Sol-Gel method was used to prepare the precursor of CaCu₃Ni _xTi₄O_{12+ x}, for x=0, 0.1, 0.2, 0.3 (samples labeled CB0, CB1, CB2 and CB3, respectively). Reagent-grade raw materials were calcium acetate [Ca(CH₃COO)₂•H₂O], copper nitrate [Cu(NO₃)₂•3H₂O], Ni(NO₃)₂ and tetrabutyl titanate [Ti(OC₄H₉)₄]. Firstly, an appropriate amount of Ca(CH₃COO)₂•H₂O, Cu(NO₃)₂•3H₂O, Ni(NO₃)₂ and Ti(OC₄H₉)₄ in the stoichiometric ratio were dissolved in ethanol separately with some glacial acetic acid. Secondly, proper amount of deionized water and little amount nitric acid were added when the sol was stirred. After continual stirring for about 20 min, a cyan precursor sol was obtained. And then, the precursor sol aged for 30 h. Thirdly, CCTO thin films were deposited on Si substrate by dip-coating technique. Finally, CCTO thin film samples were annealed at 900℃ for 1 h with a heating rate of 5 ℃/min.

For the characterization of electrical properties, the silver paste was coated on the face of samples, and the silver electrodes were formed by heating at 600℃ for 10 min. The silver electrodes were 5 mm in diameter. The voltage-current ( V- I) characteristics were measured using a V- I source/measure unit (CJP CJ1001). The nominal varistor voltages ( V_N) at 0.1 and 1.0 mA were measured and the threshold voltage was at V_T (V/mm) ( V_T = V_N (1 mA)/ d; d is the thickness of the sample in mm)^{[ 15, 16, 17]}. The leakage current ( I_L) was measured at 0.75 V_N (1 mA). The measurement accuracy for voltage was ± 0.5% and for electric current was ±2%^{[ 18, 19, 20, 21]}. In addition, the nonlinear coefficient α( α=1/lg( V_{1.0 mA}/ V_{0.1 mA})) was determined with relative error of ±5%^{[ 22, 23, 24]}. The crystalline phases were identified by X-ray diffraction (XRD, Rigaku D/max 2500, Japan) using a Cu Kα radiation. The dielectric characteristics, such as the apparent dielectric constant ( ε_r) and dissipation factor (tan δ) at different frequencies were measured by an HP4294A impedance analyzer (Agilent). The microstructures were evaluated by Atomic Force Microscope (SPM, SPA400, Seiko, Japan).

Figure 1 shows the AFM images of Ni-doped CCTO films with different Ni contents. It is apparent that the grain size of CCTO without Ni doped is larger than that with Ni doped. The average grain size decreases gradually with the increase of Ni content, With the Ni content increase, the roughness of film surface firstly increases and then decreases when x>0.1 (Table 1), which indicates that doping properly with NiO can further inhibit the growth of CCTO grains in the sintering process.

Fig. 1

	Figure Option ViewDownloadNew Window
	Fig. 1 AFM images of the surface of Ni-doped CCTO films

Table 1

Table 1

Table 1 Roughness characterization of the surface of Ni-doped CCTO films Samples CB0 CB1 CB2 CB3 Roughness, Ra/nm 65.1 141.3 76.1 10.7

Table 1 Roughness characterization of the surface of Ni-doped CCTO films

Figure 2 shows XRD patterns of the CCTO films with different dopant content. These patterns are essentially similar to that of undoped CCTO films. XRD data was indexed on the basis of a cubic unit cell which was similar to CCTO. The formation of CaCu₃Ti₄O₁₂ cubic phase belonging to the Im-3 space group (ICDD 75-2188) is apparent and reflections (220), (400) and (422) of CaCu₃Ti₄O₁₂ phase are identified for the Ni-doped CCTO films. Inaddition, the peaks of TiO₂and CaTiO₃ indicate that there are TiO₂ phase in all samples. That tetrabutyl titanate is not pure probably leads to the existence of TiO₂ phrase. The XRD peaks intensify with the increasing of Ni doping content, which suggest that dopant promotes the formation of the main crystalline phase of CCTO.

Fig. 2

	Figure Option ViewDownloadNew Window
	Fig. 2 XRD patterns of various Ni-doped CCTO film

Figure 3 illustrates the dielectric constant and dielectric loss in frequency ranging from 1 kHz to 10 MHz at room temperature. The ε_r of the pure CCTO is about 26000 at 1 kHz. The dielectric constant decreases firstly and then increases with the increase of the Ni doping content. When x=0.1 and 0.2, the dielectric constant of NiO-doped CCTO is smaller than CCTO without NiO. When x=0.3, the dielectric constant of NiO-doped CCTO is much higher than others. Furthermore, the dielectric constant decreases with the frequency increases. According to the previous work^{[ 7]}, the larger grain size will help to obtain a higher dielectric constant. In Fig. 3(b), the dielectric loss decreases firstly and then increases and the relaxation peak tends to move towards low frequency with the content of NiO increase. The dielectric loss in low frequency area is closely connected with leakage current of CCTO. The dielectric loss greatly increases with the content of NiO increase. Due to the addition of NiO, the carrier density in CCTO greatly increases, and the carrier in CCTO is electronic. Because Ni²⁺ replaces the position of Cu²⁺, the content of CuO decreases, which leads to significant increase of the electronic concentration.

Fig. 3

	Figure Option ViewDownloadNew Window
	Fig. 3 Frequency dependence of (a) dielectric constant and (b) dielectric loss in Ni-doped CCTO films measured at room temperature

XPS analysis was carried out to investigate the valence states of ions in CCTO films. Figure 4 shows the XPS spectra of O1s, Ca2p, Ti2p, Cu2p and Ni2p. The shift of core-level spectra due to charging effect is calibrated using the contaminated C1s peak located at 285 eV. O1s peak of CBO without Ni doped can be spitted into two peaks from Figure 4(a). One is due to Cu-O bonds and the other is due to Ca-O bonds. Ni-O bonds are found in the O1s of CB1, CB2, CB3 with Ni doped. The XPS spectra of Ca2p, Ti2p, Cu2p and Ni2p are shown respectively in Figure 4 (b), (c), (d) and (e). It reveals that the presence of calcium in +2 oxidation state is clear from the shape, the symmetry of the peak and the bonding energy of Ca2p in (b). The XPS patterns of Ti2p are shown in Figure 4 (c). It is clear that titanium is in the +4 state. The presence of titanium in +4 state is evident from the shape, the symmetry of the peak and the bonding energy of Ti2p. Small amounts of Cu⁺ are found in the Cu2p. The presence of copper in +1 and + 2 was clear from Figure 4 (d). Figure 4(e) shows the XPS spectra of Ni2p. It is revealed that the presence of nickel in +2 oxidation state is clear from the shape, the symmetry of the peak and the bonding energy of Ni2p in Figure 4 (e). And the graph of O1s in Figure 4 (a) presents that the nickel divalent combines with O ions to form the Ni-O bonds.

Fig. 4

	Figure Option ViewDownloadNew Window
	Fig. 4 XPS spectra of Ni-doped CCTO films(a) O1s; (b) Ca2p; (c) Ti2p; (d) Cu2p; (e) Ni2p

As can be seen from Table 2, Ni doping shows remarkable influences on the nonlinear current-voltage properties of CCTO samples. With increasing Ni content, the leakage current firstly increases and then decreases, while nonlinear coefficient and threshold voltage present an opposite tendency. When x=0.2, the lowest leakage current is obtained and the minimum value reaches to 0.564; meanwhile, the highest threshold voltage and nonlinear coefficient are 81 V/mm and 1.9, respectively. Sample CB2 displays good varistor properties which exhibits the lowest leakage current, and shows better nonlinear V- I characteristics than that of other samples. It can be accounted for that Ni doping increases CCTO grain boundary barrier, leads to nonlinear coefficient and threshold voltage rising up, and leakage current reduction. All things considered, the optimal varistor characteristic of the films was shown in the sample of x=0.2.

Table 2

Table 2

Table 2 Electrical properties of Ni-doped CCTO films Sample IL/mA VT/mm α CB0 0.620 69 1.8 CB1 0.633 56 1.6 CB2 0.564 81 1.9 CB3 0.646 64 1.5

Table 2 Electrical properties of Ni-doped CCTO films

The CaCu_{3- x}Ni _xTi₄O₁₂ ( x=0, 0.1, 0.2, 0.3) thin films were prepared by Sol-Gel method and dip-coating technique. The crystalline structure and the surface morphology of the films were markedly influenced by various NiO content. From AFM observations, the grain size of CCTO with Ni doped was smaller than that without Ni doped. In general, it is found that the CaCu_2.8Ni_0.2Ti₄O₁₂ (the sample with x=0.2) has the best properties with the highest nonlinear coefficient and threshold field, which must be accounted for that Ni doping increases CCTO grain boundary barrier, leading to nonlinear coefficient and threshold voltage rising up and leakage current reduction. The dielectric constant increases, when doping content reaches a certain level. The dielectric loss is decreased firstly and then increased. Owing to Ni does not reduce the dielectric loss, researchers can further reduce the loss to expect the promising application prospect of CCTO with NiO.