation states are nominal because of changes in the degree of ionicity/covalency when the U-to-O ratios change and because of the different polymorphs. The region largely composed of O2p orbitals has distinguished differences for the three oxides and is addressed here. In UO 2, the region has two observed peaks at approximately 5 and 7.5 eV, with the first more pronounced than the second. UO 3 is also largely composed of two components but with an inverted relative contribution at approximately 4 and 6.5 eV. U 2 O 5 is composed of one wide peak at approximately 5.5 eV in binding energy due to the convolution of the two structures. The degree of hybridization of O2p with U5f and 6d orbitals has been the subject of other research, and some information may be obtained upon comparison with these previous computation studies of the projected density of states (p-DOS) on the orbitals of the U and O ions (Tian et al., 2014; Ao et al., 2016). U5f and U6d orbitals were found to hybridize with O2p in this region. The degree of hybridization would therefore respond to the electron density around uranium cations. In this research, it is inversely proportional to the degree of orbitals filling. It is highest for U 6+ cations in UO 3, lowest in U 4+ in UO 2, and somewhere in the middle for the U 5+ cations of U 2 O 5. Therefore, UO 2 would have a higher contribution of O2p orbitals and UO 3 would have a smaller contribution. This picture, while qualitative, is important as it helps track the degree of oxidation and would be revealed as important while studying the reduction of UO 3 with atomic H, as presented next.

Figures 2A–C present the results of the exposure of fresh UO 3 to atomic H at 190 K and at the indicated exposures in Langmuir. Exposures were conducted at 10 –7 torr for different durations. The fresh UO 3 had some traces of reduced states as it was a different preparation from that in Figure 1. This trace signal was subtracted for the quantitative analyses presented next.

Three clear changes in the spectra are seen with increased exposure to H atoms.

First, the O2p region is gradually changed and becomes similar to that of the U 2 O 5 presented in Figure 1. Second, the reduction of UO 3 is clear, as seen by the increased signal of the U5f line at approximately 1.5 eV below the Fermi level. Third, there is also an increase, associated with the increased U5f lines, in the surface hydroxyls at approximately 10 eV.

The spectra are presented as obtained with no binding energy calibration, highlighting the shift to higher binding energies with increased exposure to H atoms. This shift is attributed to bend banding at the surface due to exposure. The shift to higher binding energy indicates a change in semiconducting properties to a more pronounced n-type (electron-doping) due to the increased electron density (reduction of U 6+ cations), although the presence of surface hydroxyls would certainly affect the binding energy shift, as seen on other oxides (Pacchioni and Freund, 2013; Zhang et al., 2015).

Additionally, there is a decrease in the overall O2p–U5f–U6d signal with increased reduction. This is expected because part of the O2p contribution is transformed to surface hydroxyls and part of the contribution of the 5f and 6d orbitals in the hybridization has decreased due to reduction (filling of the U5f). Denning (2007) and Neidig et al. (2013) have studied this using polarized oxygen K α X-ray emission and absorption spectra assigned to transitions from O 1s core electrons to different molecular orbitals of Cs 2 UO 2 Cl 4 Os. The participation of the 6pσ orbitals in covalent bonding is seen by the presence of a charge transfer transition in the emission. The data indicate that the 5fπ component in the U–O bond is weaker than the 5fσ component; the latter is stronger due to hybridization with the 6pσ. The data also indicate that the 6dπ interaction is stronger than the 6dσ. It is therefore inferred that changing the degree of oxidation will affect the degrees of hybridization manifested in the valence band region.

To further probe these changes, Figure 3 presents the reverse changes that occurred upon mild heating of the reduced surface. Figure 3A shows that, upon heating, all features in Figure 2A have reversed. The overall O2p (U5f and U6d) increased in intensity, the -OH and U5f contribution decreased, and the binding energy shifted back to lower values. In addition, the structure of the O2p slightly changed from that of the most reduced state to start the reverse process of oxidation. Therefore, the surface is composed of largely reduced cations and some U 6+ cations. This last point is better seen in the inset of Figure 3B, where the same spectra shown in Figure 1A are given to ease the comparison. The principal point is the presence of two satellites S 2 and S 2 * at 388.0 and 398.8 eV, not seen in UO 3 nor in UO 2; these are those of U 5+ of U 2 O 5 and are therefore evidence of partial reduction. The main lines are composed of two structures separated by approximately 1 eV. In the U4f 7/2, these are at 380.5 and 381.5 eV and fitted in consideration of the FWHM of the U 6+ (in the inset). The fitting parameters are presented in Supplementary Material Table 1.

Quantitative analyses of the signals are presented in Figure 4, which presents the computed peak areas of the U5f, OH (3 σ), and O2p-(U6d and 5f) regions. The peak areas are presented as fractions. The trace presence of U 6+ at zero exposure is subtracted. Both the U5f and OH (3 σ) signals increase with exposure and then reach saturation at high exposure (at approximately 200 L, 2000 s at 10 –7 torr). Both could be fitted reasonably well with a bi-exponential function (R 2 > 0.98). There is, however, a difference between the U5f and -OH profiles: the OH signal saturates faster than that of the U5f. This may be due to errors in computing the peak areas, particularly because the OH (3 σ) signal is at the edge of the large valence band peak. There might, however, be a distinction between the reduction of U 6+ cations and surface hydroxyls. The former may occur in deeper layers while the latter would be restricted to the surface and therefore saturate first. The kinetic parameters extracted are given in Table 1 (the peak areas were computed without further correction). As indicated above, the inverted time constants between the appearance of the U5f and OH signals are intriguing and might be linked to surface (in the case of -OH groups) versus surface and near-surface (in the case of the U5f). This is because electrons can migrate to deeper layers while protons cannot.

The reduction of U 6+ cations by H atoms of UO 3 is presented by Equation 1: where l stands for lattice, g for gas, and s for surface.

The interaction of a H atom with the surface of UO 3 occurs via a redox-type mechanism. The surface is composed of two centers: the U 6+ cations and the O 2- anions. An incoming H atom is repelled by an O 2- center due to Pauli electrostatic repulsion and the fact that an O 2, with eight electrons (2s 2 and 2p 6) in the outer shell, cannot accommodate an additional electron; it therefore interacts with a low electron density center (U 6+). This results in the transfer of an electron to one of the 5f-orbitals of U 6+, while the remaining proton interacts with a lone pair of the O2p orbital of surface oxygen. The reaction will be concerted where both events occur simultaneously because the formation of a U-H bond is not possible from U 6+ and one H atom (one electron is missing). The end-product is a pair U 5+ –OH (s), as tracked by He(II) UPS (Figures 2, 3).

The exponential dependence of the rise of the U 5+ cations may be explained by a simple kinetic reaction: where t is time at a pressure of 10 –7 torr and a and b are the reaction orders. Considering that [H. ] is largely excess compared to the UO 3 thin film, b equates to 0. We did not monitor the concentration of [U 6+ ] due to the absence of the U5f UPS signal, but we monitored the formation of [U 5+ ] cations. We assume that the total amount of reactive U 6+ cations on the surface (and near-surface) has been reduced with H atoms, since these reached saturation with an increasing exposure time above ca. 250 s (at 10 –7 torr); other measurements were conducted up to 1,230 s, with a negligible increase seen. Therefore, the concentration of [U 6+ ] t at time t approximates to of [U 6+ ] 0 at time zero approximates to [U 5+ ] final.

For a = 1, and upon integration, a plot of Ln ([U 6+ ] t /[U 6+ ] 0) as a function of t should give a straight line with the rate-constant k as the slope.

Figure 5 shows the expected initial linearity, indicating that the reaction is indeed directly proportional to available U 6+ cations while, after prolonged reaction time, deviations are seen due to considerable consumption of the reactive sites. The extracted initial pseudo-first-order rate constant is found to be ca. 10 –2 s -1. It is worth noting that the rate constant has in it the amount of matter (so it is actually k = k’*W c / A c), where W c is the catalyst weight and A c is the investigated surface area by He(II) UPS. It also indicates that, at the conditions under investigation, the reaction of H atoms with U 5+ does not lead to further reduction.

These results might be relevant for understanding the redox properties of uranium oxides. For example, one may consider that, while the surface of UO 2 would eventually be oxidized to UO 3 (as seen by Donald et al. (2017)) in an oxidizing environment or to U 3 O 8 when water vapor is present (Senanayake et al., 2005), the formation of H atoms due to radiolitic reactions would further react and partly revert the process. Our present work is, however, qualitative. To extract more meaningful rate constants that may help provide a kinetic model, other parameters should be considered, particularly the amount of matter, the effect of pressure, and the surface area.

The reaction of hydrogen atoms with UO 3 thin film resulted in the reduction of U 6+ to U 5+ cations based on the 5f 1 of He(II) UPS line shape and on the U4f XPS binding energy position and satellites. Associated with the appearance of the U 5+ are surface hydroxyls (O3σ at approximately 10 eV (He(II) UPS)). Both saturate by approximately 150 L exposure at 190 K (10 –7 torr). The shape of the valence band shows distinct changes between pure UO 3 and that which is partially reduced, which was similar to that of U 2 O 5. The changes in the valence band shape are due to changes in the degree of hybridization between the O2p and U6d and U5f orbitals. The initial pseudo-first-order rate constant at 190 K was found to be approximately 0.01 s -1. The results show a fast reduction of U 6+ to U 5+ with no evidence of further reduction to U 4+ cations.