It has been proposed that Cu stoichiometry is related to the dielectric response 13 14 . Fang et al. 13 reported that Cu-excessive CCTO samples showed improved densification and dielectric behaviors. In the present work, Cu-excessive CCTO ceramics in a composition of CaCu_3.1Ti₄O_12.1 [CC3.1TO] were fabricated. Our studies indicate that this composition exhibited a good densification and dielectric response (data not shown). The samples were fabricated using the solid-state mixed oxide method. Reagent grade CaCO₃, CuO, and TiO₂ powders were used as starting materials. The mixture of these powders was ground for 24 h in ethanol using zirconia grinding media. The suspension was then dried and subsequently calcined at 900°C for 8 h with a heating rate of 5°C/min. The calcined CC3.1TO powders were mixed with (0.5, 1, and 2 vol.%) Al₂O₃ nanoparticles (40 nm average particle size) and 1% polyvinyl alcohol [PVA] binder and were ball-milled in ethanol for 12 h using the same method as mentioned earlier. The slurry was then dried and sieved to a fine powder. The mixed powders were uniaxially pressed into pellets at a pressure of 60 MPa. The PVA binder was burnt out at 550°C with a heating rate of 1°C/min. Finally, the pellets were sintered at 1,025°C for 6 h with a heating rate of 5°C/min. The sintered pellets were investigated for phase formation by X-ray diffraction [XRD]. Density of the sintered samples was measured using the Archimedes method with distilled water as the fluid medium. The microstructures of the sintered samples were characterized using a scanning electron microscope [SEM], and the average grain size was determined using the linear intercept method. For the electrical measurement, silver paste was applied to both sides of the circular faces of the ceramics, then dried at 600°C for 15 min, and cooled naturally to room temperature. The dielectric constant and dielectric loss were then measured using a LCZ meter. The mechanical properties (hardness) of various sintered samples were studied using a Knoop microhardness tester. Indentations were applied to the polished surfaces with 0.3- and 0.5-kg loads and with an indentation period of 15 s.

The XRD results for the sintered ceramics containing up to 2 vol.% Al₂O₃ are illustrated in Figure 1. All of the patterns were similar to the unmodified CCTO diffraction peaks and were consistent with the results reported previously 18 . The peaks of the second phases such as Cu₂O and CuO could not be observed in the XRD patterns 14 . Further, no peak was observed for the Al₂O₃ phase in any of the XRD patterns. This may be due to the amount of Al₂O₃ additive which was too little to be detected at the sensitivity level of the XRD instrument.

The plot of density as a function of Al₂O₃ volume fraction is shown in Figure 2. The density slightly increased with the increasing amounts of Al₂O₃ up to 0.5 vol.% and then decreased for the 2 vol.% sample. The reduction in density for the higher Al₂O₃ samples suggests that the sintering mechanism of the samples was not complete. To obtain the best densification for compositions > 0.5 vol.% Al₂O₃, higher sintering temperatures or longer soaking times would be required.

Figure 3 displays the SEM micrographs of the as-sintered surfaces of CC3.1TO-Al₂O₃ nanocomposites. An agglomeration of Al₂O₃ nanoparticles was not explicitly observed, implying that the processing method produced a reasonably uniform distribution of the nanoparticles in the matrix of the composites. The surfaces of the CC3.1TO samples showed a duplex microstructure consisting of coarse grains (average grain size of approximately 20 μm) and fine grains (average grain size of approximately 1 μm) located around the coarse grains. This characteristic indicates an abnormal grain growth in the microstructure of the samples. The formation of a copper oxide liquid phase (in Cu-excessive CCTO), as suggested by Kim et al. 19 , may be the main reason for the formation of abnormal grain growth since the present samples have a Cu-excessive CCTO composition. The liquid phase enhanced nucleation of abnormal grains and the abnormal grains were then formed after sintering. Similar results have been reported previously for Cu-excessive CCTO ceramics 14 20 . The average grain sizes of the coarse grains were found to decrease with the additive (e.g., average grain size of the coarse grains was approximately 3 μm for the 2 vol.% sample; Figure 3b). However, the average grain size of the fine grains remained unchanged for higher Al₂O₃ content samples. Overall, the average grain size, calculated from coarse and fine grains, decreased from approximately 7.5 μm for the unmodified sample to approximately 2.0 μm for the 2 vol.% sample (Figure 2). The decrease in the average grain size is most likely caused by the mismatch of the different components. Further, Al₂O₃ might segregate to the grain boundaries which could prevent grain boundary movement during the sintering process and, as a result, inhibit grain growth.

The Knoop hardness values of the samples as a function of Al₂O₃ content are illustrated in the inset of Figure 2. The Knoop hardness data reveal that the additive improved the hardness values. The maximum hardness value in this work was 9.8 GPa (for the 1 vol.% sample) which is comparable to the value reported by Puchmark et al. for the PZT-Al₂O₃ nanocomposites. The improvement in the mechanical properties is most likely due to the nanoparticles reinforcing the grain boundaries and acting as effective pins against microcrack propagation 21 . Moreover, the enhancement of hardness can be related to the reduction in grain size, i.e., small grain size samples gave a higher measured hardness.

Figure 4 shows the dielectric constants versus the frequency at room temperature for the CC3.1TO and CC3.1TO-Al₂O₃ pellets. Compared with the CC3.1TO sample, a significant improvement in the dielectric constant of the CC3.1TO-Al₂O₃ samples was observed. For the CC3.1TO sample, the dielectric constant was 11,000 (measured at 1 kHz and at room temperature) which is close to the values reported previously 1 2 . The CC3.1TO sample also exhibited nearly dielectric-frequency independence over the frequency range of 0.1 to 500 kHz. Further, the dielectric constant increased reaching a value > 60,000 at 1 kHz for the 0.5 vol.% sample then decreased for further increases in the Al₂O₃ content. For the 2 vol.% sample, however, the dielectric-frequency curve showed a weak frequency dispersion of the dielectric constant. The reduction in the dielectric constant for the samples which were doped with more than 0.5 vol.% Al₂O₃ may be due to the fact that composites with higher additive amounts (Al₂O₃ > 0.5 vol.%) also had higher structural heterogeneity. Moreover, the formation of an impurity phase may have caused a reaction between Al₂O₃ and CC3.1TO which could not be detected using the XRD technique 22 , but it might also have contributed to the reduction in the dielectric constant. Plots of the loss tangent versus the frequency of the CC3.1TO and CC3.1TO-Al₂O₃ pellets at room temperature are presented in the inset of Figure 4. The loss tangent-frequency curve of the CC3.1TO ceramic exhibited a weak frequency dispersion for a narrow frequency range (1 to 50 kHz). However, after adding the additive, the loss tangent significantly decreased at low frequencies (< 1 kHz) which resulted in a wider range of frequency stability (10 Hz to 10 kHz). Further, the loss tangent decreased from 0.21 for the CC3.1TO ceramics to 0.09 for the 0.5 vol.% sample (at 1 kHz). A further slight decrease in the loss tangent was observed for additional additive amounts. The decrease in loss tangent shows a reduced conductivity of the CC3.1TO-Al₂O₃ samples. This result could be related to changes in the transport behavior due to an increase in resistivity at the grain boundaries where the additive nanoparticles predominantly segregated.

Figure 5 shows the dielectric constant values as a function of temperature at various frequencies for the CC3.1TO and CC3.1TO-Al₂O₃ pellets. The CC3.1TO sample showed a high dielectric constant (ε_r approximately 10,000) with temperature and frequency stability from room temperature to 60°C. After adding the additive, however, a significant improvement in the dielectric behavior was observed. The 0.5 vol.% sample showed a very high dielectric constant > 60,000 (at 1 kHz) which was nearly temperature-independent for the temperature range of approximately 35°C to 110°C. Compared to the CC3.1TO sample, this sample also displayed a pronounced frequency dependence of the dielectric constant especially for temperatures < 125°C. Moreover, the dielectric-temperature curve of the 0.5 vol.% presented a broad flat curve at high frequencies (> 10 kHz). For higher additive amounts (Al₂O₃ > 0.5 vol.%), the dielectric constant decreased with the increasing additives. Further, the dielectric frequency dispersion for the Al₂O₃ nanoparticle sample with 2 vol.% was not as strong for temperatures < 112°C. The loss tangent values as a function of temperature at various frequencies for the samples are illustrated in the insets of Figure 5. For the CC3.1TO sample, the loss tangent value was 0.21 and was stable with temperature as well as frequency from room temperature to approximately 41°C. The 0.5 vol.% sample had a loss tangent value lower than 0.10 for room temperature to approximately 50°C. From the (IBLC) model, the apparent dielectric constant ( ε r ′ ) can be related to the microstructure parameters by the formula 23 :

ε r ′ = ε gb   d t ,

where d is the grain size, t is the thickness of the grain boundary (barrier width), and ε _gb is the internal dielectric constant of the barrier material. Since the grain size of the present samples decreased with the increasing additive, Equation 1 predicts that the higher dielectric constant for the 0.5 vol.% sample is not related to the grain size, but it may be connected to a change in the grain boundary characteristics such as ε _gb and t after adding the additive. The reason for the change of grain boundary characteristic is still unclear, but it is possible that the Al₂O₃ nanoparticles had a reaction with the matrix of the CC3.1TO, and as a result, the formation of Al-metal oxide phases at the grain boundary produced other products in small amounts which could not be detect by XRD 22 . However, the higher density for the 0.5 vol.% sample can be explained by the observed higher dielectric constant in the present work.