Oxidation of a depleted uranium‐5 wt% molybdenum (U‐5Mo) alloy in UHV by AES and XPS

Early stage oxidation of a dilute‐depleted uranium‐molybdenum alloy was analysed in situ under ultra‐high vacuum conditions by AES and XPS. At the equivalent of less than 300 ns at 1‐atm O2, U‐5Mo oxidizes to form stoichiometric UO2. No molybdenum oxidation is observed. After an oxygen dose of approximately 39 L, the oxide layer approached a limiting thickness of approximately 2.4 nm. The oxidation kinetics followed a logarithmic rate law, with the best fit to the experimental data for the oxide thickness, d, being given by d = 1.26 log(0.12t + 0.56). Changes in oxygen KLL and 1s peak positions associated with transformation from chemisorbed oxygen to metal oxide were observed at similar oxygen doses of 2.3 and 2.6 L O2 by AES and XPS, respectively, which opens up the possibility of using well‐characterized XPS chemical information to inform Auger peak shifts.

radiological perspective, it is also an attractive material for uranium research purposes. 7 Alloying is a common method for increasing the corrosion resistance of uranium, and range of elements has been utilized for this purpose, including Al, Mo, Nb, Ru, Si, Ti, V, Y, Zr, and V. [10][11][12][13] Niobium is generally considered to be the most effective and has received considerable attention in the literature at various alloying amounts (2.5-14 wt%). XPS analysis has shown that increasing the Nb concentration from 2 to 8 wt% leads to a thinner oxide layer being formed. 14 Scanning Kelvin probe force microscopy (SKPFM) and potentiodynamic polarization studies have shown that increasing the Nb concentration from 2.5 to 5.7 wt% improves corrosion performance. 13 In situ studies on U-Nb alloys at ultra-high vacuum (UHV) by XPS and AES found early stage oxidation to occur initially by the formation of UO 2 , followed by successively higher niobium oxides: NbO, NbO 2 , and Nb 2 O 5 . 15, 16 It has further been proposed that a critical amount of Nb 2 O 5 , specifically, is required to achieve enhanced corrosion protection based on providing a barrier to anionic diffusion. 10 Similarly, a U-20-at% Zr was shown by XPS to form ZrO 2 following 5 L* O 2 dose, albeit to a lesser extent than uranium oxidation. 17 Alloying with molybdenum has also been shown by gravimetric methods and X-ray diffraction (XRD) to improve corrosion performance compared with unalloyed uranium, however, not to the same level as for niobium. [18][19][20] A recent XPS investigation into the effects of various surface preparation steps on the oxidation of a U-10 wt% Mo alloy found a protective role for molybdenum in the prevention of formation of higher uranium oxides, although this was following extended periods of exposure to 97% humidity. 21 To date, mechanistic studies that attempt to determine whether the initial process of U-Mo alloy corrosion exhibits similar characteristics to U-Nb alloys and whether any differences may have an effect on the lower protective performance of U-Mo are lacking. In this work, we use oxidation at extremely low-oxygen partial pressures (pO 2 < 10 −8 mbar) to study the very early stages of oxidation of the candidate U-5Mo alloy by AES and XPS.
The sample was a 5 mm diameter disc of depleted uranium alloyed with molybdenum at 5 wt%. It was polished using increasingly fine grades of SiC abrasive paper down to P2500. The sample was rinsed and washed firstly with deionized water and then with isopropyl alcohol (IPA) prior to analyses.

| Auger electron spectroscopy
AES was performed using a Thermo Scientific MICROLAB 350 scanning Auger microscope operating at a base pressure of 2.6 × 10 −8 mbar (approximate pO 2 = 5.20 × 10 −9 mbar), employing a primary electron beam energy of 10 keV and a sample current of approximately 10 nA. Monoatomic argon-ion etching was performed immediately prior to analysis to remove the native oxide and expose the clean metal surface. Etching was performed with an EX05 argon-ion gun in a 2 × 2 mm raster pattern until the O KLL signal was no longer detectable. Auger spectra were collected over an energy range of 60 to 530 eV with a constant retard ratio of 4 and dwell time of 40 ms per 1.0 eV channel. The spectral scan parameters were determined following a period of method development to achieve a good compromise between speed of analysis, spectral information, and spectral resolution for the oxidation study. The time between the end of etching and start of the spectral scan was 8 seconds (0.03 L), and the analysis time of each iteration was 35 seconds (0.14 L). Auger spectra were collected in the direct mode and were taken up to an O 2 dose of approximately 40 L (1 L = 10 −6 Torr s) with 300 iterations.

| X-ray photoelectron spectroscopy
XPS was performed using a Thermo Scientific Theta Probe photoelectron spectrometer operating at a base pressure of 6.6 × 10 −9 mbar (approximate pO 2 = 1.32 × 10 −9 mbar) with a monochromated Al Kα source (1486.6 eV). Again, monoatomic argon-ion etching was performed immediately prior to analysis to remove the native oxide and expose the clean metal surface. Etching was performed with an EX05 argon-ion gun in a 2 × 2 mm raster pattern until the O 1s signal was no longer detectable. High-resolution spectra were collected over appropriate energy ranges to record U 4f, Mo 3d, O 1s, and C 1s peaks    Figure 1, oxidation proceeded rapidly in the initial stages, but then, slowing towards a limiting oxide thickness was observed, as described by the Cabrera-Mott theory for thin-film growth. 22 It was considered that the relatively unchanging O KLL intensity beyond 7 L O 2 could result from the oxide growing to a thickness beyond the escape depth of Auger electrons; however, the XPS investigation (see §3.2) showed that this was not the case. which was attributed to initial chemisorption of oxygen species followed by transformation to oxygen within a solid metal oxide and is in agreement with similar work. 16 For a similar range of oxygen exposures, Figure 3 shows an expansion of the 50 to 300 eV energy region containing the major uranium peaks. Since it was not possible to obtain a base vacuum of lower than 10 −9 mbar in the AES instrument, speed of acquisition was favoured over spectral resolution, and pseudo-survey scans were used to cover the range of energies of interest rather than performing separate high-resolution scans. As a result, it was difficult to accurately resolve each individual uranium transition; however, a number of possible peaks were identified, as shown in the inset table in Figure 3. Assignments were made with the aid of a comprehensive Auger peak library designed specifically for this work, in which the theoretical Auger electron kinetic energy (E calc ) of all possible transitions for every element up to uranium in the periodic table (calculated using archival experimental data [23][24][25][26][27] ) is available in a spreadsheet to be searched and referenced against observed peak positions (E obs ). All of the proposed uranium peaks were shown to shift to a lower kinetic energy with oxidation, indicating the potential for extracting chemical state information with Auger uranium studies. The main molybdenum peak (Mo MNN) at 190 eV was not resolved in the spectra because of the presence of the U N 6 O 4 O 5 peak with a similar kinetic energy. In order to provide additional chemical information and expand upon the AES results, the oxidation experiment was also carried out using XPS.   An expansion of the energy range 60 to 300 eV from the data in Figure 1 for a selection of O 2 doses. Six uranium peaks originating from O and N core shell electrons were observed. Dashed lines are used to highlight peak shifts as oxidation of the U-5Mo alloy surface occurred. The inset table provides initial peak energies from the bare metal surface (E obs ), the peak transition assignments, the calculated peak energies used to derive the assignments (E calc ), and peak shift values (ΔeV), which are all to the lower kinetic energy side of the peaks in the pure metal spectrum FIGURE 4 XPS U 4f, O 1s, and Mo 3d core level spectra showing progressive in situ UHV oxidation at a selection of O 2 doses. Dashed lines show peak position shifts, or lack of, as oxidation progressed. Peak shifting in the U 4f doublet was characterized by the formation of separate oxide peaks at the higher binding energy side of the metallic peaks for both 4f 7/2 and 4f 5/2 components. The emergence of a satellite peak (SP) structure at higher binding energy was also observed for the 4f 5/2 component as is commonly observed in uranium oxide spectra. It is likely that a satellite peak for the 4f 7/2 component was obscured by the 4f 5/2 metal peak. After an approximate O 2 dose of 2.3 L, the O 1s peak exhibited a slight peak shift of approximately 1.4 eV to higher binding energy as chemisorbed oxygen species transformed to metal oxide species. The Mo 3d doublet peak exhibited no shift and was shown to reduce in intensity as the uranium oxide formed on the metal surface approximate dose of 2.3 L O 2 , also in general agreement with AES data, and again attributed to initial chemisorption of oxygen species The lack of molybdenum oxidation at this stage is contrary to that of niobium when present as an alloying addition for uranium, which has been shown by XPS to oxidize after 10 L O 2 in UHV conditions. 10,15 This observation might be expected from a purely thermodynamic perspective. Table 1 lists Gibbs free-energy values for the formation of a selection of uranium, niobium, and molybdenum oxides. 28 There is a substantially stronger thermodynamic driving force for niobium oxidation compared with molybdenum, especially for the critical oxide Nb 2 O 5 .
As shown in Figure 4, the main U 4f oxide peaks occurred at 380.8 and 391.6 eV for the 4f 7/2 and 4f 5/2 peaks, respectively, and did not shift throughout the analysis. This suggests that the composition of the oxide was invariant during the initial stages of oxidation. A significant amount of data exist in the literature regarding the XPS peak positions and peak shifts for different uranium oxide stochiometries. Table 2 is a compilation of the available data and includes relevant values from this current work. Given the error values quoted in the various studies, which range from ±0.1 to ±0.3 eV, the most likely uranium oxide assignments, based on both the free energy of formation and binding energy considerations (Tables 1 and 2), are U 4 O 9 and U 3 O 7 . However, it is known that the presence of alloying elements can cause shifts in XPS peak positions and in a comprehensive review of XPS for determining uranium oxidation states by Ilton and Bagus, it was shown that the binding energies of the U4f 7/2 peak were unreliable indicators of the chemical state based on structural and compositional factors. 34,35 Uranium oxides are typically accompanied by characteristic shakeup satellite peaks in the XPS spectra. In the same Ilton and Bagus review, it is suggested that the separation between satellite and primary peaks (ΔE s-p ) is not so markedly influenced by oxide structure/composition and therefore a more reliable indicator of oxidation state. 34 Unfortunately, satellite peak positions for oxides other than UO 2 are not reported as commonly as the primary peak positions, and so, the number of pertinent citations is significantly less (Table 3).
A number of groups assign the presence of a satellite peak at approximately 6.9 eV to the higher binding energy side of the 4f 5/2 primary oxide peak to stoichiometric UO 2 . 10,15,30,[42][43][44] Allen et al describe the appearance of additional satellite structure at ΔE s-p = 8.2 eV when UO 2 is oxidized to superstoichiometric UO 2+x . As shown in Figure 4, the only clear satellite peak observed for the UHV-grown oxide in this work was to the higher binding energy side of the 4f 5/2 oxide peak and has a ΔE s-p value of approximately 7.0 eV. This is indicative of  stoichiometric UO 2 and is in conflict with binding energy information for the 4f 7/2 and 4f 5/2 oxide peaks.
As a continuation of the XPS experiment, the sample was removed from the instrument, air exposed for a period of 5 minutes in ambient conditions, and then reanalysed (for spectra, see online Supporting with ΔE s-p = 6.9 eV for 4f 7/2 and 391.0 eV with ΔE s-p = 6.8 eV for 4f 5/2 . By consulting Tables 2 and 3, it is clear that the air-formed oxide was stoichiometric UO 2 . Considering these peak positions relative to the main O 1s peak, it is interesting to note that the separation is exactly the same as that for the UHV-grown oxide relative to the corresponding main O 1s peak for both 4f 7/2 (Δ150.2 eV) and 4f 5/2 (Δ139.4 eV). This observation supports the conclusion that the UHV-grown oxide was also stoichiometric UO 2 and is in agreement with information in the literature concerning the diagnostic use of satellite peak positions instead of absolute core-level binding energies for determining uranium oxidation states.

| Oxidation kinetics
Oxide thickness information was estimated from the XPS data, using the U 4f 7/2 oxide-to-metal peak ratio (I o /I m ) obtained by peak fitting and according to the relationship described in Equation (3), which assumes a uniform oxide layer, and was first described by Strohmeier,45 where d is oxide thickness (in nm); λ o and λ m are the inelastic mean free paths (IMFPs) for the oxide and metal, respectively (in nm); N o and N m are the volume densities of metal atoms in the oxide and metal, respectively; and sin θ is the electron take-off angle relative to the sample surface. 45  Oxide thickness was plotted as a function of time and showed a rapid initial rate of oxidation, followed by a slower growth rate approaching a limiting thickness of approximately 2.4 nm ( Figure 5).
This was in agreement with the observations for O KLL Auger peak intensity behaviour described in Section 3.1. On the basis of the best fit to the experimental data in Figure 5, the oxidation kinetics can be accurately described according by a logarithmic rate law (Equation 4), where the oxide thickness, d, is given by where k is the rate constant and a and b are separate, experimentally determined, constants. The best fit was obtained for the function d = 1.26 log(0.12t + 0.56).
Previous studies on uranium and uranium alloy oxidation describe parabolic or linear behaviour; however, these tend to be based on gravimetric methods performed at elevated temperatures and atmospheric pressure, which are therefore concerned with much later stages of oxidation. 3,47 As a result, the oxide thickness is likely to be such that ionic diffusion through the oxide layer is the rate limiting step and parabolic kinetics are expected. 48 It is generally accepted that at ambient temperatures and for thin films, as demonstrated in this current work, the rate limiting step becomes ionization of oxygen atoms to oxygen ions and hence electron transport across the oxide layer, leading to logarithmic behaviour. 48,49 The resulting oxide reaches a limiting thickness of 2 to 3 nm, which is often referred to as the "Mott thickness". Although the strict interpretation, according to the Mott-Cabrera theory of oxide growth, is associated with the instantaneous growth of an oxide layer as a result of the intensive field, that exists because of the contact potential difference between   37 7.0 37 6.9 34 6.9 34 6.8 38 No data 5.8 39 8.2 39 No data 6.9 29 No data U 3 O 8 4.0 39 8.0 39 10.0 39 No data 7.8 29 No data 8.2 40 No data γ-UO 3 3.7 39 10.6 39 No data No data 3. • Assignments were made for six uranium Auger peaks originating from N and O core electrons. All of these peaks exhibited a shift with oxidation of the uranium metal, which indicates that such shifts could be used for AES chemical state mapping studies in combination with AES/XPS spectroscopic data.
• No peak shift was observed for the Mo 3d peak, indicating that molybdenum does not oxidize during the initial oxidation stages of DU-5Mo. This lack of early oxidation is contrary to observations made for uranium-niobium alloys and indicates a lower level of corrosion protection when uranium is alloyed with molybdenum compared with niobium.
• The U 4f doublet peak exhibited a clear peak shift associated with uranium oxidation. The metallic uranium peaks remained observable throughout the experiment (equivalent to 270 ns at 1 atm), highlighting the approaching to a limiting thickness of the oxide in UHV at 40 L exposure. Stoichiometric UO 2 was the only uranium oxide formed upon oxidation, as determined primarily by the relative position of the U 4f 5/2 main oxide peak and corresponding satellite peak (ΔE s-p = 7.0).
• In UHV conditions at ambient temperature, the U-5Mo alloy obeys a logarithmic rate law kinetics, with the best fit to the experimental data for the oxide thickness, d, being given by d = 1.26 log(0.12t + 0.56). This oxidation rate behaviour is typical of oxide layer thicknesses in the order of that observed here (approximately 2.4 nm).