Probing dopant segregation in distinct cation sites at perovskite oxide polycrystal interfaces

Although theoretical studies and experimental investigations have demonstrated the presence of space-charge-induced dopant segregation, most work has been confined largely to the crystal-free surface and some special grain boundaries, and to the best of our knowledge there has been no systematic comparison to understand how the segregation varies at different types of interfaces in polycrystals. Here, through atomic-column resolved scanning transmission electron microscopy in real polycrystalline samples, we directly elucidate the space-charge segregation features at five distinct types of interfaces in an ABO3 perovskite oxide doped with A- and B-site donors. A series of observations reveals that both the interfacial atomic structure and the subsequent segregation behaviour are invariant regardless of the interface type. The findings in this study thus suggest that the electrostatic potential variation by the interface excess charge and compensating space charge provides a crucial contribution to determining not only the distribution of dopants but also the interfacial structure in oxides.

Analysis results from the electron backscatter diffraction (EBSD). a The crystallographic orientation map for grains is shown in color in the upper row, demonstrating that each grain has no preferential orientation. The grain-boundary misorientation angles obtained from the EBSD are also denoted by three different colors (red, green, and blue) on the SEM image in the lower row. As clarified in the summary, more than 90% of the grain boundaries are high-angle (>15°) boundaries (blue lines). b The random orientation of grains can be confirmed by this pole figure, which reveals no preferred orientation distribution.

Ca
Ti La Intensity (a.u.) Fig. 6 A set of EDS maps for Type-I free surface and comparison of EDS spectra acquired in a La-doped sample. a As denoted by a white arrow in the Ca map, the top-most surface is chemically verified to consist of the A-site cations including Ca. The blue dashed line denotes the La segregation layer corresponding to A-site-cation columns. b A typical EDS spectrum acquired in the La-doped sample is provided to show each of the X-ray signal peaks used for atomic-scale mapping. As the La-L α peak is quite small and close to the Ti-K α peak in the gray shadow, it is not clearly visible in this overview spectrum. c Although the positions of the La-L α (4.65 keV) and Ti-K α (4.5 keV) peaks are fairly close to each other, the energy resolution (~0.1 keV) of the EDS detector is sufficient to quantify the intensity of the La-L α peak. As denoted by red curves on each of the spectra, a higher intensity of the La-L α peak can be identified in the EDS spectrum from the surface region than from the bulk, verifying the segregation of La.  In this Note, we deal with a MO-type oxide composed of divalent metal cations, M 2+ as a model system. It is also assumed that a pristine MO crystal is of one dimension having a free surface at x = 0 and also the predominant type of lattice defects is Schottky disorder (a pair of M and O vacancies, ″ and O •• in the Kröger-Vink notation) with no preferential defect binding for simplicity. The following relationship thus is given in MO in general.
The bulk value of φ ∞ far from the surface can be readily calculated (Eq. 10) by using the charge neutral condition,

Addition of aliovalent dopants
As can be seen in Eq. 4 above, the electrical potential, φ ∞ , in the bulk of a pristine Consequently, a very large potential of φ ∞ with a positive sign is constructed in the bulk and thereby rapidly decaying φ( ) in the space-charge layer is clearly described in the right-hand diagram in Supplementary Fig. 1a. As recognized in Eq. 1, the resultant [ ″ ]( ) beneath the surface should be thus significantly reduced and the strong accumulation of positively charged M • is accompanied at the same time. The illustration on the left-hand side in Supplementary Fig. 1a depicts such circumstances, demonstrating the space-charge segregation of the donor.
The electrical potential in acceptor doping is also derived in the same manner. For example, when Na + is added as an acceptor dopant in MO and the ionic compensation by oxygen vacancies is assumed , the electrical potential in the bulk is given as In contrast to the donor doping case, a significantly large potential of φ ∞ with a negative sign is verified, resulting in acceptor segregation, as distinctively described in Supplementary Fig. 1b where and are the concentration and the effective charge of a dopant, respectively.