Defect structure of the high-dielectric-constant perovskite $\mathrm{Ca}{\mathrm{Cu}}_3{\mathrm{Ti}}_4{\mathrm{O}}_{12}$

Using transmission electron microscopy (TEM) we studied $\mathrm{Ca}{\mathrm{Cu}}_3{\mathrm{Ti}}_4{\mathrm{O}}_{12}$, an intriguing material that exhibits a huge dielectric response, up to kilohertz frequencies, over a wide range of temperature. Neither in single crystals, nor in polycrystalline samples, including sintered bulk and thin films, did we observe the twin domains suggested in the literature. Nevertheless, in the single crystals, we saw a very high density of dislocations with a Burger vector of [110], as well as regions with cation disorder and planar defects with a displacement vector $\frac{1}{4}\left[110\right]$. In the polycrystalline samples, we observed many grain boundaries with oxygen deficiency, in comparison with the grain interior. The defect-related structural disorders and inhomogeneity, serving as an internal barrier layer capacitance in a semiconducting matrix, might explain the very large dielectric response of the material. Our TEM study of the structure defects in $\mathrm{Ca}{\mathrm{Cu}}_3{\mathrm{Ti}}_4{\mathrm{O}}_{12}$ supports a recently proposed morphological model with percolating conducting regions and blocking regions.

Figure 1

Schematics of the five possible orientation domains in CCTO: the twin boundary (denoted as TB: II–IV), antiphase boundary (denoted as APB: I, II; II, III), and combinations of twin and antiphase boundaries (I–IV; III, IV) projected along the [001] direction. The squares represent the unit cell with the diamond showing its twofold symmetry. The detailed structure of the same projection with $\mathrm{Ti}{\mathrm{O}}_6$ octahedra and $\mathrm{Cu}{\mathrm{O}}_4$ squares is included at the bottom left. The $\mathrm{Cu}{\mathrm{O}}_4$ square viewed end on is shown as a double line.

Figure 2

Experimental (a) and calculated (b) convergent beam electron diffraction pattern (convergent angle: $1.64\mathrm{mrad}$) of ${\left(001\right)}^{*}$ zone showing a twofold symmetry in the crystal.

Figure 3

A calculated coherent PARODI pattern showing the interference fringes of the (110) twin boundary and its adjacent twin domains in the ${\left(001\right)}^{*}$ zone orientation. Note, the vertical fringes originate from the twin boundary, and the circular fringes from the condenser aperture that defines the edge of the disks. The twin boundary, a 90° twinning (II–IV in Fig. 1), is clearly visible in shadow images of the (310)-type reflections due to the difference in their structure factors $(F_{-310}=F_{{-310}_T}=-10.18,F_{-130}=F_{{-130}_T}=17.23)$, while invisible in the (220)-type reflections $(F_{220}=F_{-220}=F_{-2–{20}_T}=F_{{-220}_T}=45.7)$.

Figure 4

(a) A typical bright-field image showing a high density of dislocations observed in the single crystal. (b)–(d) diffraction contrast of the same two dislocations imaged using $\mathbf{g}=020$ (b), $\mathbf{g}=-101$ (c), and $\mathbf{g}=-110$ (d) reflections. With $\mathbf{g}=-110$, the dislocations are out of contrast.

Figure 5

(a) A high-resolution image viewed along the [111] direction in single crystal CCTO showing the loss of threefold symmetry, evident from the Fourier analysis of the image, with only one of the three (110) reflections present [$(-110)$ in the left region and $(-101)$ in the right]. (b) Another area in the same viewing direction showing the presence of only the 01-1 reflection. The white lines indicate a lattice shift in the region. The displacement associated with the loss of symmetry is most visible when the images are viewed along a glancing angle. (c) Selected area diffraction from a much larger area ($1\mu \mathrm{m}$ across) showing the regional loss of the CCTO symmetry along the [01-1] direction. The diffraction pattern from an undistorted CCTO in a relatively thin area is shown in (d) for reference.

Figure 6

(a) A planar defect observed in the single crystal. (b) The high-resolution image of the same defect in (a) showing a lattice shift $(R=\frac{1}{4}\left[110\right])$ adjacent to the defect. (c) A structural mode in a [001] projection illustrates the possible origin of the planar defect with a lattice shift of $R=\frac{1}{4}\left[110\right]$ caused by $\mathrm{Ca}∕\mathrm{Cu}$ cation disorder (between the $A$ atoms) and the shift of the $A$ and $B$ (Ti atoms) site. To fit the $\mathrm{Ti}{\mathrm{O}}_6$ octahedra across the interface, a small displacement $(d=0.0603[-110])$ perpendicular to the interface has to be introduced. The parallel lines indicate the $\frac{1}{4}\left[110\right]$ lattice displacement.

Figure 7

Low-temperature electron energy loss spectra recorded at $86\mathrm{K}$ from the polycrystalline sample. From left: low loss region, $L$ edge of Ca, $L$ edge of Ti, $K$ edge of O, and $L$ edge of Cu.

Figure 8

A pair of electron energy-loss spectra of the Ti $L_{2,3}$ and O $K$ edge acquired from a grain boundary and away from the boundary at room temperature after subtracting the background and correcting for multiple scattering, showing a low $\mathrm{O}∕\mathrm{Ti}$ ratio at the grain boundary.