Core level shifts in Cu-Pd alloys as a function of bulk composition and structure
J. R. Boes, P. Kondratyuk, C. Yin, J. B. Miller, A. J. Gellman and J. R. Kitchin, “Core level shifts in Cu-Pd alloys as a function of bulk composition and structure”, Surface Science 640, 127-132 (2015). https://doi.org/10.1016/j.susc.2015.02.011
This summary was written by a language model from the paper and has not been checked line by line. Rely on the published paper for details.
Question
Cu1-xPdx alloys are used as dense hydrogen separation membranes, where Cu adds mechanical strength and sulfur tolerance. Although both pure metals are FCC, the alloy forms an ordered B2 (CsCl-type, BCC) phase near x of 0.4 below about 873 K, and this phase matters for performance: it dissolves only about one fifth as much hydrogen as the FCC phase but has roughly ten times higher hydrogen permeance, while H2 dissociation is faster on the FCC surface. The paper asks whether X-ray photoelectron core level shifts (CLSs), interpreted with density functional theory, can locate the B2 phase region as a function of composition, which would give a phase-sensitive probe of the near-surface region (about 1 to 2 nm) of films and particles where diffraction is difficult (Boes et al. 2015).
Approach
The experiments used composition spread alloy films (CSAFs): roughly 100 nm thick Cu-Pd films on Mo with a continuous lateral composition gradient across about 1 cm, made in ultrahigh vacuum with either an offset filament or a rotating shadow mask tool and annealed at 800 K for 1 h. Electron backscatter diffraction (EBSD) mapped FCC versus BCC structure across the film. Spatially resolved XPS (200 um monochromated Al K-alpha spot) recorded Cu 2p3/2 and Pd 3d5/2 spectra over the whole range x = 0 to 1, after He+ sputtering to clean the surface and annealing between 400 and 800 K. On the computational side, VASP with PBE and PAW (400 eV cutoff) was used; ATAT cluster expansions identified the ground state FCC and BCC configurations, and Cu CLSs were computed with the complete screening method, in which one core electron is moved into the valence band and the shift is the difference of ionized minus ground state total energies in the alloy relative to pure Cu (Boes et al. 2015).
Key findings
EBSD showed FCC structure at Cu-rich and Pd-rich compositions, BCC structure in the known B2 range, and mixed regions in between, so the thin film behaves like the bulk alloy. The cluster expansion hulls reproduced the known ground states, with the B2 structure at x = 0.5 more stable than FCC configurations at that composition.
The measured Cu 2p3/2 CLS fell linearly with Pd content from the Cu-rich side, as in earlier literature, but from about x = 0.35 to 0.56 it deviated anomalously before the linear trend resumed, a range matching the B2 and adjacent mixed-phase regions at 800 K. The authors note this deviation had not been reported before and suggest earlier samples may have been annealed where only the FCC solid solution exists. He+ sputtering at 1 keV erased the anomaly, consistent with sputter-induced disordering, and it returned after annealing at 800 K for 1 h; in fact it was already fully restored after 1 h at 600 K.
DFT gave nearly identical Cu CLSs of about -1.1 eV for the ordered ground-state FCC structure and stoichiometric B2 at x = 0.5, both more negative than the experimental value near -0.9 eV. A 64-atom random FCC supercell (32 Cu, 32 Pd) gave a distribution of Cu CLSs whose mean lay close to the experimental value at x = 0.5, showing that configurational disorder raises the average CLS. The Cu CLS varied roughly linearly with the number of Cu-Cu nearest neighbors, and atoms beyond the first shell had little effect. Because the disordered FCC phase has more Cu-Cu neighbors than ordered B2, B2 should show a more negative CLS. For off-stoichiometric B2 near x = 0.4, the authors place excess Cu on the Pd sublattice and use Vegard’s law for the lattice constant; both the smaller lattice constant and the added Cu neighbors decrease the magnitude of the CLS, consistent with the experimental trend. They conclude that the anomaly reflects a nonstoichiometric, partially ordered B2 phase (Boes et al. 2015).
Significance
The work shows that XPS core level shifts are sensitive to chemical ordering, not only composition, so they can distinguish ordered B2 from disordered FCC Cu-Pd and map phase boundaries with nearly continuous composition resolution on a single CSAF. This gives a high-throughput route to finding the B2 window as a function of temperature, relevant to stabilizing B2 to higher temperatures for membranes and extendable to ternary alloys (Boes et al. 2015).