Uranium (U) is one of the most fundamental nuclear fuels in the nuclear industry. Corrosion induced by hydrogen (H) and oxygen (O) has been proved to be a challenge for U storage. To protect U against environmental corrosion, surface protection techniques are being researched, among which coatings of aluminum (Al) on the U surface have been reported.^[1–3] Al electroplating was utilized in procession of U fuel elements^[4] in the early years, with other techniques such as magnetron sputtering introduced later. Lu et al.^[5] investigated the impact of different bias voltages of magnetron sputtering on microstructures and residual stress of Al coatings. For the growth mechanism, Zhou et al.^[6] reported that Al films expanded in an island growth mode when sputtered on a U substrate. Moreover, diffusion on the U/Al interface has also been studied and factors including doped elements and interfacial microstructures were found to play significant roles in diffusion behaviors.^[7–9]

The first-principles calculations based on density functional theory (DFT) have been utilized to study the structures and properties of U and relevant subjects. The lattice constants, elastic constants, electronic structures, and optical properties for α and γ uranium were studied by Chen et al.^[10] Moreover, Zhang et al.^[11] investigated property variances under pressure up to 100 GPa for α, β, and γ uranium. Taylor et al.^[12] showed that H adsorption into the bulk U was endothermic, with the square-pyramidal interstitial site as the most stable one, and later he demonstrated a periodic variation in the interactions between U and a series of impurity atoms [H–Ar].^[13] It has also been suggested that O impurities in bulk U could stabilize intrinsic vacancies by diminishing the vacancy formation energy from 1.98 eV to 1.7 eV and 0.8 eV for substitutional and interstitial O, respectively.^[14]

First-principles calculations have also been employed to study the interaction between Al coatings and U substrates. Nie et al.^[15] reported that upon Al adsorption, the α-U (001) surface contracted by about 3% and that 5f electrons of U atoms on the surface became delocalized, resulting from their hybridization with 3p electrons of Al atoms. It was further observed that upon adsorption of Al atoms, the spin magnetic moments of U atoms in the first layer decreased obviously and a ferrimagnetic phase emerged, different from a ferromagnetic phase reported by the previous research on a bare α-U (001) surface.

To reveal whether Al coatings on U substrates can play an important role against oxidation and hydrogenation of U, first-principles calculations were performed to study the interaction between O/H impurities and U/Al interfaces in the present work. A supercell consisting of a U/Al bilayer and a vacuum slab was constructed. U vacancies and impurity interstitials (H, O) were introduced at different distances from the interface. Single point energies of relaxed configurations were calculated to discuss the interactions between the defects and the interfaces. Moreover, the largest numbers of impurities trapped in a U vacancy on the U/Al interface were also presented.

The Vienna ab initio simulation package (VASP) code^[16] was utilized with projector augmented wave (PAW) methods for pseudopotential and the GGA approximation of Perdew et al. (PW91)^[17] for the exchange and correlation functional. Scalar-relativistic modeling of the relativistic effects was included in the PAW-core contribution, which is believed to reasonably supply the leading relativistic correction to the electronic structure.^[18] A basis set of plane wave functions up to a kinetic energy cut-off of 400 meV was chosen in valence wave function expansion. A k-point grid of 2 × 2 × 1 was generated under the Monkhorst–Pack scheme. Convergence criteria for the electronic and ionic relaxations were 10⁻⁵ eV and 0.02 eV·Å⁻¹, respectively. The electron distribution was described by the Methfessel–Paxton scheme with sigma of 0.1. The conjugation gradient algorithm was chosen for supercell relaxation. The parameters above were tested to narrow error down to about 2 meV per atom.

Considering computation efficiency and periodic boundary conditions, we constructed a supercell consisting of a U/Al interface under an orientation match of (001)_U || (001)_Al and [100]_U || [110]_Al. In interfacial matching, the structural units of 4a₀[100]_U × 2b₀[010]_U and were specifically chosen along the interfacial plane, which yields a mismatch of about 2.5%. Moreover, an appropriate lattice displacement parallel to the interface was also implemented to create a better fit on the interface. The atom arrangement near the interface is shown in Fig. 1. The initial interface distance and vacuum slab thickness were 2.6 Å and 15 Å, respectively. For either side of the interface, five layers of atoms were included to screen the effect of the vacuum slab.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) Vertical view of atoms near the U/Al interface.

The initial supercell was relaxed with all freedoms to reduce effects of mismatch on the interface. Then vacancies and impurities were introduced around the interface zone and the atoms were further relaxed with supercell size and shape fixed. Single point energies were then calculated and compared.

Besides, the chemical potential of bulk U atoms and other reference data were also considered. For comparison with a previous research by Taylor,^[13] a supercell of (4a × 2b × 3c) α-U primitive cells was constructed, with PAW pseudopotential and GGA-PW91 functional utilized. A k-point grid of 4 × 4 × 3 was set by the Gamma scheme. The energy cutoff was 500 eV. The electronic distribution was determined by the Methfessel–Paxton scheme with sigma of 0.2. Convergence criteria for the electronic and ionic relaxations were 10⁻⁴ eV per atom and 0.05 eV·Å⁻¹, respectively. The supercell relaxation was implemented by the conjugate gradient algorithm. The calculated formation energies, E_f, of U vacancy and H/O interstitials with reference to bulk U and a single H/O atom are given in Table 1. It can be seen that the E_f error of H is within 5%, while that of O is about 10%, which might be due to refinement of k-point sampling in the present research.

As U contains f valence electrons, the DFT+U method was often adopted in the DFT calculations. In our cases, we also conducted DFT+U_eff calculation for the U/Al interface with an appropriate U_eff, and found that the results with the addition of U_eff are very close to those without U_eff. For simplicity, only the first-principles calculations without U_eff are given in the present study.

Table 1.

Table 1.

Table 1. The formation energy of intrinsic vacancies, hydrogen interstitials, and oxygen interstitials in bulk U. Results from early reports are also listed. . Ef/eV Present work Early reports Vacancy 1.802 1.69–1.98a H –1.96 –1.91a O –5.9 –5.3c a Refs. [14] and [19] b Ref. [12] c Ref. [13].

Table 1. The formation energy of intrinsic vacancies, hydrogen interstitials, and oxygen interstitials in bulk U. Results from early reports are also listed. .

In the present research, the U/Al interface was demonstrated to be a crucial adsorption position for vacancies in the U lattice, with energy difference amounting to 0.5 eV. In general, H impurities tend to converge in the U lattice while O impurities in the Al lattice, indicating a more robust role of Al coating against oxidation than against hydrogenation in protection of U. Under a relative compact atomic arrangement without vacancies, no obvious trap effect was shown for H and O interstitials on the U/Al interface. However, with a single U vacancy introduced, the interface exhibited an evident trap effect for H and O interstitials. Moreover, single U vacancies were estimated to be capable of containing clusters of at most 8 H or 6 O impurities. In general, the U/Al interface exists as a sink adsorbing vacancies and impurities from the lattice which threaten the mechanical properties of the coatings.