3.1. Scanning Electron Microscopy (SEM)
Figure 1 illustrates the morphology of ZnO-BaTiO
3 based varistor doped with different contents of Pr
6O
11 and Co
3O
4. These SEM images show that the surface morphology for varistors with a different dispersion of dopant content induced varied average ZnO grain sizes. For comparison, the microstructure of ZnO-BaTiO
3 is depicted in
Figure 1a.
Figure 1a shows that the ceramics have a two-phase microstructure with ZnO grain as the primary phase and BaTiO
3 phase as the intergranular phase. These phases in the microstructure concurred with the observation by XRD analysis in
Figure 2a and
Figure 3a at 0.0 wt.% of Pr
6O
11 and Co
3O
4. The highest peaks match well with the ZnO phase, and the lowest peaks have been matched to the BaTiO
3-rich phase. A high proportion of BaTiO
3 rich phase shows that many BaTiO
3 particles were diffused into some ZnO grains to form an intergranular layer. Another amount segregates near the grain boundaries forms solid solutions. During the sintering process, a liquid phase occurred in the grain boundaries and formed a thick layer along the grain boundary and triple junctions of matrix grains. The remaining amount of solid solutions shows that many BaTiO
3 particles formed into melt and filled the grain boundaries with a liquid phase during sintering. When the grain boundaries are completely formed at the interface state, no more liquid phase can occur in the grain boundaries. It can be attributed to the precipitation of BaTiO
3-rich phase near the grain boundaries and nodal points. It is believed that an amount of BaTiO
3 phase remains at the grain boundaries of two differently matrix which can affect the grain boundaries resistivity. In general, the varistor microstructure is composed of ZnO grains and secondary phase, which usually located at the grain boundaries and junctions of multiple grains [
28].
Figure 1b–d show the microstructure of ZnO-BaTiO
3 varistor ceramics doped with 0.5 to 1.5 wt.% of Pr
6O
11. It can be seen that the addition of Pr
6O
11 increases the size and number of BaTiO
3 phase with a slight difference in the shape of the particles in comparison with the sample at 0.0 wt.% Pr
6O
11. The difference indicates that the precipitation of Pr
6O
11 phase located in the grain boundaries leading to the increase of BaTiO
3-Pr-rich phase in the form of solid solutions. The presence of BaTiO
3-Pr-rich phase can enhance the nonlinear electrical properties due to the increase in barrier height at the grain boundaries. This tendency can be observed in
Figure 1b–d, where the Pr species have been distributed in the grain boundary and at triple points along with the BaTiO
3-rich phase. Meanwhile, segregated Pr at the grain boundaries can increase the number of grain boundary that is required for improving the varistor breakdown voltage. By increasing Pr-doping amounts to 1.5, the varistor breakdown voltage of the ceramics gradually increased. This finding are in accordance with the current density-electric field (
J-E) curves where the varistor breakdown voltage (
Eb) of the sample evidently increased when the doping amount is increased up to 1.5. Since Pr
3+ (1.113 Å) are relatively larger ions than Zn
2+ (0.74 Å), the chance for Pr ions to dissolve in ZnO matrix grains is extremely small and induced the formation of Pr-rich phase by contributing more solid solutions near the grain boundaries and triple point junctions. The microstructures of the Pr
6O
11 doping system did not differ so much, but the grain size increases when doped with 0.5 wt.% of Pr
6O
11. The average grain size decreased from 2.0 to 1.6 µm by increasing the Pr
6O
11 doping amount up to 1.5 wt.%. It has been well accepted that the ZnO grain growth can be suppressed by excessively doping with Pr
6O
11 [
29].
The addition of Co
3O
4 dopant content enhanced the grain size of ZnO compared to Pr
6O
11-doped samples, whereas the average grain size increased from 1.5 to 2.5 µm with increasing cobalt dopant. As shown in
Figure 1e–g, the addition of Co
3O
4 encourages the growth of ZnO grains and promoted densification of varistors. It was observed that ceramics doped with Co
3O
4 contained low of solid solutions with more uniform grains. This effect is due to the solubility of Co ions in ZnO grain. The ionic radii of Co
3+ (0.65 Å) which is approximately the same as Zn ion and could be induced the solubility of Co ions in ZnO lattice. The incorporation of Co
3O
4 decreased the size and number of BaTiO
3 phase with more resolve BaTiO
3 particles and diffused into grain boundaries in comparison with that of 0.0 wt.% Co
3O
4-doped samples. The Co
3O
4 dopant has induced Co ions in the interior of ZnO grains and creates interface states and deep bulk traps that are required for the formation of the potential barrier at the grain boundary and enhance the nonlinearity coefficient of ZnO varistor ceramics [
30].
Figure 1e–g indicates that Co species have been diffused into grain boundaries and formed BaTiO
3-Co-rich phase. By increasing Co-doping amounts up to 1.5, reduced the BaTiO
3-rich phase due to the segregated of Co ions in the grain boundary and generated the interface traps. A part of Co
3O
4 dopant diffused into matrix grains by increasing densification and grain growth of ZnO ceramics during the sintering process. From these findings, the Co
3O
4 dopant has increased the barrier height while promoted grain growth and densification of ZnO varistor ceramics.
3.3. X-Ray Diffraction (XRD)
The XRD study was used to examine the different phases present in the prepared and sintered samples.
Figure 2 and
Figure 3 show the X-ray diffraction patterns of the as-prepared ZnO-BaTiO
3 based varistor doped with different contents of Pr
6O
11 and Co
3O
4 dopants. The XRD patterns revealed the presence of the ZnO main phase, according to the inorganic crystal structure database (ICSD) No. 067849, with a hexagonal wurtzite structure. The additives were observed due to the presence of the BaTiO
3 phase within the ZnO ceramics, as indicated by the ICSD No. 028851. XRD spectrum demonstrates that the presence of two major phases which were ZnO phase and BaTiO
3-rich phase that co-existed with Pr-rich phase. The addition of Pr
6O
11 dopant increases the number of intergranular phases and induced solid solutions near the grain boundaries. The diffraction peaks belong to the Pr-rich phase are located at (220) peak, and no secondary phase was observed, suggesting that the Pr ions may be doped into ZnO.
Figure 2b shows the magnified region of (220) peak, in which the diffraction angle of Pr-doped ZnO-BaTiO
3 ceramics with different content shifted to smaller angle side in comparison with that of BaTiO
3-doped ZnO ceramics. This shifting was due to the difference in ionic radii of Pr
3+ (1.113 Å) and Zn
2+ (0.74 Å) which cause to the change of ZnO lattice and the observation are also similar with Pr-doped ZnO ceramic systems reported in [
31]. Therefore, the peak shift in different Pr-doped ZnO-BaTiO
3, in this case, indicates that the doping element with a larger ionic radius than Zn
2+ is replaced at the substitution sites of the ZnO crystal lattice. Further increase of Pr-doping content up to 1.5, the peak position is shifted towards the larger 2θ side, which suggests that the Pr ion has occupied the interstitial sites of ZnO. In addition, the full width at half maximum (FWHM) of diffraction peaks increase with an increase of Pr doping level, which means that the crystallite size decreases with the increase of Pr content. The mean crystallite size can be calculated using Scherrer’s formula
where λ is the wavelength of X-ray radiation (0.15406 nm),
β is the full width at half maximum of a distinctive peak, and
θ is Bragg’s diffraction angle. The micro-strain (ε) can be calculated using formula
The calculated full width at half maximum (FWHM) value, the crystallite size (
D) and micro-strain (ε) of all samples was indicated in
Table 3. It can be seen that the crystallite size of ZnO decreased from 57.95 nm to 36.94 nm when Pr concentration increased from 0.5 wt.% to 1.5 wt.%. The data indicate that the presence of Pr
3+ ions in ZnO reduced the growth of crystal grains. The decrease in crystallite size is due to the compressive strain of crystal lattice. It is clear from
Table 3 that the micro-strain is increased from 20.90 × 10
−4 to 21.20 × 10
−4. The presence of Pr content in ZnO is able to change the reaction rate as well as the growth mechanism of crystal grains. The decrease in crystal size is well understood by the grain growth suppressing effect of ZnO by Pr-doping [
24]. The decrease in the XRD peak position is generally observed when the crystallite size increased and then increased due to the decrease in lattice size. With the addition of Pr
6O
11, the Pr
3+ ions may concentrate at the grain boundaries or on the surface of ZnO in the first segregation. Afterwards, an increase of Pr
6O
11 content leads to the formation of BaTiO
3-Pr-rich phase segregation locating on the surface of the sample and in the grain boundary near the junctions between ZnO grains, which will inhibit ZnO grains.
It was observed that the ceramics doped with Co
3O
4 contained two-phase microstructure with ZnO as the major phase and BaTiO
3 phase as the minor phase. Traces of Co-rich phases were also detected where some minor peaks had been matched to Co
3O
4. The formation of intergranular phase was also induced by Co-rich phase at the grain boundaries. However, the addition of Co
3O
4 decreased the number of observable intergranular phase and enhanced the grain growth compared to the sample doped with Pr
6O
11. In the enlarged view of the (331) peak in
Figure 3b, the peak gradually moves to a lower angle with the addition of Co
3O
4 dopant, indicating increased in crystallite size as a result of Zn
2+ (0.74 Å) with the smaller radius being substituted by Co
2+ (0.75 Å).
Table 4 shows the value of full width at half maximum (FWHM), the crystallite size (
D) and micro-strain (ε) with Co-doping content. With increasing Co content from 0.5 to 1.5 wt.%, the crystallite size is decreased from 36.89 to 35.11 nm. The decrease in peak position value in this case is also due to the increased in crystal size and the effect of tensile strain. Based on this evidence, the micro-strain is increased from 21.90 × 10
−4 to 24.10 × 10
−4 and exhibited a larger crystallite size in comparison with that of 0.0 wt.% Co
3O
4 addition. Moreover, the value of FWHM decreases with the addition of Co
3O
4, which indicates that the crystallite size increases with the addition of Co
3O
4 dopant. Then the 2
θ shifted to larger angle side for 1.5 wt.% Co-doped ZnO-BaTiO
3 ceramic materials are attributed to the existence of Co
3O
4 in ZnO. Therefore, the crystal grains decreased with a further increase of Co-doping content up to 1.5 wt.%.
Moreover, some minor peaks could be observed where their peaks are located in the closed angles of the ZnO and BaTiO
3 phases. The presence of a new phase might be due to the addition of Pr
6O
11 and Co
3O
4 dopants, but some phases could not be detected in all samples, which indicates that Co and Pr ions were completely diffused into ZnO and BaTiO
3 phases. The existence of these phases promotes the formation of intergranular layer and solid solutions at the ZnO grain and grain boundary, which also acts as a donor that forms the depletion layer or as an acceptor by dominating the grain boundary layer [
32,
33]. Furthermore, the presence of these solid solutions has significant effects on the nonlinear electrical properties of the varistor ceramics.
3.4. Current Density-Electric Field (J-E) Characteristics
A preliminary test of electrical properties on finalizing the composition of the ZnO-BaTiO
3 based varistor was characterized by nonlinearity measurement [
20]. The samples added with 12 wt.% BaTiO
3 showed a relatively higher barrier height than other samples with dissimilar BaTiO
3 contents; the result is consistent with the standard nonlinear coefficient characteristics. The presence of a potential barrier at the grain boundary caused the nonlinearity characteristics of the varistor [
34]. Therefore, the presence of grain boundary contributed to the nonlinearity of the varistor ceramics.
Figure 4 shows a significant difference in the
J-E characteristics of the varistor samples doping with various of Pr
6O
11. The conduction mechanism is divided into three regions: pre-breakdown region, breakdown region and upturn region.
In the pre-breakdown region, the linear curves of
J-E response for all samples indicate that the resistivity of the varistor ceramic samples before breakdown voltage is slightly higher than the nonlinear curves of
J-E response. The
J-E curves at different concentrations of Pr
6O
11 resemble a nonlinear response with a high resistivity region at the left of the breakdown voltage and a low resistivity region at the right of the breakdown voltage. The values of the nonlinear coefficient are decreased for Pr-doped samples, whereas the
J-E curves are shifted to the right of breakdown voltage by Pr. However, the most important parameter here is the breakdown voltage. By increasing the Pr-doping amounts to 1.5, the varistor breakdown field of the ceramic evidently increased. The breakdown field of the varistor ceramics is related to the ZnO grain size and the applied electric field at each grain boundary [
35,
36]. The decrease in the average grain size with further increase in the Pr-doping amount can be clarified by the increase in the numbers of grain boundaries between the electrodes, thereby increasing the breakdown voltage and enhanced the nonlinearity of the varistor ceramics. A high nonlinear coefficient leads to the increased breakdown field of the ceramic samples [
37]. The increase in the breakdown voltage with further increase in the Pr-doping content also reduced the leakage current, indicating that it occurs at lower current density due to the increase in barrier height at the grain boundaries. The nonlinear electrical properties of the Pr-added samples is lower than that of the BaTiO
3-doped ceramic for electric fields with strengths lower than 10 V/mm. It has been reported that most of the Pr
6O
11-doped ZnO varistor system exhibited low nonlinear coefficient value at a low-voltage range [
38].
It can be observed from
Figure 5 that the
J-E curves at two regions with the pre-breakdown region at the low electric field and breakdown area at the high electric field. The sharper the knee of
J-E curves between the two regions, the higher the nonlinearity coefficient. However, the varistor samples doping with Co
3O
4 exhibited higher breakdown voltage in the breakdown region or nonlinear region and showed higher nonlinearity coefficients in comparison with that of BaTiO
3-doped samples. The varistor’s breakdown field is directly proportional to the grain boundaries number per unit of thickness where the improvement in the nonlinear behavior of the samples is attributed to the formation of the potential barrier at the ZnO grain boundaries. By increasing Co-doping amounts to 1.0, the varistor breakdown voltage of the ceramic apparently decreased first and then gradually increased when the doping amount is increased up to 1.5. The
J-E curves with different concentrations of Co
3O
4 exhibited similar shapes at electric fields below 10 V/mm, however under the varistor voltage at 1 mA (V
1mA) the varistor samples doping with 0.5 wt.% showed higher breakdown voltage compared to the ceramics doping with 1.0 and 1.5 wt.%. At V
1mA, the varistor breakdown voltage obviously decreased when the Co-doping amount is increased up to 1.5, thereby decreasing the values of the nonlinear coefficient of the samples. The reduction in nonlinearity coefficient and breakdown voltage with further increase in the Co-doping content could be attributed to the formation of oxygen vacancies at the grain boundary [
39]. It was reported that the oxygen species could increase the density of interface states at the grain boundary and enhanced Schottky barrier height [
29].
With an increased amount of doped Co
3O
4, more Co ions diffused into the ZnO grain and grain boundary which can improve the characteristic of the varistor samples, including the increment of the breakdown voltage (V
1mA) to a small extent. However, the excessive Co ions enter into ZnO grains and grain boundaries with the BaTiO
3-rich phase could destroy the original microstructure due to the accumulation of excessive Co ions which depleted the O ions and promoted the formation of O vacancies in the grain boundary. As a result, the density of interface states decreased and reduced the interface barrier height, thus increase the leakage current density within the varistor samples. The electrical parameters for the current density-electric field (
J-E) of varistor with different amounts of Pr
6O
11 and Co
3O
4 are summarized in
Table 5 and
Table 6. The knee of the
J-E curve at 0.0 wt.% of Pr
6O
11 and Co
3O
4 only appeared when a lower electric field was applied as shown in
Figure 5, which indicates that the ceramic materials exhibited low breakdown voltage. This suggests that ZnO-BaTiO
3 varistor ceramics can provide excellent protection performance at low-varistor voltage.
Figure 4 shows that the
J-E curve at 0.0 wt.% is disappeared when a stronger electric field was applied. However, with the addition of Pr
6O
11 the knee of the curve is appeared when the electric field was increased, indicating an improvement in breakdown voltage of the materials. The enhancement in the nonlinear electrical performance of Pr-doped ZnO-BaTiO
3 varistor ceramics is suitable for the application in high-voltage varistor systems. Enhancing protection performance for multilayer ceramic varistors (MLCVs) is required since it shows excellent protection performance with low-voltage and small size [
8]. Therefore, they are mostly employed in mobile devices. The improvement in nonlinear electrical properties of Co-doped ZnO-BaTiO
3 varistor systems at low breakdown voltage can provide excellent protection performance for the application in chip varistors.
The relative density of Pr
6O
11 dopants was above 96% of the theoretical density of bulk ZnO (5.61 g/cm
3), and the average grain size was in the range of 1.6 to 2.0 µm. A significant change in the grain size can produce a dense ceramic body with better varistor properties. The varistor breakdown voltage (
Eb) of Pr
6O
11 dopants increased from 12.36 to 68.36 V/mm, which indicates that doping with Pr
6O
11 decreased grain growth. The increase of varistor voltage is due to the ionic radius of rare earth oxides (Pr
6O
11) is larger than that of Zn ions. During the sintering process, the interface layer is formed by the segregation of rare earth oxides into the ZnO grain boundary, resulting in the pinning effect and therefore inhibit the grain growth of ZnO varistors [
40].
Meanwhile, the leakage current density decreased with the increase of Pr
6O
11 doping amounts. This is due to the increase in the density of varistors, where the use of Pr
6O
11 doping had restrained the formation of secondary phases and enhanced the densification of the ceramic structure [
38]. However, the increases in the leakage current densities in the Pr
6O
11 doping samples compared with the undoped sample may be due to the accumulation of excessive dopants has increased the O vacancies and promote the formation of the depletion layer. It is well known that the O species can increase the density of interface states at the grain boundary, thus enhancing the potential barrier height [
39,
41]. The addition of Pr
6O
11 dopant leads to the inhibition of the potential barrier formation in the grain boundary of the varistor. In contrast, the excessive amount of dopant can cause a variation of the electronic state at the ZnO grain boundary and deteriorate the performance of varistor ceramics. Conclusively, the electrical characteristics of ZnO varistors are greatly enhanced with small amounts of rare earth dopants but critically deteriorated when excessive quantities of rare-earth dopants are used [
42]. These varistor formers change the property of ZnO varistors by varying the inter-composition ratio of dopants. Through such changes, it will result in a complication for the multi-layer structure of chip varistors by generating a large quantity of the secondary phases with an insulating spinel phase that played no significant electrical role. This phenomenon will then result in the degradation of varistor properties [
43].
For the varistors with Co
3O
4 dopants, the α value initially increased when Co
3O
4 doping amounts were increased up to 0.5 wt.%. Later, the α decreased when Co
3O
4 doping amount was increased up to 1.5 wt.%. The performance of varistors depends on the amount of secondary phase at the grain boundaries since the presence of the BaTiO
3-rich phase can increase the effective grain boundary region and improve its performance. The increase of dopant content has changed these varistor formers, which resulted in a complication for the grain boundary layer structure by generating a large number of secondary phases. It is known that the addition of transition metal oxides such as Co
3O
4 can generally improve the nonlinear coefficient at lower current density region by trapping the electrons and increasing the barrier height [
44]. The average relative densities of Co
3O
4 dopants exceeded 97% of theoretical density, and the average grain size increased from 1.7 to 2.5 µm as the Co
3O
4 content was increased from 0.5 to 1.5 wt.%, respectively. Due to the similar ionic radius of Co and Zn, a solid solution of Zn and Co is formed because of the complete substitution of Co in the Zn structure, which neglects the presence of pores. Moreover, when a sample was strongly doped with Co
3O
4, it can increase the defect concentration and the density, which leads to a decrease in the porosity [
45].
The varistor breakdown voltage initially increased when Co
3O
4 doping content was increased up to 0.5 wt.%. Then, when the doping level was increased up to 1.0 wt.%, it decreased. Also, the
Eb value decreased after the Co
3O
4 doping amount was increased up to 1.5 wt.%. The decrease of these
Eb can be clarified by the increase of the average grain size since both are correlated [
46]. The decrease of leakage current density (
JL) at 0.5 wt.% Co
3O
4 is due to the electrons which come from the grain are trapped by the defects or dopants at the grain boundary. Therefore, the high density of samples doped with Co
3O
4 enhanced the varistor performance by restricting the current flow using high resistance and reduced the leakage current density. By increasing the Co
3O
4 doping amount up to 1.5 wt.%, the
JL value was increased due to the presence of multiple defects at the grain boundary. The state also deteriorates the surge absorption capabilities by decreasing the number of active grain boundary region [
47]. The nonlinear electrical properties are related to the Schottky barrier height at the grain boundary. Therefore, the increase of α value signifies a higher barrier height and consequently reduced the leakage current density of the varistor.