Formation of vacancies and metallic-like domains in photochromic rare-earth oxyhydride thin films studied by in-situ illumination positron annihilation spectroscopy

Rare-earth (RE) oxyhydride thin films show a color-neutral, reversible photochromic effect at ambient conditions. The origin of the photochromism is the topic of current investigations. Here, we investigated the lattice defects, electronic structure, and crystal structure of photochromic ${\mathrm{YH}}_x{\mathrm{O}}_y$ and $\mathrm{Gd}{\mathrm{H}}_x{\mathrm{O}}_y$ thin films deposited by magnetron sputtering using positron annihilation techniques and x-ray diffraction, in comparison with Y, ${\mathrm{YH}}_{\sim 1.9}$, ${\mathrm{Y}}_2{\mathrm{O}}_3$, Gd, $\mathrm{Gd}{\mathrm{H}}_{\sim 1.8}$, and ${\mathrm{Gd}}_2{\mathrm{O}}_3$ films. Positron annihilation lifetime spectroscopy (PALS) reveals the presence of cation monovacancies in the as-deposited Y and ${\mathrm{YH}}_{\sim 1.9}$ films at concentrations of $\sim {10}^{-5}$ per cation. In addition, vacancy clusters and nanopores are found in the as-prepared ${\mathrm{YH}}_x{\mathrm{O}}_y$ and ${\mathrm{Y}}_2{\mathrm{O}}_3$ films. Doppler broadening positron annihilation spectroscopy (DB-PAS) of the Y- and Gd-based films reflects the transition from a metallic to an insulating nature of the RE metal, metal hydride, semiconducting oxyhydride and insulating oxide films. In-situ illumination DB-PAS shows the irreversible formation predominantly of di-vacancies, as PALS showed that cation mono-vacancies are already abundantly present in the as-prepared films. The formation of di-vacancies supports conjectures that ${\mathrm{H}}^{\text{–}}$ (and/or ${\mathrm{O}}^{2\text{–}}$) ions become mobile upon illumination, as these will leave anion vacancies behind, some of which may subsequently cluster with cation vacancies present. In addition, in RE oxyhydride films, partially reversible shifts in the Doppler parameters are observed that correlate with the photochromic effect and point to the formation of metallic domains in the semiconducting films. Two processes are discussed that may explain the formation of these metallic domains and the changes in optical properties associated with the photochromic effect. The first process considers the reversible formation of metallic nanodomains with reduced O : H composition by transport of light-induced mobile hydrogen and local oxygen displacements. The second process considers metallic nanodomains resulting from the trapping of photoexcited electrons in an $e_g$ orbital at the yttrium ions surrounding positively charged hydrogen vacancies that are formed by light-induced removal of hydrogen atoms from octahedral sites. When a sufficiently large concentration, on the order of ∼10%, is reached in a certain domain of the film, band formation of the $e_g$ electrons may occur, leading to an Anderson-Mott insulator-metal transition like the case of yttrium trihydride in these domains.

Figure 1

PALS spectra collected at a positron implantation energy of 4 keV of (a) ${\mathrm{YH}}_{\sim 1.9}//\mathrm{Pd}$ with three-component best-fit analysis and (b) ${\mathrm{YH}}_x{\mathrm{O}}_y\text{−}1$ with four-component best-fit analysis using the poswin program. The peaks in the rectangular region in the spectra are excluded in the analysis since they correspond to backscattered positrons that annihilate at a different location in the sample chamber of the pulsed low-energy positron lifetime spectrometer (PLEPS).

Figure 2

S-W diagrams of Y- and Gd-based thin films. The Al-capped oxyhydride films are marked with blue color.

Figure 3

Normalized transmittance (averaged over the wavelength range of 450–1000 nm) for representative ${\mathrm{YH}}_x{\mathrm{O}}_y$ and $\mathrm{Gd}{\mathrm{H}}_x{\mathrm{O}}_y$ films during ultraviolet (UV) illumination (yellow regions) at 385 nm, followed by bleaching in the dark.

Figure 4

S- and W-parameter positron Doppler broadening (DB) depth profiles of ${\mathrm{YH}}_x{\mathrm{O}}_y//\mathrm{Al}$ (0.5 Pa), ${\mathrm{YH}}_x{\mathrm{O}}_y$ (0.5 Pa), $\mathrm{Gd}{\mathrm{H}}_x{\mathrm{O}}_y//\mathrm{Al}$ (0.9 Pa), and $\mathrm{Gd}{\mathrm{H}}_x{\mathrm{O}}_y$ (0.9 Pa) thin films before illumination (black symbols and lines) and after 2.5 h illumination and bleaching for ∼38 h (red circles and lines). The dashed lines indicate the positron implantation energy for the $in\text{−}situ$ illumination DB positron annihilation spectroscopy (DB-PAS) measurements shown in Fig. 5. The vertical solid lines indicate the boundaries between adjacent layers based on the vepfit analyses.

Figure 5

Time dependence of the S and W Doppler broadening (DB) parameters collected in in-situ DB positron annihilation spectroscopy (DB-PAS) measurements on ${\mathrm{YH}}_x{\mathrm{O}}_y//\mathrm{Al}$, uncapped ${\mathrm{YH}}_x{\mathrm{O}}_y$, $\mathrm{Gd}{\mathrm{H}}_x{\mathrm{O}}_y//\mathrm{Al}$, and uncapped $\mathrm{Gd}{\mathrm{H}}_x{\mathrm{O}}_y$ films before ultraviolet (UV) illumination (black symbols), under 2.5 h illumination (photodarkening, red triangles) and during ∼38 h of bleaching (blue circles) collected at positron implantation energies of 6.4, 5, 6.4, and 7.2 keV, respectively.

Figure 6

S and W values for ${\mathrm{YH}}_x{\mathrm{O}}_y//\mathrm{Al}$, uncapped ${\mathrm{YH}}_x{\mathrm{O}}_y$, $\mathrm{Gd}{\mathrm{H}}_x{\mathrm{O}}_y//\mathrm{Al}$, and uncapped $\mathrm{Gd}{\mathrm{H}}_x{\mathrm{O}}_y$ films before illumination (1, black symbols), upon illumination for 1.5–2.5 h (2, red triangles), and after illumination in the bleaching state (3, blue circles) according to the in-situ illumination Doppler broadening positron annihilation spectroscopy (DB-PAS) measurements compared with the (S,W) points of the as-deposited Y, Gd metal, ${\mathrm{YH}}_{\sim 1.9}$, $\mathrm{Gd}{\mathrm{H}}_{\sim 2}$ metal hydride, and ${\mathrm{YH}}_2{\mathrm{O}}_3$, ${\mathrm{Gd}}_2{\mathrm{O}}_3$ metal oxide films.

Figure 7

Schematic illustration of the proposed mechanisms 1 and 2.

Figure 8

Simulated positron annihilation fractions in H-rich domains as a function of average distance between the domains at positron diffusion lengths of 10, 20, and 40 nm, respectively. A volume fraction of 6 vol. % of spherically shaped domains was assumed.