Iron Enrichment at Grain Boundaries in SrTiO3: Does It Stay or Does It Go?

Le résumé fourni par la source

Grain boundary dopant adsorption is the positive or negative excess of atoms per unit area of unit interface area in comparison to the adjacent bulk phase. To study the segregation behavior of dopants in, for instance, ceramic microstructures it is of paramount importance to keep dopant concentrations below the respective bulk solubility limits. It is, however, often practiced to intentionally deposit dopants on the free surfaces of ceramic single crystals prior to their diffusion bonding. While this experimental process leads to grain boundary planes enriched with dopant atoms, but the observed interfacial dopant concentration may not reflect an equilibrium interface configuration. The latter is predicted by the Gibbs adsorption isotherm, which identifies the change of grain boundary energy γ with the amount of dopant excess Γ and chemical potential μ: In this study {100} twist grain boundaries in SrTiO3 with a nominal misorientation angle of 45° were diffusion bonded at 1200°C for 50hrs. A series of bicrystals were fabricated from either 0.01 wt% Fe-doped single crystals or undoped single crystals onto which a 0.5nm thick layer of iron was sputter deposited prior to diffusion bonding. After successful diffusion bonding one bicrystal that was formed after Fe deposition was thermally annealed for an additional 100hr. Transmission electron microscopy samples were subsequently prepared by focused ion-beam sectioning followed by Ar+ ion polishing with a Fischione Nanomill. Atomic interface structures were characterized by aberration-corrected scanning transmission electron microscopy (STEM) using the Jeol JEM 2100AC STEM instrument installed at UC Davis. Excess iron concentrations at the interface were determined by energy dispersive X-ray spectroscopy (EDXS) using the spatial difference technique [1] by solving the following equation: Sexcess is the interface specific EDXS signal, Sinterface is the EDXS intensity recorded from an area that contain the interface, and Sbulk is the EDXS signal recorded from an area that only contains bulk SrTiO3. A scaling parameter α=0.78 was estimated using previously described procedures [2] through consideration of the width of the interface of roughly 0.8nm (see images in Figure 1) and the size of the scanned areas. Figure 1 shows high-angle annular dark-field (HAADF) STEM images of the bicrystals that were diffusion bonded from Fe-doped single crystals (Fig. 1A) and from Fe-covered undoped single crystals (Fig. 1B). The bicrystal annealed for an additional 100hr is shown in Fig. 1C. All bicrystals exhibit the same interface width of roughly 0.8nm, which is consistent with undoped bicrystals with the same misorientation previously investigated [3]. Figure 2A shows an HAADF-STEM survey image that illustrates the scanned areas utilized for EDXS acquisition using the spatial difference technique. Figure 2B shows a direct comparison of the interface specific EDXS signals obtained from the three different bicrystals. The doped bicrystal revealed a Fe/Ti atomic ratio of 0.0264, while the bicrystal formed after Fe deposition revealed a significantly higher Fe/Ti atomic ratio of 1.105. After additional annealing for 100hrs the Fe/Ti atomic ratio decreased to 0.098. Thermal annealing for an additional 200hrs after diffusion bonding is currently underway and results will be discussed during the presentation. The preliminary results of this study demonstrate that dopant atom deposition on the free surfaces of single crystals prior to diffusion bonding can lead to interface concentrations that are not consistent with interface excess predicted by the Gibbs adsorption equation. Additional thermal annealing may be used to force “de-segregation” of dopants, i.e., diffusion into the bulk so that the dopant enrichment at the interface approaches the expected equilibrium adsorption. HAADF-STEM images of twist grain boundaries in SrTiO3 that were diffusion bonded from Fe-doped single crystals (A) or undoped single crystals that were deposited with metallic Fe layers prior to diffusion bonding. (A) HAADF-STEM survey image, (B) normalized EDXS spectra recorded from the bicrystals displayed in Figure 1.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.