Grain Growth Control of ZnO-V2O5-Cr2O3 Based Varistors by PrMnO3 Addition


 In this work, the grain growth behaviour of ZnO+V2O5(1 mol%)+Cr2O3(0.35 mol%)-based ceramics with 0.25–0.75 mol% additions of PrMnO3 was systematically investigated during sintering from 850°C to 925°C with the aim to control the ZnO grain size for their application as varistors. It was found that with the increased addition of PrMnO3, not only did the average grain size decrease, but the grain size distribution also narrowed and eventually changed from a bimodal to unimodal distribution after a 0.75 mol% PrMnO3 addition. The grain growth control was achieved by a pinning effect of the secondary ZnCr2O4 and PrVO4 phases at the ZnO grain boundaries. The apparent activation energy of the ZnO grain growth in these ceramics was found to increase with increased additions of PrVO4; hence, the observed reduction in the ZnO grain sizes.


Introduction
ZnO-Bi2O3 ceramics are an established class of ZnO-based varistors.It is well known that the nonlinear current-voltage (I-V) characteristics of these ZnO-Bi2O3 varistors are directly dependent on their microstructures, mainly the average grain size and size distribution of ZnO.Typically, a large average grain size of >30 μm is required for low-voltage applications, and a small average grain size of <10 μm is needed for high-voltage applications [1].Additionally, a narrow grain size distribution is also critical for achieving stability of the electrical field strength of these materials [2].
It has been found that ZnO-V2O5 ceramics also exhibit a nonlinear I-V behaviour comparable to ZnO-Bi2O3 ceramics [3,4] and thus have the potential to be a new class of varistors.One of the significant advantages of ZnO-V2O5 ceramics is that they can be sintered at a relatively low temperature of ~900°C [3][4][5][6].This outstanding feature allows these ceramics to be co-fired with Ag (m.p. 961°C).This enables Ag, instead of expensive Pd or Pt, to be used as inner electrodes for applications in multilayer chip components [7].Moreover, V2O5 is also a better sintering aid compared to Bi2O3 for ZnO, enabling ZnO-V2O5-based ceramics to be densified to the same density at a lower temperature than ZnO-Bi2O3-based ceramics [8,9].However, ZnO-V2O5-based ceramics suffer from a distinct disadvantage in that they have been shown to exhibit abnormal ZnO grain growth [10,11].This is commonly attributed to the high reactivity of the V2O5-rich liquid phase formed during sintering.This phase assists in the diffusion of Zn 2+ and thus promotes ZnO grain growth, which results in the formation of abnormally grown grains (AGG) of ZnO.
One approach to curb the formation of AGG in ZnO-V2O5 ceramics is to induce the formation of a secondary phase to hinder ZnO grain growth by the addition of a third oxide to the system [12,13].The two commonly used additives are Cr2O3 and Sb2O3, as Cr2O3 can react with ZnO to form a ZnCr2O4 spinel phase, while Sb2O3 and ZnO form spinel phases of ZnSb2O6 and Zn2.33Sb0.67O4[12][13][14].However, these two additives have their own specific limitations.With Cr2O3, an addition greater than 1 mol% is required to effectively counter the AGG and refine the ZnO grain size in ZnO-V2O5 ceramics [13].However, at this required addition level, Cr2O3 has also been found to segregate at grain-boundary regions, leading to the deterioration of I-V behaviour due to high leakage current [12].With Sb2O3, an addition of up to 2 mol% is needed to control the grain growth of ZnO-V2O5; however, the addition of Sb2O3 increases the required sintering temperature for the system.At an Sb2O3 content greater than 0.5 mol%, the sintering temperature could increase to 1200°C [13], nullifying the advantage of having the ZnO-V2O5-based system in the first place.This work investigated the effect of PrMnO3 addition on the grain growth of ZnO-V2O5-Cr2O3 ceramics with the aim to control the average grain size and size distribution of ZnO to improve their electrical properties as practical varistors.

Sample preparation
All the reagents used in this work were of analytical grade (>99.5% purity) and were supplied by SINOPHARM.PrMnO3 powder was prepared in-house by mixing Pr6O11 and MnCO3 at a molar ratio of 1:6 and allowing the mixture to react at 1000°C for 5 h in air.The formation of PrMnO3 phase was confirmed by X-ray powder diffraction (XRD) analysis.
Table 1 summarises the nominal compositions of constituent powders used to prepare four ZnO-V2O5-Cr2O3 ceramic samples and their designated sample names.The PrMnO3 powder was introduced with the aim to regulate the grain growth in these ceramics.For each nominal composition, the powder mixture was homogenized by ball milling in absolute alcohol using zirconia balls in a polypropylene container on a planetary mill for 24 h.After milling, the powder mixture was dried at 80°C for 24 h.The dried powder mixture was then mixed with 5 wt% polyvinyl alcohol (PVA) binder and pressed into pellets of Φ12 mm × 1 mm under a uniaxial pressure of 130 MPa.The green pellets were first fired at 500°C for 1 h to remove the binder before being subjected to different sintering temperature/time regimes in an alumina crucible at a heating rate of 4°C/min.In this work, seven sintered ceramic discs for each nominal composition were produced at 850°C, 900°C, and 925°C for 4 h and at 875°C for 2, 4, 6, and 8 h.After sintering, the furnace was powered off to allow the sintered discs to cool down naturally within the furnace.

Sample characterization
The phase compositions of the sintered and quenched samples were analysed by XRD patterns obtained using a diffractometer (PANalytical X'pert Powder) with Cu Kα1 radiation (λ = 0.1541 nm).
The microstructure of the sintered samples was observed and analysed by scanning electron microscopy (SEM) using a ZEISS Supra 55 electron microscope.A polished cross-sectional area along the thickness of the sintered disc was prepared and chemically etched in a dilute HCl solution to reveal the grain and phase structures for SEM observation.
The average ZnO grain size and grain size distribution were analysed using a representative SEM image of each sintered sample.
To determine the average grain size, the method reported by Mendelson [16] was used.
Basically, the average grain-boundary intercept length ( ) was first determined by measuring along 10 randomly drawn lines across the SEM image.The average grain size G is then calculated by Eq. ( 1): To determine the ZnO grain size distribution, the dimensions around 500-800 ZnO grains of each sample were measured from its SEM images using a dedicated image analysis software.The surface area (S) of each grain was then estimated, and the equivalent diameter (d) of each grain was obtained by transforming the surface area of the irregularly shaped grain into a circle of the same area using the method reported by Daneu [17].
The grain growth behaviour was analysed using a phenomenological kinetic grain growth equation established by Nicholson et al. [18], which has been widely used for ZnO-based ceramic systems [8,10,18,19].The equation is expressed as follows: where G is the average grain size of the ZnO ceramic at time t, G0 is the initial grain size of the ZnO powder, n is the kinetic grain growth exponent, Q is the apparent activation energy, R is the universal gas constant, T is the absolute temperature, and K0 is a T-independent constant.In the case where G0 is significantly smaller than G, then Eq. ( 2) is simplified to ZnO-V2O5-Cr2O3 samples studied in this work.

Sample I-V characteristics
The DC current-voltage (-) behaviour of the sintered samples was measured using a withstand voltage tester (MS2671A, Xi'an Instruments Inc., China).The sintered discs (~Φ8 mm × 1 mm thickness) were painted with a silver paste on both surfaces to enable the measurements.The − curve was determined by measuring the current at a stepwise-increased applied voltage until reached 10 mA.The field strength ( ) is defined as the applied voltage per sample thickness, and the current density ( ) is the measured current per sample area.The switching field strength E1 mA/cm 2 is the field strength at J=1 mA .cm −2 .The nonlinear coefficient (α) is calculated by Eq. ( 4): where 10 • −2 and 1 • −2 are the field strengths at 10 mA .cm -2 and 1 mA .cm -2 , respectively.The higher the value of is, the better it is for a varistor.

Phase compositions and distributions in sintered samples
Fig. 1 shows the XRD patterns of the four ZVCP samples sintered at 875 °C for 4 h.For the ZVC sample, in addition to the main ZnO phase, several minor phases were also detected, including Zn4V2O9, α-Zn3(VO4)2, and ZnCr2O4.These minor phases have been commonly observed and reported in sintered ZnO-V2O5-Cr2O3 systems [12].With the addition of PrMnO3, a new minor phase of PrVO4 was also observed in the ZVCP25, ZVCP50, and ZVCP75 samples.
The formation of the PrVO4 phase is believed to resulte from the reaction between α-Zn3(VO4)2 and PrMnO3 during sintering following the reaction: This has also been reported by Zhao et al. [20].In some cases, ZnO grains were observed to grow around either ZnCr2O4 clusters or PrVO4 grains.Fig. 3 shows the TEM image reveal the difference in microstructure between the second phase particles PrVO4 and ZnCr2O4.The electron diffraction pattern in Fig. 3 further confirmed the existence of PrVO4 and ZnCr2O4.The ZnCr2O4 grains are very small that the grain size is below 0.2 µm, but the PrVO4 grains have a relatively larger size than ZnCr2O4 which ranges from 0.5µm to 4µm.Most grains of ZnCr2O4 and PrVO4 were found located at the ZnO grain boundaries, and ZnCr2O4 grains tend to aggregate to form a heap of ~1µm in size.This observation suggests that these secondary minor phases could exert a 'pinning' effect, which hindered the movement of the grain boundaries of ZnO, thus limiting its grain growth.

Average ZnO grain sizes in sintered samples
For all fours sintered samples, the average grain size of each sample under all sintering conditions was determined following the procedure described in 2.2, and the results are summarized in Table 2.

Kinetic grain growth parameters
To reveal the effect of PrMnO3 addition on the sintering behaviour of the ZnO-V2O5-Cr2O3 ceramics, the isothermal grain growth behaviour at 875°C was first analysed using a rearranged Eq. ( 3), as shown below: Figure 4 illustrates vs. curves, which show the isothermal grain growth behaviour of the four samples sintered at 875°C for 2-8 h.A linear fit of the curves enabled the determination of the kinetic grain growth exponent from the slope of the linear fit for each sample.The values so determined are also shown in Fig. 4.
Notably, with the increased addition of PrMnO3, not only did the average grain size decrease, the grain growth rate also decreased, as indicated by the increased values.To determine the apparent activation energy Q associated with the grain growth, Eq. ( 3) Q can then be calculated from the gradient of the Arrhenius plot of log(G n /t) vs. (1/T).
Figure 5 shows the Arrhenius plots of the four samples from the G values obtained at different sintering temperatures shown in Table 2.The apparent activation energy values determined from Fig. 5 are summarized in Table 3, together with the values determined from Fig. 4.  As seen in Table 3, from both of this work and other published works, the basic ZnO-V2O5 binary systems had a kinetic grain growth exponent value spanning 1.5-1.8.
This value was lower than the = 3.0 observed for pure ZnO, indicating that the addition of V2O5 encouraged the grain growth of ZnO.This was attributed to the high reactivity of the V2O5-rich liquid phase during sintering, promoting the formation of abnormally grown ZnO grains.
The addition of a third component-e.g., the 0.35 mol% of Cr2O3 in this study or 0.5 mol% Sb2O3 as reported elsewhere [21] been shown to increase the value to 2.9 and 4.0, respectively.This means that the addition of a third component could hinder ZnO grain growth in basic ZnO-V2O5 binary systems.In both cases, this was attributed to the formation of spinel ZnCr2O4 or a ZnSb2O4 secondary phase, which exerts a pinning effect on ZnO grain boundaries and thus hinders ZnO grain growth [13,21,22].
It was clear that the addition of PrMnO3 to the ZnO+V2O5 (1 mol%) + Cr2O3 (0.35 mol%) ceramics resulted in a further increase in the value, which increased to 6.1 after a nominal addition of 0.75 mol% of PrMnO3 for ZVCP75, demonstrating the high effectiveness of PrMnO3 in suppressing ZnO grain growth.As seen in the XRD and SEM analyses, the addition of PrMnO3 resulted in the formation of an additional spinel PrVO4 phase, which was also found to be distributed at the grain boundaries of ZnO similarly to ZnCr2O4 in the sintered ceramics.A similar pinning effect was thus believed to be the main reason for the hindered ZnO grain growth by the addition of PrMnO3.
As expected, it is clearly seen in Table 3 that the apparent activation energy values for the grain growth followed an inverse behaviour compared to the values.Obviously, the higher the activation energy, the higher the values, the slower the grain growth rate, and the smaller the average ZnO grain sizes.

Grain size distribution in sintered samples
Figure 6 shows SEM images and grain size distribution histograms of the four samples sintered at 875°C for 4 h.It was clear from the SEM images that with increased additions of PrMnO3, not only did the average grain size of ZnO decrease, but the size distribution also narrowed significantly.Obviously, the ZVC sample without the addition of PrMnO3 exhibited a broad and obviously bimodal grain size distribution.In the context of this work, larger grains are considered to be the AGG, and the relatively smaller grains are considered as the normally grown grains (NGG).The area fractions of AGG and NGG for each sample are also included in the size distribution histograms in the figure.The addition of PrMnO3 not only resulted in the reduction and narrowing of the grain size range, but also a diminishing of the bimodal distribution.For ZVCP75, the grain size distribution became unimodal.
Figure 7 shows the average grain sizes of the AGG and NGG vs. the nominal addition of PrMnO3.This revealed the impact of PrMnO3 addition.Notably, the average grain size of NGG did not change much; however, the average grain size of the AGG decreased dramatically.This indicated the effectiveness of PrMnO3 addition in suppressing the coarsening of ZnO grains in this system.The number fractions of AGG ( ) and NGG ( ) were also determined for the sintered samples and are provided in the histograms.This significant stabilization of the switching field strength was attributed to the uniformization of the ZnO grain size as the result of PrMnO3 addition.As seen in Fig. 6, a 0.75 mol% addition of PrMnO3 in ZVC effectively eliminated the formation of abnormally grown ZnO grains.The resulting ZVCP75 sample was shown to have a markedly improved stability in its switching field strength.This was also consistent with the results reported by Hng et al. [12], which showed that the presence of just a few abnormally grown ZnO grains could have a pronounced destabilization effect on the switching field strength of ZnO-V2O5 varistors.

Electrical behaviour of sintered samples
Furthermore, with a 0.75 mol% addition of PrMnO3, there was also an marked increase in the 1 • −2 value for the ZVCP75 sample.This increase in switch field strength was mostly due to the refined ZnO grain size.It was noted that all the samples possessed a high sintering density at close to 95% of the theoretical density of ZnO.While the addition of PrMnO3 had little effect on the sintering density, it effectively reduced and eliminated the fraction of AGG and progressively refined the grain sizes of the resulting ceramics.
The homogenization and reduction of ZnO grain size were attributed to the noticeably enhanced electrical properties of the ceramics, most significantly the stabilization of the switching field strength.The nonlinear coefficient also increased marginally from 7.2 to 8.9 from ZVC to ZVCP75, respectively, as seen in Table 4.
In summary, when a small amount of PrMnO3 was added to the ZnO-V2O5 (1 mol%)-Cr2O3 (0.35 mol%) ceramics, an additional second phase, PrVO4, formed in addition to the commonly existing ZnCr2O4.Both of these phases tended to locate at the ZnO grain boundaries, hindering ZnO grain growth.Unlike the randomly distributed ZnCr2O4 particles, the PrVO4 particles seemed to selectively participate in the regions where abnormal ZnO grain growth would take place, thus effectively suppressing the AGG.This was believed to be due to the formation of PrVO4 consuming V2O5, which would only occur in the V2O5-rich liquid phase.This weakened the fast solution-precipitation process of the V2O5-rich liquid phase during sintering, thus demoting the abnormal ZnO grain growth.

Conclusions
In this study, the effect of PrMnO3 addition (0.25 to 0.75 mol%) on the grain growth behaviour of ZnO-V2O5(1 mol%)-Cr2O3(0.35 mol%) ceramics was systematically investigated during sintering from 850°C to 925°C with the aim to refine the ZnO grain structure.The main findings were as follows.
(1) The addition of PrMnO3 not only uniformized but also refined the ZnO grains in the resulting ceramics.At a 0.75 mol% PrMnO3 addition, the bimodal grain size distribution was completely eliminated, and the average ZnO grain size was reduced by more than 45% compared to the sample without PrMnO3 addition.
(2) The addition of PrMnO3 resulted in remarkably improved stability of the switching field strength, narrowing its range of variation from 1580 V/cm (without PrMnO3 addition) to only 91 V/cm (with 0.75 mol% PrMnO3 addition).The homogenization and reduction of the ZnO grain sizes were responsible for the observed stabilization of the switching field strength.
(3) A phenomenological analysis of the ZnO grain growth kinetics showed that the kinetic grain growth exponent increased from 2.9 without PrMnO3 to 6.1 after a 0.75 mol% PrMnO3 addition, corresponding to an increase in apparent activation energy from 202±29 to 697±66 kJ/mol, respectively.
(4) The formation of a PrVO4 secondary phase as the result of PrMnO3 addition was responsible for the grain growth behaviour observed.The PrVO4 phase was found to mostly locate at the ZnO grain boundaries, thus hindering and eventually eliminating the abnormal growth of ZnO grains.

Fig. 2
Fig.2shows a typical microstructure of a sintered ZVCP50 sample.As expected, the dominant morphology included grains of the ZnO phase, and the focus here was on the appearance and distribution of the minor phases.As seen in the micrograph, in addition to ZnCr2O4 (ZC) clusters (circled in white-dashed lines), PrVO4 (PV) grains (circled in yellow-dashed lines) were clearly seen to also distribute at the junctions of ZnO grains.

Fig. 3 .
Fig. 3. TEM micrographs and electron diffraction patterns from secondary particle grains: (a) bright-field image of ZnCr2O4 grains from a sample with the composition of ZVCP75 sintered at 875 °C for 4 h, (b) bright-field image of PrVO4 grains from a sample with the composition of ZVCP75 sintered at 875 °C for 4 h, (c) electron diffraction pattern from the [001] zone of ZnCr2O4 grain in (a) and (d) electron diffraction pattern from the [010] zone of PrVO4 grain in (b).

Fig. 6 .
Fig. 6.SEM microstructures and grain size distribution histograms of the samples sintered at 875 °C for 4 h: (a) ZVC, (b) ZVCP25, (c) ZVCP50, and (d) ZVCP75.NGG and refers to the normally grown grain and its number fraction.AGG and refer to the abnormally grown grain and its number fraction.

Fig. 7 .
Fig. 7. Effect of nominal addition of PrMnO3 on the average grain sizes of the AGG and NGG in ZVCP samples sintered at 875 °C for 4 h.

Figure 8
Figure 8 shows the switching field strength 1 • −2 of the four samples sintered at 875°C for 4 h.For each sample, 10 specimens were prepared, their I-V curves were measured, and their 1 • −2 values were determined.Figure 8(a) shows all measured 1 • −2 data for four samples.As seen, the basic ZVC sample displayed a very broad range of 1 • −2 .Obviously, a 0.25 mol% nominal addition of PrMnO3 resulted in a remarkable narrowing of the 1 • −2 range for the ZVCP25 sample.Further increasing the PrMnO3 addition to 0.75 mol% produced a continuous narrowing of the 1 • −2 range.

Fig. 8 .
Fig. 8. Measured switching field strength data vs nominal addition amount of PrMnO3.

Table 1 .
Basic compositions of the ZnO ceramics studied in this work

Table 2 .
Average grain sizes estimated from SEM analysis of ZVC and ZVCP samples sintered at different conditions

Table 3 .
Summary of ZnO grain growth exponents and apparent activation energy for grain growth

Table 4
summarises the relative density and number fraction of AGG and the average grain size of the sintered samples, together with the nonlinear coefficient and the average and range of the switching field strength of these samples.

Table 4 .
Relative density, number fraction of AGG and average grain size, and electrical properties of the sintered samples at 875 °C for 4 h.