Using a spherical aberration (Cs)-corrected scanning transmission electron microscopy (STEM) and electron energy-loss spectroscopy (EELS), we investigated a 6° low-angle [001] tilt grain boundary in SrTiO3. The enhanced spatial resolution of the aberration corrector leads to the observation of a number of structural variations in the edge dislocations along the grain boundary that neither resemble the standard edge dislocations nor partial dislocations for SrTiO3. Although there appear to be many variants in the structure that can be interpreted as compositional effects, three main classes of core structure are found to be prominent. From EELS analysis, these classifications seem to be related to Sr deficiencies, with the final variety of the cores being consistent with an embedded TiOx rocksalt-like structure.

Extended abstract of a paper presented at Microscopy and Microanalysis 2005 in Honolulu, Hawaii, USA, July 31--August 4, 2005

The properties of ceramic oxides being developed for such varied applications as fuel cells, ionic transporting membranes, high-Tc superconductors, ferroelectrics and varistors are dominated by the presence of grain boundaries. Key to controlling the electronic properties of the grain boundaries in these materials is a fundamental understanding of the complex relationship between structure, composition and local electronic structure. The ability to characterize and directly correlate these parameters on the atomic scale is afforded by the combination of Z-contrast imaging and electron energy loss spectroscopy (EELS) in the scanning transmission electron microscope (STEM). Furthermore, the recent development of in-situ heating capabilities in the JEOL 201 OF STEM/TEM permits atomic resolution analysis to be performed at elevated temperatures and the interactions of grain boundaries with the oxygen vacancies determined.

Figure 1 shows an example of the type of experiment that can be performed using these methods.

Grain boundaries have long been known to have a deleterious effect on the superconducting critical current that can be carried by YBa2Cu3O7-δ. Recent theoretical analyses have proposed that the origin of this behavior may be band bending, which results in the depletion of charge carriers at the grain boundaries. For this to occur in these p-type superconductors there must, by definition, be a high density of localized donor states in the boundary plane. Here we describe a structural feature intrinsic to all [001] tilt boundaries that may be the origin of these localized states.

Direct atomic-resolution images of asymmetric [001] tilt grain boundary structures have been obtained using the Z-contrast imaging technique. The grain boundaries are observed to be composed of distinct structural units, as shown in figure 1. Within these structural units, a 2×1 reconstruction of the CuO columns is seen to occur. This reconstruction, which is caused by the constraint imposed on the structure by the sizes of the component atoms, leads to effective oxygen vacancies in the grain boundary plane (figure 2). Further oxygen annealing of the boundary cannot fill these vacancies as there is no space in the structure for more oxygen atoms.

Recent theoretical analyses [1] have suggested that the origin of the reduced Jc at grain boundaries in high-Tc superconductors may be band bending, which results in the depletion of mobile charge carriers at the boundaries. For this to occur in these p-type superconductors there must, by definition, be a high density of localized donor states in the boundary plane. Here we report a “structural unit” analysis of [0011 tilt grain boundaries in YBa2Cu 3O7-δ (YBCO) that indicates there may be a simple structural origin for such localized donor states.

Due to the extemely short coherence lengths of the high-Tc superconductors (around 30 Å in the a-b plane), defects such as grain boundaries are obvious barriers to the flow of supercurrent. Within a few months of the discovery of these materials, it was shown how the critical current dropped four orders of magnitude as the grain boundary misorientaion increased from zero to 45°. Even today, there is no quantitative understanding of this behavior. A qualitative understanding is however possible through atomic resolution Z-contrast imaging on YBa2cu3O7-δ and SrTiO3 bicrystal grain boundaries, combined with bond-valence-sum analysis.

The Z-contrast image of a YBa2cu3O7-δ low angle grain boundary in Fig. 1 shows the same kind of reconstructed dislocation cores as seen in SrTiO3, containing reconstructions on both the Cu and Y/Ba sublattices.

Although grain boundaries are known to dominate the bulk properties of many technologically important materials, in most cases there is no fundamental atomic scale understanding of why they should have such an effect. One of the problems in developing this understanding is that conventional structure determination techniques, such as phase contrast imaging in TEM or Z-contrast imaging in STEM, produce only a 2-dimensional projection of the crystal structure. Atomic scale compositional changes must be simulated and a unique interpretation is clouded by boundary reconstructions and strain effects. Furthermore, neither technique provides any information on the local changes in the electronic structure that are critical for both the electrical and mechanical properties of the boundary.

EELS provides a means to quantify local changes in both composition and electronic structure. However, without a knowledge of the structure, interpretation of any observed changes at grain boundaries is extremely difficult.

Although grain boundaries typically account for only a small fraction of a material, they can have far reaching effects on the overall bulk scale properties. These effects are usually simply linked to the boundary having a different atomic arrangement to the bulk. A necessary first step in understanding the structure-property relationships is therefore a detailed determination of the boundary structure.

One means of obtaining detailed information on the structure of grain boundaries is through correlated Z-contrast imaging and electron energy loss spectroscopy (EELS). The Z-contrast image generates a map of the grain boundary which can be used to position the probe in defined locations for spectroscopy. In the case of oxides, a structural model of the metal atom positions can be determined directly from the image. Furthermore, using a simple bond-valence sum minimization routine, the oxygen atoms can be placed so that the structure contains atoms that have valences consistent with their expected formal valence state.

The short coherence length in high-Tc superconductors (5-15Å) makes an atomic scale understanding of the electronic properties at defects and interfaces essential for device applications. This understanding is particularly relevant for grain boundaries in YBa2CU3O7-δ (YBCO), where although extensive studies have shown a clear exponential decrease in critical current with misorientation angle, the absolute value can vary by several orders of magnitude at any given misorientation angle.

Figure 1 shows Z-contrast images of an [001] low-angle tilt boundary and a 30° [001] asymmetric tilt grain boundary. An interesting feature of both of these boundaries is that there appear to be sites where two atom columns are too close together. However, the problem of like-ion repulsion can be avoided if the columns are taken to be partially occupied. Insight into the effect of this partial occupancy can be obtained through the use of bond-valence sum analysis. Here, the formal valence of an atom is made up of contributions from all of its nearest neighbors, the magnitude of which are determined by the bond length.

Electron energy loss spectroscopy (EELS) is ideally suited for the study of oxide materials. The large cross section for the oxygen K-absorption edge generates enough signal for fine-structure analysis even when the probe is small enough for atomic resolution imaging (∼2Å). Therefore, by using the Z-contrast imaging technique in the STEM to locate defects and position the probe with atomic resolution, detailed spectroscopic information can be obtained from a known atomic location. by selecting the appropriate acquisition conditions, these spectra can have the same atomic resolution and incoherent features as the Z-contrast image. The primary benefit of this methodology is that the image provides a structural model for interpretation of the spectrum through theoretical modeling.

One means of interpreting this spectral fine-structure is through multiple scattering analysis. In the multiple scattering formulation, the fine-structure of the absorption edge is considered to arise from interference effects resulting from the scattering of an excited photoelectron from neighboring atoms.