Available online at www.sciencedirect.com

www.elsevier.com/locate/actamat

ScienceDirect

Acta Materialia 61 (2013) 7849–7858

Sintering behavior and non-linear properties of ZnO varistorsprocessed in microwave electric and magnetic fields at 2.45 GHz

Alexandre Badev a,⇑, Sylvain Marinel a, Romain Heuguet a, Etienne Savary a,b,Dinesh Agrawal c

a CRISMAT Laboratory, UMR 6508 CNRS-ENSICAEN 6, Boulevard du Marechal Juin, 14050 Caen Cedex 4, Franceb LMCPA Laboratory, Universite de Valenciennes et du Hainaut-Cambresis, ZI du Champ de l’Abbesse, 59600 Maubeuge, France

c Microwave Processing and Engineering Centre, The Pennsylvania State University, 107 MRL, University Park, PA, USA

Received 6 May 2013; received in revised form 13 September 2013; accepted 13 September 2013Available online 4 October 2013

Abstract

A study of the densification behavior and grain growth mechanisms of ZnO-based varistors composed of 98 mol.% ZnO–2 mol.%(Bi2O3, Sb2O3, Co3O4, MnO2) has been carried out. The pressed samples were sintered in microwave electric (E) and magnetic (H) fieldsusing a single-mode cavity of 2.45 GHz. The effect of the sintering temperature (900–1200 �C), holding time (5–120 min) and sinteringmode (E, H) on the microstructure and electrical properties of the sintered varistor samples were investigated. The grain growth kineticswas studied using the simplified phenomenological equation Gn = kte(�Q/RT). The grain growth exponent (n) and apparent activationenergy (Q) values were estimated for both electric and magnetic heating modes and were found to be n = 3.06–3.27, Q = 206–214 kJ mol�1, respectively. The lower value of n estimated in the E field was attributed to a volume diffusion mechanism, whereas thehigher n value in the H field sintering was correlated mainly to a combined effect of volume and surface diffusion processes. Samplessintered in the H and E fields showed high final densities. Moreover, the ones sintered in the H field presented slightly higher densityvalues and bigger grains for all sintering temperatures than E field heated ones. The optimal sintering conditions were achieved at1100 �C for a 5 min soaking time for both H and E field processed samples, where respectively densities of 99.2 ± 0.5% theoretical density(TD) and 98.3 ± 0.5% TD along with grain size values of G = 7.2 ± 0.36 lm and G = 6.6 ± 0.33 lm were obtained. Regarding the elec-trical properties, breakdown voltage values as high as 500–570 V mm�1 were obtained, together with high non-linear coefficients a = 29–39 and low leakage currents (Jl � 5 � 10�3 mA cm�2), respectively, for E and H field sintered varistor samples. Moreover, samples sin-tered in an H field systematically exhibited higher breakdown voltage values compared to the ones sintered in the E field. This was attrib-uted to an improved coupling between the H field and the present dopants within the ZnO matrix, this latter being mostlysemiconductive, thus leading to an enhanced reactivity and improved properties of the electrostatic barrier.� 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Microwave; Varistor; Microstructure; Sintering; Breakdown voltage

1. Introduction

Zinc oxide (ZnO) varistors are non-linear, two-terminal,semiconductor voltage-dependent resistors, which arewidely used as transient voltage surge suppressors to limitvoltage surges in the range from low voltages (5 V) to high

1359-6454/$36.00 � 2013 Acta Materialia Inc. Published by Elsevier Ltd. All

http://dx.doi.org/10.1016/j.actamat.2013.09.023

⇑ Corresponding author. Tel.: +33 (0)2 31 45 13 69.E-mail address: alexandre.badev@ensicaen.fr (A. Badev).

voltages (up to 1 MV). The non-linear J–E characteristicsresult from the nature of the grain boundary layer, whichmainly consists of ionic additives such as Bi2O3, Sb2O3,Co3O4, etc. Many researchers have reported the sinteringbehavior of several doped ZnO systems, such as Bi2O3-doped ZnO [1], Sb2O3-doped ZnO [2] and Al2O3-dopedZnO [3]. The electrical properties of ZnO varistors areobviously related to the composition and microstructure,such as grain size, density, morphology and the

rights reserved.

7850 A. Badev et al. / Acta Materialia 61 (2013) 7849–7858

distribution of second phases. As a consequence, the non-ohmic properties of these materials are strongly dependenton the fabrication process, which in turn regulates thedefect chemistry in the depletion region and in the vicinityof the grain boundary. It is commonly accepted that themicrowave sintering process can densify ceramic materialsin a very short time and mostly at lower temperatures thanconventional sintering methods. Several authors havereported the non-thermal effects induced by the microwavefield (E, H) [4], acting as an additional driving force (elec-tromigration, ponderomotive force etc.) for diffusion mech-anisms. In 1994, Cherradi et al. [5] were the first to noticethat the H microwave field can be used for heating semi-conducting materials in a single-mode microwave cavity.The basic experimental setup allowed them to correlatethe distribution of the E and H fields inside the cavity withthe temperature distribution through a CuO bar, this latterbeing microwave-heated in either an H field or an E field.This H field component has been successfully used whenusing a specific assembly for heating up semiconductingmaterials (mainly oxides). Subsequently, a systematicinvestigation of the heating profile of various materials inE and H field separation was carried out at Penn State Uni-versity [6,7] in a single-mode microwave cavity at2.45 GHz. As stated in one of our previous works basedon the sintering behavior of pure ZnO in both E and H

fields, the material exhibited denser and more homoge-neous microstructure during the sintering in the microwavemagnetic (H) field [8]. This study, which had not been car-ried out previously, was also applied for the investigationof the sintering behavior of ZnO with several dopantsadded. Indeed, one could expect the specificities of the elec-tric and magnetic field during the microwave selective sin-tering of ZnO-based compositions. Due to the highabsorption of electromagnetic power and the conductivenature of various dopants in ZnO varistor compositionssuch as Bi2O3 [9], Co3O4 etc., it is expected that the heatingwill be very high when sintered in a microwave magneticfield.

We report the effect of the sintering temperature andsoaking time on the microstructure of ZnO-based ceramicsheated in microwave E and H fields. The correlationbetween the microstructure properties and their J–E prop-erties has been also established. For each heating mode,different sintering cycles were used and microstructureand electric characterizations have been carried out. Theelectrical properties related to the processing conditionsare presented and discussed according to the heatingconditions.

2. Experimental procedure

2.1. Powder synthesis, characterization and shaping

High-purity nano-ZnO powder (Nanogard, 99.99% pur-ity, 60 nm mean particle size) and pure grade dopants inthe proportion of 98 mol.% ZnO, 0.5 mol.% Bi2O3,

0.5 mol.% Sb2O3, 0.5 mol.% Co3O4 and 0.5 mol.% MnO2

were used for the preparation of ZnO-based varistors.The mixture was ball-milled in an agate mortar using5 mm diameter zirconia balls in ethanol media. After dry-ing the slurry at 100 �C for 24 h, an organic binder (Rho-doviol 4%, Prolabo) was introduced to the dry mixture toensure a good cohesion between particles. The latter wasthereafter dried under an infrared lamp in order to evacu-ate the water content of the binder solution. The shaping ofthe samples was done by uniaxial pressing at 110 MPausing a cylindrical steel die with a diameter of 8 mm, fol-lowed by cold isostatic pressing at 300 MPa. The pelletswere then calcined in air at 650 �C for 2 h in order toenhance powder reactivity. The crystalline phases wereidentified by X-ray diffraction (XRD) using Cu Ka radia-tion (Phillips X’Pert diffractometer). For the microstruc-tural observations, scanning electron microscopy (SEM)observations were made on 1 lm polished and H3PO4

etched sample surfaces. Bulk densities of the samples weredetermined using their weights and dimensions. All sam-ples held a green density of 66% of the theoretical density(TD) and weighed �0.5 g. Based on the accounted absoluteerrors on the samples dimensions and weights, the overallpercentage error on the final densities of the samples wasestimated to be ±0.5% of the calculated data.

Grain size measurements were carried out on the micro-graphs of the etched samples using the following equation:

G ¼ 1:56L ð1Þwhere G is the average grain size and L is the average grainboundary intercept length of nine random lines on two dif-ferent micrographs of each sample. Each line accounted for�30 grain interceptions. The percentage error on the grainsize values was estimated to be ±5% of the calculated data.

2.2. Microwave furnace

The microwave equipment used in this study consists ofa 2 kW, 2.45 GHz magnetron source (SairemGMP20KSM), a water-cooled aluminum circulator, a rect-angular twist and a waveguide (WR340) that conducts themicrowaves to a tailor-made TE10p single mode cavitywith ports for vacuum and gas feed. The operating modesused in the cavity are TE103 (maximum E field region inthe center of the cavity where the H field is minimum)and TE102 (maximum H field region in the center of thecavity where the E field is minimum), and were adjustedby tuning the length between the coupling iris and the shortcircuit piston from both side ends (Fig. 1).

The sample was placed in an alumina-silicate box Fiber-fax (DURABOARD), which is microwave transparent andensures a homogeneous thermal insulation of the sample byreducing the thermal radiation from the sample surface(Fig. 1b). The box and sample were placed in the middleof the cavity and the cavity length was tuned to either E

or H field mode. An infrared pyrometer was placed abovethe sample and allowed precise optical temperature

Fig. 1. Schematic representation of the varistor sample placed in the microwave cavity under microwave radiation.

A. Badev et al. / Acta Materialia 61 (2013) 7849–7858 7851

measurements to be done within the range 348–2500 �C.The heating cycles were interfaced on a computer, whichallowed an estimation of the heating rate and reproducibil-ity of the heating cycles.

2.3. Microwave sintering

As reported previously [10], the use of microwavesallows the transfer of energy directly into the materials,where it is converted to heat through various absorptionmechanisms such as ionic conduction, dipole relaxation,photon–phonon interaction and the Joule effect. Thus,microwave-sintering techniques mostly allow a volumetricheating, providing high heating rates, markedly optimizingthe processing conditions as compared to the conventionalsintering routes.

The microwave sintering was performed in air using a250 W incident power. Four different temperatures werechosen (900, 1000, 1100, 1200 �C) with four dwell times(5, 30, 60 and 120 min) at each sintering temperature,and the heating ramp was �250 �C min�1 for each heatingcycle. During the sintering process, the incident power wasfixed at 250 W, and in order to increase the temperatureinside the sample, the reflected power was adjusted by tun-ing a stub placed prior to the cavity.

In order to compare and analyze the different likelypeculiar effects during microwave sintering of ZnO-basedvaristors, the samples were respectively placed in configura-tions of maximum electric and magnetic fields distributionsat the center of the cavity, corresponding to microwaveelectric (E) and magnetic (H) sintering modes. During thedwelling step at each sintering temperature, the maximumstable temperature of the sample surface was measuredby an infrared pyrometer (Modline 5 model).

2.4. Characterizations of the sintered samples

The microstructures of the microwave E and H fieldsintered samples were examined by SEM (Supra 55,Zeiss). In order to quantify the different phases presentin ZnO grains and at the grain boundary regions, a localchemical composition was investigated using energydispersive spectroscopy (EDS) coupled with SEM, asthe microscope is equipped with the EDAX Genesissystem.

Electric field–current density (E–J) experiments of thesintered pellets were measured at room temperature usinga Keithley 237 unit. Before the measurements, the sampleswere coated with a silver paste at both faces. Ohmic contactwas ensured after a heat treatment at 750 �C only for30 min in order to prevent significant microstructuralchanges that might occur. The breakdown voltage (EB)was determined for a current density of 1 mA cm�2, andthe leakage current density (JL) was estimated at 0.8EB.The non-linear coefficient was calculated using theequation:

a ¼ ðlog J 2 � log J 1Þðlog E2 � log E1Þ

� �ð2Þ

with J2 = 10 mA cm�2 and J1 = 1 mA cm�2.

3. Results and discussions

3.1. Powder synthesis and phase composition of the sintered

samples

After the ball-milling and calcination processes, ZnO-based powder particles are nanosized with a near sphericalshape (Fig. 2).

7852 A. Badev et al. / Acta Materialia 61 (2013) 7849–7858

The average grain diameter (d) value of the powder par-ticles was estimated from the SEM image measurements,and was found to be �65 nm. Then an estimated value ofthe specific surface area (S) was deduced from

d ¼ 6000

S � q ; S ¼ 17 m2 g�1 ð3Þ

using the theoretical density of the main compound(qZnO = 5.66 g cm�3), and assuming spherical and monodi-spersed powder.

The XRD patterns of the samples, previously polishedand sintered in different conditions, are shown in Fig. 3.

In the XRD diagram are shown the patterns of samplessintered at 1100 �C for a 5 min soaking time. These thermalconditions were estimated to be the optimal ones in termsof densification of the sintered samples. Therefore theproperties of microwave E and H field processed samplesusing the above-mentioned heating cycle will be mainlydiscussed.

Moreover, a set of samples sintered in a conventionalfurnace were compared in terms of phase composition withthe ones heated in both microwave E and H fields, usingidentical thermal conditions.

The XRD patterns of the microwave E and H field sin-tered samples at 1100 �C revealed crystal structures similarto the conventionally sintered ones. The identified phaseswere hexagonal ZnO (PDF: 01-089-0511, main phase);Bi2O3 (PDF: 00-002-0988) present for samples calcined at650 �C; pyrochlore phase (Bi3Zn2Sb3O14), also known asBi-rich phase, which resulted from the reactivity of theBi2O3 phase with ZnO and Sb2O3 above 700 �C; and a spi-nel type phase Zn7Sb2O12 that resulted from the decompo-sition of the pyrochlore phase at temperatures higher than1000 �C. As a result, it can be seen that the microwave

Fig. 2. SEM image of the varistor powde

process did not change the main phases in our ZnO-basedvaristor compositions.

3.2. Microwave sintering of ZnO varistors in electric (E)

and magnetic (H) fields

The weight losses of ZnO varistors sintered in the range900–1200 �C for 5–120 min were plotted (Fig. 4). The figureshows that the weight loss evolution with the temperaturefollows a near linear trend.

The weight loss increased with the sintering temperatureand soaking time as expected. This is likely due to the evap-oration of bismuth for temperatures above 1000 �C, asreported in Ref. [11]. Furthermore, higher sintering tem-peratures and longer processing time led to a further evap-oration of the bismuth phase, as well as some othercompounds such as cobalt and antimony. Density mea-surements were performed on samples sintered in bothmicrowave electric and magnetic fields (Fig. 5a and b).

Fig. 5 shows that for both E and H field sinteringmodes, the maximum density of the samples are obtainedat 1100 �C for a 5 min soaking time (98.3 and99.2 ± 0.5% of the theoretical density respectively). Asreported in Ref. [12], the Bi-rich phase that appears at�850 �C homogeneously wets ZnO grains, thereby pro-moting the densification rate of ZnO varistors by accelerat-ing mass transport, through liquid phase assisted diffusion.

Moreover, samples sintered in a microwave H fieldexhibited slightly higher final densities and larger grain size(Fig. 6b). As mentioned in one of our previous works [8], insemiconductors such as ZnO-based ceramics, it is under-stood that the H field induces current loops which areresponsible for the heating up of the samples. The interac-tion of the current loops with the H field could give rise to

r synthesized by the solid state route.

Fig. 3. XRD pattern of (a) pure ZnO raw powder, (b) calcined varistor powder, (c) conventionally sintered varistor samples, (d) microwave E fieldsintered varistor samples, (e) microwave H field sintered varistor samples.

Fig. 4. Weight loss of microwave sintered samples at different tempera-tures for 5–120 min holding times.

A. Badev et al. / Acta Materialia 61 (2013) 7849–7858 7853

radial Laplace forces applied onto the granular compact,which in turn would lead to a better contact between par-ticles and enhance grain growth and densification. Thiswould also create a particle rearrangement in the material,consequently increasing the kinetics of densification andgrain growth.

For temperatures above 1100 �C, the weight lossincreased and the final relative densities of the samplesslightly decreased to 97.7 and 98.7 ± 0.5% TD, respectively

for E and H field heated samples. As reported in the workof Mazaheri [13], this could be related to the antidensifica-tion phenomenon that occurs at high temperatures due tothe partial evaporation of some dopants and grain growth.

The antidensification phenomenon was also observedwhen the samples were subjected to longer soaking times.As shown in Fig. 5a and b, for soaking times ranging from5 to 30 min, the relative density of the sintered samplesincreased when the temperature increased from 900 �C to1000 �C. Then a decrease of the latter was afterwardsobserved for temperatures above 1000 �C. For longer soak-ing times, the important evaporation of bismuth is likely tobe responsible for the decrease of the relative density athigher sintering temperatures. An abnormal grain growthwas also observed when the sintering temperature wasabove 1100 �C and for soaking times longer than 60 min,as marked by the extremely higher grain size shown inFig. 6.

As reported in Ref. [9], grain growth and coarseningthrough coalescence play a major role in the densificationprocess of ZnO varistors. However, when the sinteringtemperature reached 1200 �C, the grain growth had littleeffect on the densification since most of the dopants presentwithin the composition began to evaporate, producing lar-ger porosity.

In this section, the grain growth kinetics will be dis-cussed in detail.

Fig. 5. Relative densities of microwave (a) E field, (b) B field sintered ZnOvaristors at various temperatures for 5–120 min soaking times.

Fig. 6. Sintering paths for microwave (a) E field, (b) H field sintered ZnOvaristors for various soaking times at different temperatures.

7854 A. Badev et al. / Acta Materialia 61 (2013) 7849–7858

3.3. Microstructure characterization and grain growth

kinetics

Fig. 7 exhibits SEM micrographs of polished and etchedvaristor samples sintered at 900–1200 �C.

It is to be noted that for all sintering temperatures, whitespots are seen in SEM photos around ZnO grain boundaryregions, corresponding to the Bi-rich (pyrochlore) phase.As the temperature was raised to 1100 �C, this liquid phasespread uniformly throughout the grain boundaries, and thepresence of few round shaped small white spots (Fig. 7C)throughout ZnO grains marked the formation of the spinelphase, resulting from the partial decomposition of thepyrochlore phase. For samples sintered at 1200 �C, thequantity of pyrochlore phase present between ZnO grains(Fig. 7D) is less than the one present in the samples sinteredat 1100 �C (Fig. 7C).

Regarding the porosity, samples sintered at 900 �C pre-sented small isolated pores and few white spots in ZnOgrain boundary regions that marked the beginning of liquidBi-rich phase formation. The presence of a mass transportmechanism through the liquid phase at 1100 �C and higher

mobility of grain boundaries caused the entrapment of iso-lated pores and led to larger isolated intergranular pores, asshown in Fig. 7C. Furthermore, the presence of many iso-lated larger pores present in samples sintered at 1200 �Cresulted mostly from the evaporation of the Bi-rich phasealong ZnO grains (Fig. 7D).

The grain growth kinetics can be determined using thesimplified phenomenological kinetics (Eq. (4)). The graingrowth exponent (n) can be found at isothermal conditionsfollowing the kinetic equation:

n log G ¼ log t þ ½log K0 � 0:434ðQ=RT Þ� ð4Þwhere n is the grain growth exponent, t is the sinteringtime, K0 is a constant, Q is the sintering activation energy,R is the universal gas constant and T is the sinteringtemperature.

This law is valid for ceramic materials holding relativedensities above 95%.

The n value can be calculated from the slope (1/n) of thelog (grain size) vs. log (time) plot. Plots were made for iso-thermal conditions at each sintering temperatures and the(n) values were calculated from the slopes of the plots

Fig. 7. SEM micrographs of microwave H field sintered ZnO varistors for a 5 min soaking time at (A) 900 �C, (B) 1000 �C, (C) 1100 �C, and (D) 1200 �C.

A. Badev et al. / Acta Materialia 61 (2013) 7849–7858 7855

constructed. Lin and co-workers have reported the n valuesfor microwave and conventionally sintered ZnO varistorsto be 3 and 4, respectively [14]. Gunay et al. have estimatedthe n value for the ZnO–Sb2O3–CoO system to be 5.2 [12].They also indicated that the n value in a system was asso-ciated with a grain-growth inhibition mechanism criterion.

Fig. 8 shows that the n factor decreased mostly with theincrease of the sintering temperature for both microwave E

and H field sintered samples.The n value for the E field sintered ZnO varistors was

found to be �3, which is consistent with a volume diffusionmechanism that takes place during the final stage of thesintering process [14]. This could be explained by the pen-etration depth of the electric field being within the centime-ter range, thus promoting an increment of volume diffusionmechanisms within the ceramic material. On the otherhand, the higher n value for microwave H field sinteredsamples suggests a major contribution of volume diffusionmechanisms with a non-negligible contribution of a surfacediffusion mechanism (where the n factor is known to beabove 3), which is consistent with the H field penetrationdepth being within the micrometer range. As the H fieldwould be mainly concentrated on the surface and betweenZnO grains, it would promote grain boundary diffusionmechanisms.

If Eq. (4) is expressed in the form:

logðGn=tÞ ¼ log K0 � 0:434ðQ=RT Þ ð5Þ

The apparent activation energy (Q) of a grain growthprocess can be calculated from the Arrhenius plot of log(Gn/t) vs. 1/T (K�1). Such plots have been constructedfor the microwave E and H field systems, by taking intoaccount the average values of n calculated for each system(Fig. 9).

The apparent activation energy of 206 kJ mol�1 hasbeen determined for microwave E field sintered samples,which is slightly lower than the one estimated duringmicrowave H field sintering (214 kJ mol�1). Lin and co-authors reported the activation energies of 225 kJ mol�1

and 363 kJ mol�1, respectively, for microwave and conven-tional sintering systems. The estimated values of the activa-tion energies are in good agreement with the data found inthe literature.

We can assume that the lower values of n, as well as thelower apparent activation energies obtained throughmicrowave sintering of ZnO varistors, compared to theones obtained through conventional sintering, areconsistent with the often-reported theory of microwavenon-thermal effect [7] induced during the microwavesintering.

3.4. Electrical properties

Fig. 10 shows the J–E characteristics of the microwave E

and H field sintered varistors. As shown in Fig. 10, at lowvoltage, the current passing through the ceramic is low. As

Fig. 8. Isothermal grain growth of microwave (a) E field, (b) H fieldsintered ZnO varistors.

Fig. 9. Arrhenius plots for the grain growth of microwave (a) E field, (b)H field sintered ZnO varistors.

Fig. 10. J–E characteristics of the sintered varistors for a 5 min soakingtime at different temperatures.

7856 A. Badev et al. / Acta Materialia 61 (2013) 7849–7858

a result, the J–E response is mainly governed by the resis-tivity of the intergranular or grain boundary phase, whichis mainly composed of the pyrochlore (bismuth-rich) phase[15]. For high field regions, the electrical response is mainlygoverned by the ZnO grains’ resistivity. It can be observedfrom Fig. 10 that almost all microwave E and H field sin-tered samples indicate a sharp transition from the low cur-rent zone to the non-linear region. Almost all microwavesintered samples exhibit non-linear coefficient values (a)higher than 20, thus presenting varistor type behavior(Table 1).

It can be seen from Table 1 that the optimal electricalproperties (high a, low Jl, high EB) for both microwave E

and H field sintered varistor samples were obtained for a5 min soaking time at 1100 �C. These results are consistentwith the highest values of the relative density obtained forsuch a heating cycle. For both heating modes (E or H field)and for this optimized thermal cycle, the leakage current isquite low (�4–5 lA cm�2), which accounts for a high grainboundary resistivity [15], and this is in adequacy with thehomogeneous phase distribution obtained within the sam-ples, and previously shown on the microstructures

(Fig. 7c). Considering the slightly larger grain size(7.2 ± 0.36 lm) of samples sintered in the H field, thanthe ones sintered in E field (6.6 ± 0.33 lm) (Fig. 6), it isexpected a more pronounced decrease of the breakdown

Table 1Electrical properties of the sintered varistors at different temperatures andfor a 5 min soaking time.

Tx (�C) a ± 0.2 Eb ± 2 (V mm�1) Jl ± 0.005 (mA cm�2)

Microwave (E field) sintered varistors (5 min at Tx)

900 29.6 376 0.0051000 29.53 361 0.00681100 29.08 503 0.0041

1200 20.87 172 0.07

Microwave (H field) sintered varistors (5 min at Tx)

900 28.13 625 0.00151000 37.66 581 0.0071100 39.05 575.6 0.005

1200 38.9 550 0.0065

The values in bold are the optimal electrical parameters at the optimalsintering conditions.

A. Badev et al. / Acta Materialia 61 (2013) 7849–7858 7857

voltage values for microwave H field processed samplesthan for those processed in E field. Actually the oppositewas observed: for all thermal cycles investigated, the sam-ples sintered in an H field systematically exhibited a higherbreakdown voltage values compared to the samples sin-tered in E field (Table 1). In our previous paper [11], with-out distinguishing the E and H field mode, and using aZnO-based susceptor, it was mentioned that the volumetricheating distribution provided by microwaves, promotedthe reactivity of dopants with the ZnO matrix, particularlywithin the grain boundary regions. As a consequence, theproperties of the electrostatic barrier were improved,resulting in better non-linear properties as compared withconventional heated samples. In the present work, no sus-ceptor was used (hence it is direct microwave heating) andE and H field modes are compared regarding the micro-structure and the non-linear properties of ZnO varistortype component. Although the grain size is bigger for

Fig. 11. Schematic representation of the microwave interaction with thedifferent phases present in ZnO varistors.

samples sintered in the H field than for those sintered inE field, the breakdown voltage is also significantly higher.This strongly suggests that those samples have an electro-static barrier higher than the ones sintered in the E field.This is understood by an improved reactivity betweendopants (especially Bi2O3, which is highly semiconducting)and ZnO matrix within the grain boundary region. It isbelieved that in H field mode, the intergranular region,mostly composed of semiconducting type materials(Bi2O3, CoO, MnO etc.), preferentially couples with micro-waves, resulting in locally enhanced reactivity. In E fieldmode, the ZnO matrix, which is mainly insulating, prefer-entially couples with microwaves, resulting in a lower reac-tivity within the grain boundary region, where thesemiconducting phases are localized. These two distinctbehaviors could explain to a large extent the differencesobserved in breakdown voltages as a function of the E orH field sintering mode. Fig. 11 shows the proposed interac-tions of the field with the material regarding the above-mentioned discussion.

4. Conclusion

A study of the densification behavior and grain growthwas carried out on varistor samples composed of 98 mol.%ZnO–2 mol.% additives (Bi2O3, Sb2O3, Co3O4, MnO2)using a microwave radiation in E and H field configurationmodes. The highest density values, non-linear coefficientand breakdown voltage values were obtained for micro-wave E and H field sintered samples at 1100 �C for a5 min soaking time.

The lower values of the grain growth exponent (n) andactivation energy (Q) estimated during the microwave E

field sintering process were attributed to a volume diffusionmechanism, and the higher (n) and (Q) value related to themicrowave H field sintering were correlated mainly to avolume diffusion process, with a non-negligible contribu-tion of a surface diffusion.

Moreover, the contribution of the magnetic (H) field onthe microstructure resulted in enhanced electrical proper-ties of the sintered samples. Following the optimal sinteringconditions (1100 �C, 5 min soaking time), samples sinteredin a magnetic H field showed slightly higher densities(D = 99.2 ± 0.5% TD) and bigger grains (G = 7.2 ±0.36 lm) than the ones sintered in electric (E) field(D = 98.3 ± 0.5% TD, G = 6.6 ± 0.33 lm). Regarding theelectrical parameters, breakdown voltage values as highas 503–570 V mm�1 were obtained together with highnon-linear coefficients a = 29–39 and low leakage currents(Jl � 5 � 10�3 mA cm�2), respectively for microwave E

and H field heated varistor samples. These values arehigher (a = 39, Eb = 570 V mm�1) for samples sintered inthe H field compared to the ones sintered in the E field(a = 29, Eb = 503 V mm�1), which is likely due to animproved coupling between the H field and the differentdopants within the grain boundaries region, being of asemiconductive nature. Consequently, the dopants–matrix

7858 A. Badev et al. / Acta Materialia 61 (2013) 7849–7858

reactivity has been enhanced, together with the propertiesof the electrostatic barrier. At last but not least, the abilityto uncouple the microwave electric field from the magneticone allowed us to tailor the electrical properties of suchfunctional ceramics, which is a major breakthrough inthe microwave sintering process.

Acknowledgements

The authors would like to express their gratitude for theANR Grant No. 2011 BS08 014 01, which supported thiswork. The authors are also thankful to the team for givingvaluable guidance all along this study in the area of micro-wave sintering of ceramics, grain growth mechanisms,varistor composition synthesis and electrical measure-ments. Special thanks to Mr. Jerome Lecourt and Mr.Jean-Franc�ois Lefebre from CRISMAT Laboratory,UMR 6508 CNRS-ENSICAEN, for their help and enthu-siasm during the experimental parts.

References

[1] Senda T, Bradt RC. J Am Ceram Soc 1990;73:1.[2] Senda T, Bradt RC. J Am Ceram Soc 1991;74:6.[3] Hu S et al. IEEE Trans Compon Hybrids Manuf Technol 1985;8:4.[4] Rybakov KI, Olevsky EA, Semenov VE. Scripta Mater 2012;66:12.[5] Cherradi AD, Provost G, Raveau GB. Electroceramics IV 1994.[6] Cheng J, Roy R, Agrawal D. Mater Res Innovat 2002;5:3–4.[7] Roy R et al. Mater Res Innovat 2002;6:3.[8] Badev A, Heuguet R, Marinel S. J Eur Ceram Soc 2013;33:6.[9] Dey D, Bradt RC. J Am Ceram Soc 1992;75:9.

[10] Semenov VE, Rybakov KI. Advances in microwave and radiofrequency processing. Berlin, Heidelberg: Springer; 2006.

[11] Savary E et al. J Alloys Compd 2011;509:21.[12] Gunay V, Gelecek-Sulan O, Ozkan OT. Ceram Int 2004;30:1.[13] Mazaheri M. Ceram Mater 2010;62:4.[14] Cong Lin ZX, Peng Hu, Sun Dan-Feng. Am Ceram Soc Bull

2007;86:1.[15] Cai J et al. J Am Ceram Soc 2007;90:1.