Interaction of Molecular Hydrogen on α‑Pu at 300 K: Nucleation of Hydrides and Pu-Catalyzed Carbides

Abstract

An α-Pu sample was studied in its “as-received” and “sputtered” state by both X-ray photoelectron spectroscopy (XPS) and time-of-flight secondary ion mass spectroscopy (ToF-SIMS). Analysis of the “as-received” surface indicated significant carbon and oxygen, with oxygen signal pertaining to plutonium dioxide (PuO₂) and sesquioxide (Pu₂O₃). The “sputtered” clean metal surface was then dosed with hydrogen (H₂) gas at 300 K and found to nucleate hydride species (PuH _x ). Interestingly, H₂ was observed to be critical for nucleation of carbides (Pu _x C _y ) by acting as a catalyst for Pu’s reaction with background gases. Because Pu metal has a high affinity for oxygen, reoxidation occurred under UHV, forming Pu _x O _y and disrupting both PuH _x and Pu _x C _y growth. Observations from the experiments were modeled for the α-Pu(020) surface, illustrating the ability of H₂ to readily dissociate onto sputtered and speciated surfaces at 300 K. The projected density of states from these calculations were compared to experiments, showing good agreement between the Pu 5f, 6d, and 7s states and the H 1s, O 2p, and C 2p valence states. Overall, the above results indicated that the formed hydride layer is unstable at 300 K in UHV conditions.

1. Introduction

The actinide (An) elements are unusual because the 5f electronic behavior is dynamic, varying between localized or itinerant. − This leads to complex energetics when studying bond contribution in multielement systems within the An family, which results in materials with unusual, but interesting structural, electronic, and magnetic properties. Moreover, the relevancy of plutonium (Pu) as the critical point, in which the transition between itinerant and localized electrons occurs, as well as its importance in nuclear materials, fuels, and energy, necessitate fundamental understanding of itself and its corresponding compounds. This is specifically essential regarding safe handling of Pu and conditions of its storage environments.

Some properties of Pu include its existence in six allotropes at ambient pressure, where the monoclinic alpha (α) phase is stable at 300 K. However, because α-Pu is hard and brittle, it is typical to alloy with gallium (Ga), aluminum (Al), or cerium (Ce) to stabilize the face-centered cubic (fcc) delta (δ) phase. As a result, δ-Pu (Pu–Ga) alloys have been the subject of several experimental − and theoretical studies, − and used to formulate conditions and procedures for Pu processing and storage. Tangentially, δ-Pu aging has raised further questions on the formation of its products through gas–solid reactions, such as the chemical stability of Pu oxides (Pu _x O _y ), ,− ,,, hydrides (PuH _x ), ,, and oxycarbides (Pu _x O _y C _z ). ,−

Pu metal is highly reactive, and like most other metals, exposure to oxygen (O₂) and water (H₂O) at room temperature (300 K) will result in the passive oxidation of the surface, where the layering of oxide scales is in the following sequence: PuO₂ – Pu₂O₃ – Pu_2–x O_3–y – Pu. The major concern with passive oxidation is that plutonium dioxide (PuO₂) will form, spall, and then disperse from the Pu surface as particulates. In addition, the Pu metal will readily expand from oxygen absorption and create hydrogen (H₂) gas when in the presence of H₂O vapor. For example, the experiments by Haschke et al., , demonstrated the catalytic nature of Pu and PuO₂ and their abilities to drive corrosion in the presence of O₂ and H₂O vapor under the following schemes

$$\mathrm{Pu}(\mathrm{s})+{\mathrm{O}}_2(\mathrm{g})\to \mathrm{Pu}{\mathrm{O}}_2(\mathrm{s})$$ 1

$$\mathrm{Pu}{\mathrm{O}}_2(\mathrm{s})+\frac{x}{2}{\mathrm{O}}_2(\mathrm{g})\to \mathrm{Pu}{\mathrm{O}}_{2+x}(\mathrm{s})$$ 2

$$\mathrm{Pu}(\mathrm{s})+2{\mathrm{H}}_2\mathrm{O}(\mathrm{g})\to \mathrm{Pu}{\mathrm{O}}_2(\mathrm{s})+{\mathrm{H}}_2(\mathrm{g})$$ 3

$$\mathrm{Pu}{\mathrm{O}}_2(\mathrm{s})+x{\mathrm{H}}_2\mathrm{O}(\mathrm{g})\to \mathrm{Pu}{\mathrm{O}}_{2+x}(\mathrm{s})+x{\mathrm{H}}_2(\mathrm{g})$$ 4

Because Pu corrosion is unavoidable, numerous experimental − ,, and theoretical − studies have explored the reactivity of α-Pu and δ-Pu surfaces and their chemical products when exposed to different gases. Even more interesting, the analysis of δ-Pu in ultrahigh vacuum (UHV) has revealed the unique challenges in studying Pu metal surfaces, such as the preference in forming Pu sesquioxide (Pu₂O₃) over PuO₂ by autoreduction, , and the observation of Pu metal oxidizing to Pu _x O _y despite sitting under UHV. ,, Many of these phenomena are postulated to be controlled by oxygen solubility and diffusion, either laterally from adatoms coming from the oxidized surface onto the metal surface or from subsurface diffusion through the Pu lattice by oxygen anions (O^2–). For this reason, corrosion experiments that test different aspects of Pu reactivity are extremely valuable for validation of theoretical models.

In this paper, we approach Pu corrosion through surface science measurements and have performed all our experiments in UHV with α-Pu samples, allowing studies of surface interactions with ultrahigh purity (UHP) gas to not be affected by alloyed interstitials (e.g., Ga in δ-Pu). The following experiments were chosen for a few reasons; (1) H₂ dosing of metal Pu is relevant to nucleation of PuH _x (x = 1, 2, 4, etc.) species, ,,− ,,, (2) Removal of adsorbed oxygen (O_ads) from Pu is difficult because of its high propensity to dissolve in the lattice, meaning interstitial O persists even after noble gas ion sputtering. Therefore, H₂ dosing experiments exemplify how residual oxygen impacts gas dosing experiments and the overall corrosion mechanism, and (3) The interaction between α-Pu and H₂ is the simplest prototypical reaction from a modeling standpoint. To illustrate, consider these proposed reactions

$$\mathrm{Pu}(\mathrm{s})+{x\mathrm{H}}_2(\mathrm{g})\to \mathrm{Pu}{\mathrm{H}}_x(\mathrm{s})$$ 5

$${\mathrm{Pu}}_2{\mathrm{O}}_3(\mathrm{s})+{3\mathrm{H}}_2(\mathrm{g})\to 2\mathrm{Pu}(\mathrm{s})+{3\mathrm{H}}_2\mathrm{O}(\mathrm{g})$$ 6

Below we performed two types of surface analysis, enabling the α-Pu surface to be examined at two probing depths. First, X-ray photoelectron spectroscopy (XPS) was used to gather information at ∼10 nm depth, providing detection of core electrons and to those within the vicinity of the Fermi level (E _F). ,,,,,,, Second, time-of-flight secondary ion mass spectrometry (ToF-SIMS), was carried out to extract information on the topmost 1–2 monolayers (<1 nm) of the sample. ToF-SIMS on δ-Pu has been investigated before (e.g., Venhaus et al.) and in electrodeposited Pu, , proving the method’s sensitivity and ability to detect PuH _x species. The gas dosing experiments described below used polished α-Pu samples, which were sputtered prior with argon ions (Ar⁺) until only residual oxygen remained. To validate our experimental findings, density functional theory (DFT) calculations were completed on various surfaces of α-Pu (crystalline, sputtered, speciated), describing the molecular dynamics with H₂. Comparisons were then made with the experiment to provide insight on the mechanism of H₂ with α-Pu at 300 K.

2. Experimental Section

2.1. Preparation of α-Pu for UHV Analysis

The experiments used an α-Pu sample (5 × 10 mm²) that was mechanically polished to a 1200 grit finish using De-Solv-it fluid in an argon glovebox, and a final ethanol rinse was applied at the end of polishing. For the XPS tilting experiments, in order to determine the presence of any potential bulk impurities (i.e., oxygen and carbon), the sample was repolished in a nitrogen glovebox with decreasing grit from 9 μm, 3 μm, and 1 μm diamond suspension.

2.2. XPS Experimental Details

The α-Pu sample was unloaded from a transfer device and placed into an XPS instrument’s (Versaprobe III, Physical Electronics, Chanhassen, MN) preparation chamber, which was held at base pressure (P _base) of ∼7.5 × 10^–10 Torr. A monochromatic Al K_α light source (1486.7 eV) was used to analyze the α-Pu surface, and properties of the X-ray’s beam included a spot size of ∼100 μm diameter and power of 25 W. Calibration of the system was conducted with silver (Ag) at the 3d_5/2 position to ensure that it was located at 368.2 eV. Before collecting spectra, the sample was aligned in the Z-direction. Further work confirmed that the gold (Au) 4f and 4d peak positions were at their expected values to confirm that a large range of kinetic energies were well calibrated. Source intensity was confirmed by comparison of Ag to the installation standards for full width at half-maximum (FWHM) vs intensity. To monitor the sample before and after experiments, a digital camera (Canon, EOS R8, Tokyo, Japan) was used to photograph the sample through a viewport, whose image is displayed in Figure S1. Note how the α-Pu sample was oxidized, seen by its blackened surface, and several sputter craters were noticeable from previous experimentation on the sample.

After loading into the preparation chamber, the sample was moved to the main chamber, which was held at P _base = 3.75 × 10^–10 Torr. The sample was pinned, aligned, and the stage configured to a tilt angle of 45° (relative to the hemispherical analyzer). Parameters used to collect both high-resolution survey and region scans are displayed in Table S1. As indicated below, the regions of interest were the following: Pu 4f, O 1s, C 1s, and the valence band. In addition, tilting experiments were performed by changing the eucentric tilt angle of the stage to 60°, 75°, and 90°, i.e., angle of emission at 30°, 15°, and 0° to become more depth sensitive to the α-Pu sample.

Once the “as-received” surface was surveyed, the α-Pu sample was sputter cleaned using an UHP Ar⁺ sputter gun to remove any adsorbed contaminants from previous atmospheric exposure. The sputter gun was operated in unrastered mode at ∼7.5 × 10^–9 Torr, and conditions included accelerating voltage = 4 kV and current = 4 mA. Sputtering was done in 15 min intervals, scanning the Pu 4f and O 1s regions periodically until reaching metal. Here the α-Pu surface was sufficiently cleaned (4f was metal, but residual oxygen was in the O 1s), followed by UHP H₂ being administered via a manifold connected to the XPS instrument’s analysis chamber. The partial pressure of H₂ (P _H2) was ∼3.59 × 10^–7 Torr. To check P _H2 and P _base compositions during experiments, a residual gas analyzer (RGA) attached to the XPS instrument was used to collect mass spectra. Both are plotted in Figure S2, and the spectra illustrates that P _base consisted of trace amounts of H₂, H₂O, Ar, CO, and CO₂. During the H₂ dose, the highest signal, as expected, was H₂ (2 amu). Finally, all measured spectra were analyzed using the CasaXPS software.

2.3. ToF-SIMS Experimental Details

ToF-SIMS is a highly surface-specific analytical technique that provides measurement of 1–2 monolayers on the surface (<1 nm), whereas the probe depth of XPS is ∼5–10 nm. Due to the sensitivity of ToF-SIMS, it can determine impurities at parts-per-million (ppm) levels and all chemical speciation of the surface, including hydrogen and its compounds. The instrument used in this study was a Kore Technology ToF-SIMS (Ely, Cambridgeshire, U.K.) which was equipped with a 25 kV Ga⁺ liquid metal ion gun (LMIG) as the primary ion source at a current of 4 nA (steady state) at an incident angle of 45° to the normal. The beam was pulsed at a repetition rate of 3.3 kHz and a pulse width of 10 ns, providing a total flux of 1.3 × 10⁹ ions/cm²/s. The ion beam was rastered over an area of 500 × 500 μm². The ToF spectrometer entrance lens was operated at a mass resolution of approximately 1000 (Δm/m), and the analysis time was 20 s, yielding a total fluence of ∼3 × 10¹⁰ ions/cm², which provided minimum effect on the surface chemistry during the analysis time, i.e., “static SIMS.” A 4 kV Ar⁺ sputter gun was operated in a 3.3 kHz interlaced mode, and an 85 ms pulse width (duty cycle of 28%) was used for surface cleaning before exposure to UHP H₂ gas. The raster area for the sputter gun was 1000 × 1000 μm². Gas exposures of H₂ were performed at ∼1 × 10^–7 Torr at room temperature (300 K) and low temperature was achieved by cooling the sample stage with liquid nitrogen.

2.4. Density Functional Theory Calculations

Experimental observations from XPS, ToF-SIMS, and other surface sensitive methods have indicated that α-Pu metal surface, even when ion beam sputtered, does not remain free of chemical impurities and over time builds up regions of oxygen. Thus, to make the simulations of H₂ interaction with α-Pu surface realistic in comparison to experimental observation, oxygen impurities were first placed on the α-Pu surface to form an extremely thin “oxide” layer on the order of ∼2.5 Å in thickness. This resulting surface was then exposed to one monolayer (1 ML) of H₂ gas. Figure shows the structural representation of the α-Pu(020) surface with O impurities exposed to H₂ gas. The entire system was optimized into a minimum energy state followed by room temperature ab initio molecular dynamics using DFT. ,

1.

Structural model of an α-Pu surface with a thin “oxide” layer exposed to 1 ML of H₂ gas. The Pu, O, and H atoms are colored gray, red, and blue, respectively.

The DFT calculations were performed using the SIESTA electronic structure code. − In SIESTA the Kohn–Sham eigenstate is represented as a linear combination of strictly localized pseudoatomic orbitals, together with norm-conserving pseudopotentials for the description of ion-electron interactions. The Pu atoms were described using a recently developed optimal strictly localized basis functions, while double-ξ basis with polarization orbitals (DZP) basis functions were employed to describe the O and H atoms. The Perdew–Burke–Ernzerhof (PBE) formulation of the generalized gradient approximation (GGA) to the DFT exchange-correlation functional was also used. The Pu ion-electron interactions were described using norm-conserving Hamann pseudopotentials, while norm-conserving Troullier-Martins pseudopotentials were used to describe the ion-electron interactions in H and O.

The α-Pu­(020) surface was modeled by a periodic slab with 8 atomic layers and 16 atoms per layer (total of 128 Pu atoms). The choice of a (020) surface is based on X-ray diffraction studies on α-Pu, where the (211) and (020) planes are exceptionally strong. Although, the (020) surface is the more stable and the only one that is atomically flat. To start, the bulk lattice vectors from which the surface lattice vectors were constructed were taken from in silico experiments by Atta-Fynn. A large vacuum space of 75 Å was added to the slab along the surface direction to isolate the slab’s faces, and an antiferromagnetic spin configuration, which consists of alternating layers of down-spin and up-spin on the Pu atoms along the surface direction, was employed. A 2 × 2 × 1 k-point mesh generated from the Monkhorst–Pack method was used for the Brillouin zone integration. The first-order Methfessel-Paxton Fermi surface smearing function with a width of k _B T _electron, (T _electron = 300 K) was applied to accelerate the convergence of the electronic self-consistent iterations. The following convergence criteria was then used for the total energy and stress minimizations: the maximum magnitude of the atomic force along each coordinate direction was <0.01 eV/Å, and the maximum magnitude of the stress tensor element <0.01 GPa. Constant temperature ab initio molecular dynamics (AIMD) simulations were performed using a time step of 1 fs, with the temperature controlled with the Nose-Hoover thermostat chains.

The simulation protocol is as follows:

• (i).First, O₂ molecules were placed on a clean α-Pu(020) metal surface, and the total energy and stress are minimized; this resulted in dissociative O chemisorption on the surface to form the thin “oxide” layer. After minimization, the thin oxide layer consisted of 40% Pu₂O₃ units and 60% PuO₂ units, with the average Pu–O distance being 2.3 Å.
• (ii).Second, 1 ML of H₂ gas (16 H₂ molecules) was placed on the surface in step (i) and the total energy and stress tensor of the entire system is minimized (e.g., Figure ).
• (iii).AIMD simulation was performed for 3 ps at 300 K on the system in step (ii).

Additional AIMD simulations at 300 K were performed for the following systems to study the influence of sputtered surface defects and interstitial O and C atoms (the O and C concentrations are equal) on the XPS spectra

• (a)H₂ on pristine α-Pu(020) surface
• (b)H₂ on α-Pu(020) surface with defects mimicking sputtering
• (c)H₂ on α-Pu(020) with defects mimicking sputtering and O atom interstitials
• (d)H₂ on α-Pu(020) with defects mimicking sputtering and C atom interstitials

In Figure , the representative structural configurations for the simulations in (a)-(d) are depicted. More details for each of the simulations are described further below after first describing experimentation on the α-Pu sample by XPS and ToF-SIMS.

2.

Structural models of α-Pu(020): (a) H₂ on pristine surface; (b) H₂ on defective surface mimicking sputtering; (c) H₂ on surface with defects and O interstitials; (d) H₂ on surface with defects and C interstitials. The Pu, O, H and C atoms are colored gray, red, blue, and pink, respectively.

3. Results

3.1. XPS of Polished α-Pu: As-Received vs Sputtered Surface

Experiments on the α-Pu sample began by surveying the as-received surface and scanning from 1300 to 0 eV. This spectrum is shown as a black line in Figure . As reported elsewhere, ,,,, major chemical features include core electron spectra of Pu, O, and C, along with X-ray induced auger transitions for O KLL and Pu NOV, NVV, and N₇P₃V. For the core orbitals, the highest intensity peaks are Pu 4d and Pu 4f. Other smaller peaks include Pu 4p_3/2, Pu 5p_1/2, Pu 5p_3/2, and Pu 5d. Lastly, Pu 6p_3/2 and 6p_1/2 are in the valence region just before Pu 5f and the Fermi level.

3.

Survey scans of α-Pu before and after 30 min of Ar⁺ sputtering. The as-received surface is in black, and various Pu, C, and O peaks are indicated. After sputtering (red line), the α-Pu reveals a metal surface.

Because XPS probes the surface at ∼10 nm depth, Ar⁺ sputtering is helpful for removing contaminant elements and exposing the Pu metal. In Figure , the “clean” α-Pu’s survey spectrum is depicted as a red line after 30 min of unrastered Ar⁺ sputtering at 4 kV and 5 mA. A significant increase in the intensities of Pu features is noted when compared to the as-received surface. In contrast, a decrease in signal is observed for the C 1s, O 1s, and O KLL peaks. Tabulation of α-Pu peaks and their positions after sputtering is shown in Table and compared to other XPS results in the literature. As noted, the measurements agree with those previously reported. ,,,,

1. Peak Positions of Different Pu Features in α-Pu by XPS .

Pu (this work)	position (eV) (this work)	refs	position (eV)
4p3/2	1123.6	Baptist	α-Pu, 1124
		Roussel	α-Pu, 1123.3
4d3/2	846.8	Baptist	α-Pu, 844
4d5/2	798.0	Baptist	α-Pu, 799
		Roussel	α-Pu, 797.4
4f5/2	435.1	Larson	α-Pu, 434.9
		Baptist	α-Pu, 435.1
		Roussel	α-Pu, 435.1
4f7/2	422.1	Larson	α-Pu, 422.2
		Baptist	α-Pu, 422.1
		Roussel	α-Pu, 422.2
5p1/2	284.8	Baptist	α-Pu, 280
5p3/2	210.8	Baptist	α-Pu, 212
5d5/2	101.6	Baptist	α-Pu, 100
		Roussel	α-Pu, 101.3
6p1/2	28.2	Roussel	δ-Pu, 28.9
		Courteix	Pu2O3, 31.0
6p3/2	17.7	Roussel	α-Pu, 17.7
		Roussel	δ-Pu, 17.8

^aComparison to literature is provided.

After surveying, high resolution region scans were gathered for the as-received and sputtered α-Pu surface. These included: Pu 4f, O 1s, C 1s, and the valence region. As illustrated in Figure A, the spin–orbit coupled Pu 4f peaks (5/2 and 7/2) are plotted for the two surfaces to provide comparison. In the top half, the as-received surface had a PuO₂ passivating layer, which occurred when the α-Pu sample was exposed to atmosphere. Note that the sample was not held in UHV long enough (∼1 h) to observe complete autoreduction. ,, The 4f spectrum was consistent to previous XPS measurements. Here positions of 4f_5/2 and 4f_7/2 peaks for PuO₂ were approximately equivalent to Garcia-Flores et al. at 438.9 and 426.4 eV. Second, the 4+ “shake-up” satellites at 445.7 and 433.2 eV were also present. Third, partial autoreduction is observed at 424.4 and 437.2 eV in the Pu 4f, where the formation of Pu₂O₃ (3+) is beginning to add shoulders to the primary spectral peaks. In the bottom half of Figure A, sputtering reduced the intensities of the Pu 4f and shakeup peaks associated with PuO₂/Pu₂O₃. As this occurred, the metal (0) peaks for Pu 4f_5/2 (435.1 eV) and 4f_7/2 (422.1 eV) appeared and increased in intensity. In addition, the satellites at 437.8 and 424.9 eV were noted. In Figure B, the top half plots the valence band from 35 to 0 eV. Analogous to Figure A, the as-received surface had lower signal-to-noise than the sputtered one, however, both the 6p_1/2 and 6p_3/2 peaks were visible at positions of 30.4 and 17.8 eV. For the 5f peak, its position was ∼2.9 eV, similar to Roussel et al., suggesting features are shifted to higher binding energies for the oxidized surface. Interestingly, there was also a small peak at ∼10 eV. This feature would correspond to O 2p′, which has been observed before by PES. In Figure B’s lower half, signal was gained after sputtering. As expected, the 6p peaks shifted to slightly lower binding energy and the 5f peak moved toward the Fermi level.

4.

Region scans of α-Pu sample before and after sputtering. (A) Pu 4f region showing a PuO₂/Pu₂O₃ vs Pu metal surface. (B) Pu 6p and valence. Increase in signal and shift to lower binding energies is indicative of metallization.

Collection of both the O 1s and C 1s peaks were carried out due to their high intensities in Figure . Both are plotted in Figure to view differences between the as-received and sputtered surfaces (different y-scales for each). In Figure A, the as-received surface is plotted in black and depicts the O 1s peak. The peak is composed of two major components with the first one at ∼530.1 eV, indicative of a metal–oxygen (M _x O _y ) bond, i.e., Pu _x O _y . The second is apparent from 531–533 eV. This region corresponds to oxygen–carbon bond pairing of both σ (C–O) and π (CO) states and oxygen–hydrogen bonding in the form of hydroxyl groups (−OH). In Figure A’s lower half, the sputtered surface is shown, and here the O 1s peak is at ∼530.4 eV, which is close to values reported by Flores et al. Here the O 1s peak was not fully removed, despite 30 min of sputtering. As described in the Supporting Information (SI), the atomic percentage (atom %) of Pu and O, from analyzing Figures – indicated 88.7% and 11.3%, respectively. The relative sensitivity factors (RSFs) used for Pu 4f and O 1s were 20.42 and 0.711, respectively. When completely removed, it is postulated that the oxygen is transported back onto the metal by either lateral diffusion across the surface, or from below by diffusing upward toward the lower concentration gradient at the surface. This occurs in the time frame of minutes to hours for δ-Pu samples. , α-Pu also forms Pu _x O _y , but the time scales are longer, and Figure S3 shows various spectra (Pu 4f, valence, O 1s, and C 1s) taken after sputtering and leaving the α-Pu sample in UHV (9 × 10^–10 Torr) for periods of 20 and 40 min. This displays the expected change in the surface under identical timing to our H₂ gas experiments described below. Note that there were no changes to either the Pu 4f, valence, or C 1s regions. However, the O 1s signal did slowly increase with time. Analysis of Figure S3 for Pu 4f and O 1s, using RSFs, indicated an initial concentration of 95.8 atom % for Pu and 4.2 atom % for O. As the sample sat in UHV for 20 and 40 min, the increase in O was 5.1% and 5.4%. From these values, thickness can be estimated for the oxide layer if homogeneity is assumed. To do so, the modified Hill Equation (described in the SI) is used where the effective attenuation length (EAL) coefficient through a thin Pu oxide layer was ∼1.35. The EAL value was determined by correcting the inelastic mean free path determined from the Tanuma-Powell-Penn (TPP-2M) equation. Under these conditions, the oxide thickness is ∼0.08 Å and ∼0.16 Å. This represents a fractional coverage of ∼0.07 ML if the thickness of the oxide layer is ∼2.5 Å (Figure ), i.e., the surface is still mostly metal when sitting in UHV. Comparison to Figure A, where the O atom % was ∼2.7 times greater, it is not expected to majorly enhance the UHV growth rate. In addition, the removal of carbon and hydrogen from the O 1s peak signifies that any adventitious carbon that had deposited from atmospheric exposure was no longer present. This is confirmed in Figure B, where the top half shows the C 1s peak (black) of the as-received α-Pu sample. The lower portion of the frame shows the effect of sputtering. When the carbon was removed, there was an increase in the Pu 5p_1/2 peak, , which spans from 300–270 eV.

5.

Region scans of α-Pu sample before and after sputtering. (A) O 1s shows both an oxidized and metal Pu surface. (B) C 1s shows the removal of adventitious carbon to reveal the underlying Pu 5p_1/2 peak.

Lastly, tilting experiments were performed to test whether the reoxidation of the α-Pu surface originates from oxygen interstitials trapped deeply in the bulk or due to the effects of near-surface bound atoms. Figure shows area-normalized Pu 4f spectra taken at eucentric angles of 45°, 60°, 75°, and 90° (emission angles of 45°, 30°, 15°, and 0°). All of these spectra were taken after sputtering the surface until the O 1s and C 1s peaks were well below the XPS instrument’s detection limit (Figure S4), and the purpose of this experiment was to see how the Pu 4f peaks changed with additional sensitivity to the bulk. As shown, at 45° the spectrum was similar to Figure A and there was minimal change when increasing to 60°. When the sample was tilted to 75° and 90°, there was a change in the ratios of the metal and satellite features for both the 4f_5/2 and 4f_7/2 peaks. The metal components increased while the satellites decreased, which suggests the satellites must be driven by surface phenomena. Note there was no observation of a 3+ (Pu₂O₃) peak located within the Pu bulk confirming that the dissolved oxygen is limited to the selvage region just below the surface. Another interpretation of the result is that the metal satellite is related to Pu 3+ species trapped inside the selvage. However, without experiments on single crystals of Pu, it is hard to verify since sputtered-induced defect sites may also contribute to the satellite’s intensity. In addition, the O 1s and C 1s intensities were kept near the XPS’s detection limit (Figure S4), which proves that a stable 3+ species was not in appreciable amounts during the tilting experiments.

6.

Pu 4f_5/2 and 4f_7/2 spectra as a function the eucentric tilt angle. As angle is increased, more sensitivity to the bulk is gained. Here an increase in metal and decrease in the satellites is observed.

3.2. Analysis of XPS When Dosing Sputtered α-Pu Surfaces with H₂ Gas

After the α-Pu sample had been sputtered to metal, the surface was exposed to H₂ gas at room temperature (∼300 K). The experiment involved delivering H₂ at ∼3.6 × 10^–7 Torr for 20 min, i.e., ∼432 Langmuir (L). After the first dose, the H₂ gas was shut off, and high-resolution scans of Pu 4f, O 1s, C 1s, and the valence region were collected. Following that, a second H₂ dose was performed increasing the total exposure to ∼864 L. Changes from H₂ are presented in each of the regions by comparison at 0 L (red), 432 L (blue), and 864 L (green) for the α-Pu surface.

Consider Figure that plots the Pu 6p_1/2, 6p_3/2, and 5f peaks for the metal and H₂ dosed surfaces. Each spectrum was area-normalized and overlaid with each other to illustrate the effects of H₂. From 35–10 eV, there were no differences between the Pu 6p_1/2 and 6p_3/2 peaks, indicating no interaction with H₂. However, at ∼0.3 eV, a small reduction in the 5f’s proximity was noted (Figure , inset). At 432 L the peak’s intensity decreased by ∼2.8%, and then an additional 1.7% when exposure was increased to 864 L. While this change is small, ultraviolet photoelectron spectroscopy experiments aimed at studying valence bonding between Pu and H₂ have found that adsorbed hydrogen does decrease the intensity in this region. Unfortunately, the signal-to-noise ratio and resolution is significantly lower with the XPS method, but the decrease in signal is still indicative of hydrogen interaction with Pu.

7.

Pu 6p_1/2, 6p_3/2, and 5f peaks of α-Pu after H₂ doses of 432 and 864 L. A decrease in the 5f intensity is observed just before the Fermi level.

H₂ dosing was also studied by monitoring the response of core electrons in the Pu 4f, O 1s, and C 1s spectra. The Pu 4f peaks provide an indirect probing of H₂, and Figure shows the raw data (open black circles) and the fit function (red line) going through the Pu 4f region. Based on Roussel et al., fitting parameters were applied to both Pu 4f_5/2 and Pu 4f_7/2. Line-shapes and extracted fit parameters are shown in Table . The two major components in the Pu 4f spectrum consisted of 7/2 and 5/2 metal peaks, which are colored black and green in Figure A. In the 4f_7/2, the metal peak converged at 422.0 eV. For the 4f_5/2, its position was 435.0 eV, both close to previously reported values. Two metal satellite peaks (orange and blue) were also observed at higher binding energy in the 4f_5/2 and 4f_7/2 peaks. ,, Their positions were 437.3 and 424.7 eV, respectively. Lastly, from 426–434 eV, there is poor fitting between the fit functions and the experiment. It is not understood why the experiment has higher intensity in this region, but it may be related to X-ray scattering effects, contributions of Auger peaks at higher binding energies, or from O and C interstitials, thereby raising the background. Yet, as demonstrated by Roussel, , this region is typically left unfitted. In addition, an inset is provided in Figure A that depicts the change in the 4f region after dosing with H₂ gas at 432 L (red) and 864 L (blue). The enlarged plot is shown in Figure S5.

8.

Analysis of Pu 4f on α-Pu. PuH _x and Pu _x C _y species are seen after H₂ doses. (A) Clean metal. Inset – Spectral evolution from H₂ exposure (Figure S5). (B) 864 L H₂ exposure with fits (Table S2). The gray waveform represents the difference between frames (B) and (A), describing peak positions related to PuH _x and Pu _x C _y . (D) Atomic percentages of the Pu, O, H, and C species are plotted with respect to H₂ exposure.

2. Fitting Parameters of Pu 4f Peaks from Figure A .

Pu 4f	lineshape	position (eV)	FWHM (eV)	area (counts)
7/2 – this work	LF(0.42,1.25,110,195)	422.0	0.99	308,085
7/2 – Roussel	LF(0.42,1.25, 110, 195)	422.0	0.98	-
5/2 – this work	LF(0.42,1.25,110,195)	435.0	1.27	248,592
5/2 – Roussel	LF(0.42,1.25,110,195)	435.1	1.14	-
Sat. 7/2 – this work	LF(0.43,0.75,50,233)	424.7	2.8	108,137
Sat. 7/2 – Roussel	LF(0.43,0.75,50,233)	424.9	2.1	-
Sat. 5/2 – this work	LF(0.43,0.75,50,233)	437.3	3.1	56,430
Sat. 5/2 – Roussel	LF(0.43,0.75,50,233)	437.8	2.5	-
BG – this work	Shirley	450–416	N/A	N/A
BG – Roussel	Shirley	448–420	N/A	N/A

^aComparison to literature is also provided.

In Figure B, the Pu 4f peaks are plotted after an 864 L exposure of H₂. As seen, there was a reduction in both the metal and satellite components of the 7/2 and 5/2 peaks. It is not possible to account for these spectral changes with metal only line-shapes, and the gray colored difference spectrum (B – A) reveals the appearance of four positive peaks. The additional peaks are described as plutonium hydride (PuH _x ) and plutonium carbide (Pu _x C _y ). PuH₂ has been observed on δ-Pu by XPS when dosing with H₂, but at much higher exposures (1 × 10⁴, 6 × 10⁷, and 4.8 × 10⁸ L). These experiments indicated the formation of a broad peak at ∼425.0 eV with a FWHM of 3.6 eV, positioned directly next to the satellite peak at 424.9 eV. Pu _x C _y species have been denoted in the Pu 4f for plutonium oxycarbide and carbide, , with positions at 423.4 and 436.4 eV. To account for these features in Figure B, a fifth (pink, 7/2) and sixth (violet, 5/2) component was used that corresponds to plutonium hydride (PuH _x ). The fit resulted in PuH _x peaks at 425.4 and 438.6 eV. The Pu _x C _y peaks also needed seventh (cyan, 7/2) and eighth (light gray, 5/2) components, which had fits with positions of 423.7 and 436.8 eV. Discrepancies between our measurements with Favart et al. are thought to be related to signal-to-noise, where the two species (PuH _x and Pu _x C _y ) are nucleation events and near the detection limit of the XPS. A table of line shapes, positions, FWHM, areas, and boundary conditions are provided in Table S2.

The summary of the H₂ gas dosing is described in Figure C by plotting atom % of Pu, O, H, and C throughout the experiment. On the left axis in black, the Pu atom % (black squares) is shown for sputtered cleaned Pu and after 432 and 864 L of H₂ exposure. First, the Pu was ∼88.7% because the surface was freshly sputtered, and the O atom % (blue squares) was ∼11.2% from interstitial oxygen atoms. Upon H₂ exposure, both decreased to 82.3% and 5.8% from the nucleation of both PuH _x and Pu _x C _y species. This is noted by an increase of both H (2.0%) and C (9.9%) at 432 L. At 864 L, both PuH _x and Pu _x C _y growth rates began to slow. The final value for PuH _x was ∼3.8%, and Pu _x C _y was ∼10.6%. The decrease in the growth of PuH _x is attributed to being destabilized by surface oxygen and the onset of Pu _x O _y formation. Here the final oxygen percentage was ∼7.2%, but a significant fraction remained in the subsurface, which is why Pu₂O₃ did not appear on the difference spectrum. Analogous observations were seen by Allen et al., where uranium hydride (UH₃) was oxidized by trace water vapor or oxygen shortly after H₂ dosing.

To verify this observation, collection of both the O 1s and C 1s peaks during the H₂ doses were completed. In Figure A, a plot shows the evolution of the O 1s peak from 533–528 eV. Initially, the peak was centered at ∼530.2 eV, indicative of Pu _x O _y . This is similar to reports by others that oxygen is always present in Pu and will slowly oxidize the metal when under UHV. , In blue, the first H₂ dose led to a decrease in the O 1s intensity. This indicates that H₂ was able to adsorb and attenuate the Pu _x O _y signal. However, in the second H₂ dose (green line), there was an increase in the O 1s signal. This would be in line that oxygen was affecting the growth rate of the PuH _x . No hydroxyl (−OH) peaks were detected at 532 eV. More on the Pu 4f and O 1s behavior is discussed below.

9.

(A) O 1s peak showing both reduction and increase in signal from H₂ dosing. (B) C 1s peak shows growth of PuC from H₂ dosing.

In Figure B, effects of H₂ are shown on the C 1s peak. For metal Pu, the broad Pu 5p_1/2 peak is observed. Indeed, there is a noticeable, and small feature just before 282 eV, but given the signal-to-noise of the scan, it is not possible to resolve and should not be labeled a chemical species. However, after the two doses of H₂, there was formation of a narrow and prominent peak at ∼281.8 eV. This peak appears to be related to plutonium carbide (PuC), which has been observed previously. , Note that during the second H₂ dose, the PuC intensity was unchanged.

3.3. Results of ToF-SIMS

Shown in Figure are the negative ions for the low mass regime (0–70 m/z) at 300 K. At 0 L the ion spectrum indicates a sputtered clean surface, as only ¹⁶O^–, ¹⁶O¹H^–, ¹⁹F^–, ¹²C₂ ¹H^–, ¹²C₂ ¹H₃ ^–, ¹⁶O₂ ^–, ³⁵Cl^–, and ³⁷Cl^– are present. This is due to the sensitivity of the ToF-SIMS instrument in detecting halogens due to the electropositivity, which will be detected even at low levels of ppm amounts in the material. Upon exposure to H₂ gas (90, 180, 300, and 400 L), an immediate increase of ¹⁶O¹H^–, ¹²C₂ ¹H^–, ¹²C₂ ¹H₂ ^–, and ¹²C₂ ¹H₃ ^– peak fragments were detected, noted by the growth of a carbonate (¹²C¹⁶O₃ ^–) peak. In the high mass spectra from 238–245 m/z (Figure A), the ²³⁹Pu¹H _x ^– fragment patterns were observed in the sputter cleaned 0 L spectra, indicative of clean Pu metal, but upon exposure to H₂ gas there was no apparent trend in the intensity changes. However, in Figure B (252–268 m/z), there was an increase in the ²³⁹Pu¹⁶O¹H₂ ^– indicating physisorption of the H₂ gas. There were also increases in ²³⁹Pu¹⁶O^– from reoxidation in UHV and, interestingly, an increase in intensity of ²³⁹Pu¹²C₂ ^– from the addition of H₂. By cooling down the sample with liquid nitrogen, the oxygen diffusion kinetics can be slowed to detect any apparent intermediate species at 1–2 ML. The low mass region (Figure ) is similar to the room temperature, where the ¹⁶O¹H^–, ¹²C₂ ¹H₁ ^–, ¹²C₂ ¹H₂ ^–, and ¹²C₂ ¹H₃ ^– increased upon exposure to the H₂ (90, 150, and 300 L), but the relative intensity between the ¹²C₂ ¹H₂ ^– and ¹²C₂ ¹H₃ ^– fragment pattern differed where ¹²C₂ ¹H₂ ^– was greater than the ¹²C₂ ¹H₃ ^– fragment peak. There was also a minor increase in the ¹²C¹H^– fragment, whereas ¹²C^– was essentially stable. In addition, a ¹²C₂ ¹H₅ ¹⁶O ^– fragment peak appeared, which was not present in the room temperature low mass spectra (Figure ). The detection of a carbonate ¹²C¹⁶O₃ ^– peak was also seen. As for the high mass spectra at low temperature (Figure A), there were slight increases in the ²³⁹Pu¹H _x ^– peaks, with an intensity decrease observed 1020s after the H₂ gas was turned off. In Figure B, there was not only the intensity increase in the ²³⁹Pu¹⁶O¹H₂ ^– as observed in the room temperature exposure, but also a slight increase of a ²³⁹Pu¹⁶O¹H₃ ^– fragment peak. Overall, the low temperature results imply that the H₂ is weakly chemisorbed, causing the molecules to dissociate and nucleate PuH _x sites but also reacting with adsorbed oxygen species. This is concluded from the increase in the ¹⁶O¹H^–, ¹²C₂ ¹H₅ ¹⁶O ^–, ²³⁹Pu¹⁶O¹H₂ ^–, and ²³⁹Pu¹⁶O¹H₃ ^– fragment peaks observed at low temperatures. It is well-known that Pu metal is highly reactive, therefore the intermediate species ¹²C₂ ¹H₅ ¹⁶O^– and ²³⁹Pu¹⁶O¹H₃ ^– would not be distinctly observed at room temperature, because they would be immediately dissociated and, thus concluding, an apparent physisorb reaction. ,,,, Lastly, the increase of ²³⁹Pu¹²C₂ ^– and ²³⁹Pu₁₂C₂H^– indicates that H₂ must also be promoting the formation of Pu carbide.

10.

ToF-SIMS low mass spectra from 0 to 70 m/z of H₂ gas exposure on α-Pu at room temperature. Note that 0 L is an Ar⁺ sputtered-clean metal surface.

11.

ToF-SIMS high mass spectra from (A) 238 to 245 m/z and (B) 252–268 m/z of H₂ gas exposure on α-Pu at room temperature. Note that 0 L is an Ar⁺ sputtered-clean metal surface.

12.

ToF-SIMS low mass spectra 0–70 m/z of H₂ gas exposure on α-Pu at low temperature using liquid N₂. Note that 0 L is an Ar⁺ sputtered-clean metal surface.

13.

ToF-SIMS high mass spectra (A) 238–245 m/z and (B) 252–268 m/z of H₂ gas exposure on α-Pu at low temperature using liquid N₂. Note that 0 L is an Ar⁺ sputtered-clean metal surface.

3.4. Results of Density Functional Theory Calculations

As shown in the XPS and ToF-SIMS measurements, α-Pu will have both oxygen and carbon surface species when exposed to atmosphere and when handled in UHV. To understand the different scenarios of H₂ interaction with Pu, consider the structural representations outlined in Figures –. In Figure , no H₂ dissociation was observed from the room temperature AIMD simulation consistent with the ToF-SIMS room temperature experiments. This is because the thin oxide layer acts as a protective barrier and subsequently prevents charge transfer from the surface Pu atoms to the H₂ molecules. This suggests that H₂ dissociation on the thinly oxidized α-Pu surface is an activated process and that the presence of O on the surface prevents H₂ from readily dissociating to form surface metal hydrides. The average Pu–H distance was 3.2 Å and the adsorption energy was 0.2 eV/H₂, indicating that the nature of the interaction of the H₂ molecules with the surface is physisorption. Furthermore, no appreciable charge transfer from Pu to H was observed.

For the simulations shown in Figure A, we note that some of the H₂ molecules partially dissociated on the pristine α-Pu(020) surface, while some of the molecules desorbed from the surface. The desorption process arises from the strong intramolecular repulsion between the molecules on the pristine surface. A lone H₂ molecule has a physisorption energy of −0.2 eV, but with several adsorbed H₂ molecules (16, Figure ), that energy increases to 0.03 eV/H₂, which is unfavorable. In Figure B–D, about half of the H₂ molecules on the sputtered surface dissociated and formed bonds with surface Pu atoms, however no H₂ molecular desorption was observed in contrast to the pristine surface. The partial H₂ dissociation in Figure A–D arises from the fact that the Pu 6d and 7s states readily interact with the H 1s states to form hybrid states (i.e., the interstitial impurities do not prevent charge transfer from the surface Pu atoms to H). Furthermore, no appreciable migration of the O and C interstitial atoms from the interior region to the surface layers were observed. It must be mentioned, however, that the AIMD simulations were for only 3 ps (the computational cost for long-time AIMD simulations is quite high for such large system sizes) and hence slow diffusive processes may not be observed within the simulation time span, especially at room temperature.

In Figures –, we show the projected density of states (PDOS) for Pu, O, and H in the structural configuration shown in Figure (H₂ physisorbed on thinly oxidized α-Pu surface) and Figure C (partially dissociatively chemisorbed H₂ on defected α-Pu surface mimicking sputtering with O interstitials). In both cases, we observe the following: (i) the Fermi level is dominated by the Pu 5f electron states, with some contribution from the Pu 6d states; (ii) the narrow feature of the Pu 5f states, which is often attributed to 5f electron localization, can be seen; (iii) the thin surface oxide layer in Figure and local bulk oxide in Figure C are formed via the hybridization of the O 2p states with the Pu 6d and 7s states. In Figure , the energies of the H 1s states display very little overlap with the Pu and O valence states; this is a signature of physisorption, which agrees with the computed weak H₂ adsorption energy and large H distances from the surface observed in Figure . In Figure , the hybridization of the H 1s state with the other electronic states can be seen in the −6 eV to −3 eV region; this arises from the interaction of dissociatively chemisorbed H atoms with the metal.

14.

Projected density of states of Pu (5f, 6d and 7s), O (2p), and H (1s) for the configuration in Figure . The Fermi level is denoted by the dashed vertical line.

15.

Projected density of states of Pu (5f, 6d and 7s), O (2p), and H (1s) for the configuration in Figure C. The Fermi level is denoted by the dashed vertical line.

To compare the DFT simulations directly to experiment, we computed the valence band XPS spectra of the system. The intensity, I, of the valence band XPS spectra was computed as

$$I(E)=g(E)f{(E,T)}^{*}{I}_{\gamma }{(E)}^{*}{I}_{\sigma }(E)$$

where the asterisk symbol * denotes convolution and E is the binding energy measured relative to the Fermi level. Here, g(E) = ∑ _i δ­(E – E _i ) is the global density of states of the single-particle Kohn–Sham DFT eigen energies {E _i }, $f(E,T)=\frac{1}{1+\exp \mathrm{}(E/(k_{\mathrm{B}}T))}$ is the Fermi–Dirac function at electronic temperature T, $I_{\gamma }(E)=\frac{1}{\pi }\frac{\gamma }{E^2+{\gamma }^2}$ is a Lorentzian function of width γ which accounts for the photon lifetime broadening, and $I_{\sigma }(E)=\frac{1}{\sqrt{2\pi {\sigma }^2}}{\mathrm{e}}^{-E^2/(2{\sigma }^2)}$ is a Gaussian function of width σ which accounts for the instrumentation broadening. I(E) was computed using T = 300 K and Lorentzian and Gaussian widths of 0.2 and 0.5 eV, respectively.

In Figure , the DFT valence band XPS spectra for the structure in Figure C is depicted alongside experiment after exposures to H₂. The match between theory and experiment in the immediate vicinity below the Fermi level is in good agreement; this feature originates from the localized nature of Pu 5f states. The small satellite feature in the DFT spectrum at −3 eV is due to the Pu 6d electron states. While there are numerical differences between theory and experiment in the E < −4 eV region (this region is dominated by the O 2p and H 1s states); the qualitative match with experiment is adequate. Overall, the DFT calculations seem to capture the valence DOS features of Pu quite well in comparison with experiment.

16.

Comparison of the simulated valence band photoemission spectra with experiment. The experimental data has been scaled to match the maximum of the simulated spectra. The Fermi level is denoted by the dashed vertical line.

In Figure , we depict the plot of the valence band XPS spectra for the four-structures in Figure A–D. The shape of the spectra in the vicinity of the Fermi level can be associated with the nature of the Pu 5f bands. In general, the spectra are narrower for the sputtered surfaces compared to the pristine surface; this is due to the increased density of Pu dangling bond defects induced by sputtering. Defect states, and more specifically the Pu 5f electron states at the defect sites, are localized. The sputtering-induced Pu 5f localization, therefore, manifests in the form of the narrow features in the spectra near the Fermi level. The region below −4 eV are due to the admixture of Pu 6d and 7s states with H 1s, O 2p and C 2p states.

17.

Comparison of the simulated valence band photoemission spectra for structures in Figure A–D. The Fermi level is denoted by the dashed vertical line.

4. Discussion

An α-Pu sample was analyzed in its “as-received” state and after having Ar⁺ sputtered the surface until Pu metal was reached. With the XPS measurements, the chemical state was determined and used to study the varying effects of H₂ at exposures of 432 and 864 L at 300 K. Comparatively, ToF-SIMS assessed the Pu metal before and after H₂ dosing from 0–400 L at 300 K and from 0–300 L at cold temperatures via liquid N₂ cooling. From these experimental results, DFT calculations and AIMD simulations were then performed to bring further understanding of the mechanism of H₂ interaction with Pu at 300 K in the presence of O and C impurities on the surface and interstitials within the subsurface.

4.1. Analysis of α-Pu Surface in As-Received and Sputtered Conditions

Before discussing the H₂ experiments, it is important to consider and recognize the state of the α-Pu surface before and after Ar⁺ sputtering. As shown in Figures –, the as-received surface had significant fractions of oxygen and carbon pertaining to adventitious carbon and Pu oxides. Figure indicated that composition was a mixture of primarily PuO₂ with onset of Pu₂O₃ from UHV autoreduction. This was based on the peak positions of the Pu 4f_5/2 and Pu 4f_7/2 for both types of oxides, and the position of the 4+ satellites. The spectral region from 35–0 eV also had higher binding energies for the Pu 6p_1/2, 6p_3/2, and 5f peaks, i.e., representative of an oxidized surface. These observations are both aligned to behavior of α-Pu and δ-Pu after atmospheric exposure, which is growth of PuO₂ followed by its autoreduction to Pu₂O₃ in UHV. However, the length of time for autoreduction is ∼hours to days for surfaces exposed to atmospheric pressures, , explaining why the 4⁺ features were still measurable. Both the O 1s and C 1s peaks also provide additional information on the type of carbon on the as-received surface. For example, the O 1s had satellites where C–O/CO and O–H bonds would occur, and the C 1s had appreciable amounts of C–C/CC carbon.

To confirm that both the oxygen and carbon were not bulk features but layered materials, the α-Pu surface was Ar⁺ sputtered cleaned, and Figures – showed the removal of both elements. First, the Pu 4f peaks saw total removal of PuO₂ and Pu₂O₃, leaving behind a clean Pu metal surface with the 4f_7/2 peak at 422.1 eV and the satellite at 424.9 eV. The Pu valence band also shifted its peaks (6p_1/2, 6p_3/2, and 5f) to lower binding energy, along with a significant increase to the collected signal. For the O 1s and C 1s, all the adventitious carbon was removed and nearly all of the oxygen. This is shown in Figure A, where the O 1s peak indicated that residual Pu _x O _y was still on the surface after sputtering for 30 min. This was attributed to not sputtering the surface below the oxygen detection limit. However, to prove that this was inconsequential, Figure S3 indicated that reoxidation in UHV for 40 min only permitted the O 1s to grow back ∼0.07 ML of oxide, and both the Pu 4f and valence bands remained unchanged during this time scale. This is important to mention, given that reoxidation of the Pu surface can interfere with the results when gas dosing. Fortunately, this seems to have minimal effects with α-Pu given the time length and pressures of the H₂ experiment. In Figure B, the Pu 5p_1/2 was observed in the metal surface, and there may have been residual carbon at ∼282 eV, but this peak was not quantifiable.

4.2. Analysis of α-Pu and H₂ on the Sputtered Surface by XPS and ToF-SIMS

The DFT calculations from Figure are important because they illustrate the effect of H₂ interaction with the thinly oxidized surface, which simulates an α-Pu surface in UHV for an extended period of time (∼hours). The AIMD simulation informed that a significant oxygen fraction on the surface led to a Pu _x O _y layer (40% Pu₂O₃ and 60% PuO₂), causing the H₂ to become physisorbed. In contrast, dissociation and chemisorbed H was found to occur on the metal surface or “sputtered” metal with O and C impurities present. Since the XPS and ToF-SIMS depicted changes in spectra from H₂ addition on a sputtered surface, the DFT calculations from Figure are more aligned to the experiments. For instance, the pristine and sputtered α-Pu surface both had ∼50% H₂ dissociation and bonding with Pu atoms. Several H₂ molecules desorbed from the pristine surface, but for the sputtered surface, no H₂ molecular desorption was observed. Chemisorption of H has been computed for various surfaces of δ-Pu, such as (111), (100), and (110). − ,,, Thus, it is not surprising that similar findings would hold true for α-Pu(020), especially when containing high amounts of dangling Pu atom bond defects.

The XPS experiments required Ar⁺ sputtering for >30 min to reach metal, thus the freshly sputtered surface is consistent to the structural representation of Figure B, i.e., highly defective. However, because the O 1s in Figure A was not sputtered below the oxygen detection limit, it is reasonable to consider Figure C as the most accurate representation of the experiment. From the XPS and ToF-SIMS results, the H₂ gas was dosed at ∼3.59 × 10^–7 Torr and ∼1 × 10^–7 Torr, respectively. The H₂ collision rate would be ∼1–5 collisions/sec/nm at 300 K between the two methods. Relative to the AIMD simulation, the α-Pu slab, whose surface area = 129.88 Ų, would become saturated with H₂ molecules in less than two seconds if one assumes a sticking coefficient at unity. Given that the H₂ dosing experiments were ∼40 min for XPS and ∼60 min for ToF-SIMS, the surface would have been consistently passivated with H atoms, while remaining in proximity to both surface and subsurface O and C atoms.

The XPS results reveal a few major points to the DFT results. (1) The Pu valence band showed a decrease in intensity from H₂ dosing (Figure ), while UHV exposure did not (Figure S3F). (2) The Pu 4f_7/2 and Pu 4f_5/2 peaks were affected by H₂. (3) The O 1s and C 1s had varying behaviors as the H₂ reaction was carried out. For point 1, H₂ was not observed to affect the Pu 6p_1/2 and 6p_3/2 peaks in Figure , and a decrease was specifically in the 5f peak’s vicinity. DFT confirmed that Pu 6d and Pu 7s electrons primarily interact with H 1s to form hybrid states, and the nearby localized Pu 5f states were not involved. This explains the reason for a modest change in the XPS valence, since most of the intensity would originate from Pu 5f. Due to XPS’s poor resolution in the valence, it is not possible to deconvolute the 6d and 7s orbitals from the 5f, but the signal decrease is indicative of H₂ bonding, which was confirmed by Butterfield et al. in δ-Pu.

Regarding points 2 and 3, H₂ can be indirectly probed with measurement of Pu 4f and O 1s peaks. Figure A showed fitting of the metal and satellites on both the Pu 4f_5/2 and Pu 4f_7/2 peaks based on parameters by Roussel. When H₂ was dosed at 432 and 864 L, the Pu 4f spectrum required additional functions to fit with the experiment. It was chosen to use a component, described as PuH _x and measured by Larson et al. These experiments involved H₂ doses that were ∼10,000 times greater, and a rapid reduction in the metal peak intensity was observed. In Figure , the PuH _x peaks nucleated from addition of H₂ and caused a decrease in the Pu 4f metal peaks while simultaneously lowering the satellite intensities. During the second dose, the PuH _x features continued to grow, but the growth rate had slowed dramatically on the 4f_7/2. The explanation is that Pu’s propensity to oxidize leads to the addition of oxygen onto the surface, which then occupies Pu surface sites. It is likely that H absorbs into the Pu lattice or desorbs as H₂, and once an O atom binds to Pu, the site becomes poisoned. This was validated with δ-Pu, using H₂O vapor by Fricke and Hernandez The calculated adsorption energies for the Pu–O bond is approximately −5.0 eV, ,, and −0.43 eV for the Pu–H bond. Thus, as the oxygen builds up, the nucleation of Pu _x O _y is noticeable, which is picked up in proximity to the satellites of the Pu 4f spectrum. Likewise, Figure C clearly displays that the PuH _x growth was self-limiting, and Figure A shows that O 1s had complementary behavior. At first, the O 1s signal decreased, which would imply nucleation of PuH _x and coverage of residual oxygen in the subsurface, but during the second H₂ dose, it increased due to a passivating Pu _x O _y layer, disrupting the PuH _x growth. It also must be pointed out that −OH features did not appear on the O 1s peak. The reason is, at this point in the experiment, there is not sufficient oxygen coverage to facilitate formation of hydroxyls. Note that for Pu₂O₃ films grown by O₂ dosing, there is an −OH feature at ∼532.5 eV, which would be formed due to reaction with background H₂.

To provide a thickness estimate of PuH _x /Pu _x C _y layers from Figure , a similar methodology is applied as for oxygen on UHV exposed α-Pu (Figure S3). In this case, the Pu 4f, O 1s, 5p_1/2, and C 1s are analyzed. The RSF values used for Pu 4f = 20.42, O 1s = 0.711, and C 1s = 0.296. It was found that the metal surface had atom % of Pu = 88.7% and O = 11.3%. The oxygen resides as interstitials within the Pu subsurface. When H₂ was exposed to 432 and 864 L, the Pu decreased to 82.3% and 78.5%. The decrease in metal is spurred by the formation of PuH _x and Pu _x C _y , which increased atom % of H to 2.0% and 3.8% and atom % C to 9.9% and 10.6%. The thickness of this grown layer assumes a homogeneous mixing of hydride and carbide over an α-Pu, O interstitial surface using atom % of H = 3.8% and C = 10.6% (final atom % of O = 7.2%). Bear in mind that some oxygen is affecting the film, since the O atom % did increase after the second H₂ dose. Despite, the modified Hill equation with an EAL value of ∼1.4 estimates this speciated layer to be ∼1.9 Å. Given the limits of this approximation, it is not possible to extract a monolayer coverage of PuH _x and Pu _x C _y , but analysis considering all PuH _x or all Pu _x C _y affords coverage below 1 ML. Considering the time scales of these experiments, the PuH _x layer is quite unstable at 300 K in UHV.

The ToF-SIMS results were able to validate the XPS measurements and concluded that H₂ dosing led to immediate increases in ¹⁶O¹H^– signal at low mass and ²³⁹Pu¹⁶O¹H₂ ^– and ²³⁹Pu¹⁶O^– at high mass. These indicate fragmentation events of complexes from reaction with H₂. Note that the ToF-SIMS’s H₂ exposure went up to 400 L, thus Figures – would be most similar to the first H₂ dose in the XPS experiment. By doing cold experiments via ToF-SIMS, the lifetimes of surface adsorbates are extended, increasing the sensitivity in capturing the dissociation of H₂. It was observed in Figures – that the same peaks at 300 K were present, but slight increases were seen in ²³⁹Pu¹H _x ^– peaks and the ²³⁹Pu¹⁶O¹H₃ ^– fragment peak. All things being considered, the conclusion is that H₂ dosing of Pu at 300 K leads to the nucleation of PuH _x , but it becomes disrupted by the subsequent reoxidation of the Pu surface back into Pu _x O _y from oxygen addition. It is also noted that with enough oxygen coverage, H begins to form −OH species. Both the XPS and ToF-SIMS results are in good agreement with the AIMD simulation of the sputtered α-Pu slab in Figure C, where H₂ adsorbs and then dissociates, despite being in the presence of O atoms.

The last point to consider is the effect that H₂ and α-Pu have on residual carbon species in the UHV chamber, C adatoms on the surface, or C interstitials in the subsurface. The AIMD simulations in Figure D indicated that C atom interstitials did not negatively impact dissociative chemisorption of H₂. Instead, the majority of H₂ molecules dissociated and formed bonds with Pu atoms. Simulations only lasted for 3 ps, so it is unclear whether diffusion rates of either subsurface oxygen or carbon interstitials are affected by dissociated H. However, the XPS results in Figure B showed that exposure to H₂ forms a PuC feature, which was not observed in UHV alone (Figure S3D). A consideration is the catalytic abilities of Pu and H₂ being able to dissociate residual carbon gas into elemental C, i.e., the Bosch reaction (eq ) or to promote C adatom desorption.

$$\mathrm{CO}(\mathrm{g})+{\mathrm{H}}_2(\mathrm{g})\to \mathrm{C}(\mathrm{s})+{\mathrm{H}}_2\mathrm{O}(\mathrm{g})$$ 7

It appears nucleation of Pu–C bonds then translates to lower activation energies for subsequent C–O bond cleavage, which is interesting considering the XPS instrument’s P _base is ∼2 orders of magnitude less than P _H2. It is also important to note that the PuC feature in the C 1s spectrum is in the same position as Pu oxycarbide films studied by Roussel. , Finally, the ToF-SIMS results at 300 K indicated that H₂ gas led to an immediate increase in ¹²C₂ ¹H^–, ¹²C₂ ¹H₂ ^–, and ¹²C₂ ¹H₃ ^– peak fragments and ¹²C¹⁶O₃ ^–. For the high mass range, ²³⁹Pu¹²C₂ ^– was observed to grow in. The liquid N₂ experiments also showed the addition of ¹²C₂ ¹H₅ ¹⁶O ^–, which was not stable at 300 K. Thus, both the XPS and ToF-SIMS are consistent that H₂ is critical for aiding in the nucleation and formation of stable PuC surface species.

4.3. Mechanism of H₂ Interaction on α-Pu Surface at 300 K

In our experiments, the α-Pu surface was Ar⁺ sputtered to remove any contaminant oxygen and carbon, which would have deposited when exposed to the atmosphere. After the Pu metal was reached, it is safe to assume that the α-Pu surface is highly defective and contains several dangling Pu bonds and very low concentrations of oxygen (Figure C). If the surface is allowed to sit under UHV for hours, then reoxidation occurs under the following reaction scheme

$$x\mathrm{P}\mathrm{u}(\mathrm{s})+\mathrm{y}{\mathrm{O}}^{*}(\mathrm{ads})\to {\mathrm{Pu}}_x{\mathrm{O}}_y(\mathrm{s})$$ 8

However, if the freshly sputtered surface is immediately exposed to H₂ gas, both the experiments and DFT calculations indicate that H₂ adsorbs, saturates the surface, and ∼50% of the H₂ molecules dissociate with the remaining molecules staying adsorbed. For dissociated H, the atoms are free to bond with Pu forming nucleation sites of PuH _x , which were detected by XPS (Figures –8) and ToF-SIMS (Figures –). The reaction would be described as

$$\mathrm{Pu}(\mathrm{s})+{x\mathrm{H}}^{*}(\mathrm{ads})\to \mathrm{Pu}{\mathrm{H}}_x(\mathrm{s})$$ 9

This is in line with dissociative H₂ chemisorption that has been modeled for various facets of δ-Pu and α-Pu. − ,,, Therefore, after a period of time, a steady state is reached, where fractional amounts of the H₂ molecules dissociate, H atoms absorb into the Pu bulk, or H₂ molecules desorb. Meanwhile, as occupied Pu surface sites momentarily free up, the balance between H₂ adsorption is in competition by O adsorption. As noted above, the adsorption energy of atomic O is ∼10 times of atomic H. Therefore, if a Pu–O bonds forms, the surface site is poisoned, which is consistent to the simulations of H₂O dissociation on δ-Pu by Fricke and Hernandez

One of the questions then is how does the mechanism for reoxidation of the Pu surface operate under UHV? A few possibilities are (1) The H₂ reacts with Pu₂O₃ outside the sputter crater, forming H₂O vapor, which then reacts with the Pu surface. (2) The O and H atoms are sourced from residual H₂ and H₂O, and given enough time, the surface becomes passivated. (3) The O₂ is already dissolved into the Pu lattice, and sputtering forms a concentration gradient, allowing O₂ to diffuse toward the surface to minimize the difference. For #1, this explanation is avoided because H₂ dosing did not lead to an observable increase in H₂O signal on the RGA. Point #2 is of the perspective that sputtered Pu metal is highly catalytic and will cleave adsorbed H₂, H₂O, and O₂ into constituent functional groups. Several models have confirmed that dissociation occurs for each of these gases. − ,− ,− Thus, competition between adsorbed H and O indicates that Pu _x O _y is more thermodynamically stable than PuH _x because of the difficulty in breaking the Pu–O bond. This explains why Figure C showed a decline in PuH _x growth rate throughout the extent of the experiment. However, P _base on the time scale of these experiments is not high enough magnitude to affect the XPS or ToF-SIMS measurements. For example, P _base = 3.75 × 10^–10 Torr would require ∼45 min to reach 1 L exposure, and reoxidation of the α-Pu surface at ∼40 min had features similar to gas doses >100 L. This demonstrates that point #2 is important to consider, but not the main driving factor behind Pu’s UHV behavior. For #3, if oxygen is in the bulk, then it may diffuse to the surface. This was tested by tilting the sample with the XPS instrument to increase sensitivity to the Pu bulk. Figure showed that there was an increase in the metal signal and a decrease in the satellite for the Pu 4f peak, but no finding of a Pu₂O₃ feature, implying the bulk does not contain excess oxygen deposits after sputtering the surface. Instead, it is reasonable that O is trapped in the first several MLs, which could be denoted as the selvage. Thus, #3 has merit as being the main oxidation mechanism since the distance is small enough for oxygen to diffuse to the surface within the experiment time scales. It is also interesting that hydroxyl species were not detected by XPS but were seen by ToF-SIMS measurements. This observation has to do with the depth sensitivity of XPS vs ToF-SIMS. In XPS, significantly more oxygen must be present to detect a stable Pu _x O _y film, which would be seen in the corresponding Pu 4f, O 1s, and valence spectra. At this point, enough O is present for −OH to be detected in the O 1s, which is commonly observed on Pu surfaces after O₂ dosing.

Considering point #3, reaction of H with Pu _x O _y is able to drive reactions such as

$${\mathrm{Pu}}_x{\mathrm{O}}_y(\mathrm{s})+{\mathrm{H}}^{*}(\mathrm{ads})\to {\mathrm{Pu}}_x{\mathrm{O}}_y+{\mathrm{OH}}^{*}(\mathrm{ads})$$ 10

For ToF-SIMS, sensitivity to the topmost 1–2 ML is enough to detect the interaction of H with O interstitials and adsorbates. This explains why it was difficult to measure growth of PuH _x since the oxygen quickly overtakes the hydrogen, and there is less access to detection of absorbed H in the selvage region. For example, PuH _x signal increased initially (Figure ), but then decreased, while Pu _x O _y fragmentation continued to increase along with evidence of hydroxylation (contribution from adsorbed H atoms). This is supported by the AIMD simulation in Figure that once the surface is oxidized, H₂ molecules no longer form hydrides.

Lastly, catalytic qualities of α-Pu are further supported by how PuH _x on the surface was able to react with carbon gas impurities. For example, the Bosch reaction (eq ). If α-Pu dissociates H₂O and H₂, then the activation energy is lowered for residual carbon gas, where adsorbed H₂ assists in the formation of elemental C. This would appear to be the mechanism because exposing a sputtered metal surface to the UHV environment for 40 min (Figure S3D) did not show an appreciable increase in C 1s signal. However, under a passivating H layer, there was growth of carbide and carbonate signals among both the XPS and ToF-SIMS measurements (Figures –). Thus, it is interesting that H₂ could be helpful and insightful information for forming oxy-carbide films similar to those grown by Roussel. ,

5. Conclusion

A polished α-Pu surface was analyzed before and after Ar⁺ sputtering, confirming that samples exposed to atmosphere are initially covered in adventitious carbon and PuO₂/Pu₂O₃. When the metal surface is dosed with H₂ gas, PuH _x is nucleated and grown. Although, due to Pu’s high affinity for oxygen, reoxidation of the surface quickly leads to formation of Pu _x O _y from diffusion of interstitial oxygen and reaction with residual background gases, which impedes PuH _x growth at 300 K making this layer unstable. DFT calculations and AIMD simulations confirmed this observation and indicated that interstitial O and C do not appear to have any effect on the adsorption or dissociation kinetics of H₂ molecules onto α-Pu, since it behaved similarly to a “sputtered” impurity free metal. This was demonstrated by fitting theory to the experiment in the valence region, showing the bonding is between Pu 6d and 7s states with H 1s, O 2p, and C 2p. It was also observed that H₂ exposure is an assistor in the nucleation of PuC, which is not observed to grow in its absence.

Glossary

Abbreviations

α-Pu

α plutonium

XPS

X-ray photoelectron spectroscopy

ToF-SIMS

time-of-flight secondary ion mass spectrometry

PuO₂

plutonium dioxide

Pu₂O₃

plutonium sesquioxide

PuH _x

plutonium hydride

Pu _x C _y

plutonium carbide

An

actinide

Ga

gallium

Al

aluminum

Ce

cerium

fcc

face-centered cubic

δ-Pu

delta-plutonium

Pu _x O _y

plutonium oxide

Pu _x O _y C _z

plutonium oxycarbide

O₂

oxygen

H₂O

water

H₂

hydrogen

UHV

ultrahigh vacuum

O^2–

oxygen anion

UHP

ultrahigh purity

O_ads

adsorbed oxygen

E _F

Fermi level

Ar⁺

argon ions

DFT

density functional theory

P _base

base pressure

Ag

silver

Au

gold

P _H2

partial pressure hydrogen

RGA

residual gas analyzer

LMIG

liquid metal ion gun

ML

monolayer

DZP

double-ξ basis with polarization orbitals

PBE

Perdew–Burke–Ernzerhof

GGA

generalized gradient approximation

AIMD

ab initio molecular dynamics

M _x O _y

metal–oxygen bond

C–O

σ-states

CO

π-states

EAL

effective attenuation length

TPP-2M

Tanuma-Powell-Penn

L

Langmuir

–OH

hydroxyl

PDOS

projected density of states

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.6c03998.

• Data tables for all the other figures (XLSX)

• Image of the α-Pu sample (Figure S1), table of the XPS instrument collection parameters (Table S1), residual gas analysis of the XPS vacuum chamber’s background (Figure S2), additional information on determination of atomic percentages and monolayer coverage of the Pu _x O _y layer, XPS region scans of Pu 4f, valence, O 1s, C 1s, and overlays of Pu 4f and valence of α-Pu after sitting in vacuum of ∼9 × 10^–10 Torr for 20 and 40 min (Figure S3), the O 1s and C 1s spectra during the XPS tilting experiments (Figure S4), overlay of Pu 4f after the α-Pu sample was dosed with 432 and 864 L of H₂ (Figure S5), and the fitting parameters for Figure B (Table S2) (PDF)

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.