Effect of valence state and incorporation site of cobalt

Available online at www.sciencedirect.com

www.elsevier.com/locate/actamat

Acta Materialia 61 (2013) 1489–1498

Effect of valence state and incorporation site of cobalt dopants onthe microstructure and electrical properties of 0.2PZN–0.8PZT ceramics

M.-P. Zheng a, Y.-D. Hou a,⇑, F.-Y. Xie b, J. Chen b, M.-K. Zhu a, H. Yan a

a College of Materials Science and Engineering, Beijing University of Technology, Beijing 100124, Chinab Instrumental Analysis & Research Center, Sun Yat-sen University, Guangzhou 510275, China

Received 26 April 2012; received in revised form 16 October 2012; accepted 13 November 2012Available online 12 December 2012

Abstract

A CoCO3-added Pb((Zn1/3Nb2/3)0.20(Zr0.50Ti0.50)0.80)O3 (0.2PZN–0.8PZT) system was prepared and investigated. The results revealthat Co ions are present in the mixed valence form of +2 and +3, and their relative mole ratio depends on doping content. Co dopinginduces a phase transformation from the MPB to the tetragonal phase side, and the corresponding domain size increases accordingly.The solubility limit of Co ions in the perovskite matrix is near 0.2 wt.% in CoCO3 form. Below the solubility limit, Co ions enter the B

sites of the oxygenic octahedral center, and the charge-compensating oxygen vacancies accelerate mass transport, which assists in thedensification of the specimens. Above the solubility limit, excess Co ions gather in the grain boundaries and triple junctions, which facil-itate the formation of a liquid phase with excess PbO and lead to remarkable grain growth. The resulting size effect plays a dominant rolein improving the piezoelectric properties, which compensates for the hardening effect of acceptor doping. The optimum piezoelectricparameters are obtained at 0.8 wt.% CoCO3-doped 0.2PZN–0.8PZT system: d33 = 310 pC N�1, kp = 0.66, and emax = 14,600.� 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Perovskites; Ferroelectricity; Phase transformations; Grain growth; Grain-boundary structure

1. Introduction

Pb(Zr,Ti)O3. (PZT)-based ceramics are an importantpiezoelectric material widely applied in the electronicindustry, mainly in the manufacture of actuators and sen-sor devices [1–3]. By introducing various additives to thePZT matrix, piezoelectric and dielectric properties can betailored to a wide extent [4–7]. The effect of dopants is acomplex matter. However, a number of important general-izations have been proposed regarding aliovalent substitu-ents in perovskites. Donor dopants (e.g., La3+ [8], Nb5+ [9]and W6+ [10])—those with higher charges than the ionsthey replace—are compensated by cation vacancies,whereas acceptor dopants (e.g., Fe3+ [11], Cr3+ [12], Li+

1359-6454/$36.00 � 2012 Acta Materialia Inc. Published by Elsevier Ltd. All

http://dx.doi.org/10.1016/j.actamat.2012.11.026

⇑ Corresponding author. Tel.: +86 10 67392445; fax: +86 10 67392445.E-mail address: [email protected] (Y.-D. Hou).

[13] and Mn3+/Mn2+ [14])—ions with lower charges thanthe ions they replace—are compensated by oxygen vacan-cies. Donor doping in PZT could be expected to reducethe concentration of oxygen vacancies, leading to a reduc-tion in the concentration of domain-stabilizing defect pairs.The resulting increase in wall mobility causes an observedincrease in dielectric constant and loss, piezoelectric coeffi-cients and coupling factors, and a reduction in the mechan-ical quality factor Qm, while the acceptor doping increasesthe concentration oxygen vacancies, pinning the domainwall and causing an observed contrary effect comparedwith the donor doping. The two types of substitutionbehaviors result in the “softening” and “hardening” effect,respectively. In addition, the dopants are also known to beimportant factors influencing microstructural develop-ment. Some works indicate that the addition of dopantsabove the solubility limit can strongly reduce grain size[15–17]. The solid solution impurity drag mechanism has

rights reserved.


mailto:[email protected]


Fig. 1. (a) XRD patterns and (b) comparison of (002)T, (200)R and(200)T reflections (from left to right) for 0.2PZN–0.8PZT ceramics withdifferent CoCO3 additions.

1490 M.-P. Zheng et al. / Acta Materialia 61 (2013) 1489–1498

been proposed to explain this phenomenon, but withoutany corroborating evidence [18–20]. Conversely, in otherworks [21–23], gradual increase in grain size has beenobserved in the doped ceramic system, even with the dop-ant level above the solubility limit. However, the relatedmechanism is still lacking. Hence, although doped PZT-based ceramics have been explored for many years, thedoping mechanism remains debatable. Related substantialsupporting evidence, especially at the nanoscale, is rare.

In a previous work [24], a ternary solid solution systemconsisting of PZT and a complex perovskite-type com-pound, Pb(Zn1/3Nb2/3)O3 (PZN), was developed. The dis-covered composition of 0.2PZN–0.8PZT was near themorphotropic phase boundary (MPB), showing super pie-zoelectric properties. In the present paper, as an extensionto the research on doping behavior, CoCO3 was selected asa representative of dopants to modulate the microstructureand the electrical properties of the 0.2PZN–0.8PZT system.A systematic research was conducted to clarify the con-fused views on Co doping in previous works[17,23,25,26]. X-ray diffraction (XRD), X-ray photoelec-tronic spectroscopy (XPS), scanning electron microscopy(SEM) and high-resolution transmission electron micros-copy (HRTEM), in conjunction with energy-dispersive X-ray detection (EDX), were applied to investigate thedomain structure evolution and the liquid-phase sinteringbehavior induced by Co doping. Emphasis was placed onthe valence state distribution and the incorporation sitesof Co ions within the sample, as well as the formation ofsecondary phases when doping above the solubility limit.The influence of Co doping on electrical response (i.e.,dielectric relax behavior, ferroelectric and piezoelectricproperties) was also investigated and analyzed in correla-tion with the microstructure.

2. Experimental procedure

The specimens were prepared using a conventionalmixed-oxide process. The compositions used in this studyare as follows: Pb(Zn1/3Nb2/3)0.20(Zr0.50Ti0.50)0.80O3

+ xCoCO3, where x = 0–1.5 wt.%. Reagent-grade oxidepowders (Pb3O4, ZrO2, TiO2, ZnO, Nb2O5 and CoCO3)were used as starting materials. The powders were weighedand mixed by ball milling in alcohol for 24 h, with partiallystabilized zirconia balls as media. After drying, the mixturewas calcined in a covered alumina crucible at 850 �C for2 h. The calcined powders were remilled for 24 h and thenpressed into disks 11.5 mm in diameter at �100 MPa. Thegreen disks were sintered at 1000 �C for 2 h in a sealed alu-mina crucible. To minimize PbO loss, a PbO-rich atmo-sphere was maintained by placing powders of PbZrO3

used as packing powders inside the crucible.Specimen density was measured by the Archimedes

method. The crystal structures of the samples were exam-ined by XRD (Bruker D8 Advance, Karlsruhe, Germany)in the h–2h configuration using Cu Ka radiation. Thevalence state of the Co ions was determined by XPS

(ESCALAB 250, Thermo Electron, Waltham, MA).Micromorphology was detected on a thermally etched sur-face by SEM (Hitachi S4800, Japan). The mean grain sizewas calculated by the line intercept method. HRTEM wascarried out using an instrument (Hitachi FEI Tecnai F20,Japan) equipped with an EDX detector operated at anaccelerating voltage of 200 kV. Specimens for the TEMstudies were prepared by the standard procedure ofmechanical thinning, dimpling and ion milling towardselectron transparency before TEM observation. Ar+ ionbeam milling was applied to prepare the specimens, usinga 5 kV potential combined with a gun current of 2.2 mAat an incidence angle of 15� to the rotating sample.

To measure the electrical properties, silver paste wascoated on both sides of the sintered pellets and then firedat 560 �C for 30 min to form electrodes. The dielectricproperty and its dependence on temperature were mea-sured using a multi-frequency inductance capacitance resis-tance (LCR) analyzer (Agilent E4980A, Santa Clara, CA)with an automated temperature controller. Ferroelectricbehavior was studied using a ferroelectric tester (PremierII, Radiant Technologies Inc, Albuquerque, NM) at1 Hz. Prior to the testing of piezoelectric properties, thespecimens were poled in a silicone oil bath at 120 �C byapplying a DC field of 35 kV cm�1 for 30 min and thenaged for 24 h. The piezoelectric constant d33 was measuredusing a piezoelectric d33 meter (ZJ-6A, Institute of Acous-tics, Academic Sinica, China) at 100 Hz. The electrome-chanical coupling factor kp was determined by a precisionimpedance analyzer (4294A; Agilent Technologies, SantaClara, CA) through the resonance–anti-resonance methodbased on IEEE standards.

3. Results and discussion

3.1. Crystal structure and microstructure

The XRD patterns of 0.2PZN–0.8PZT ceramics withdifferent CoCO3 additions in the 2h range of 20–60� are

M.-P. Zheng et al. / Acta Materialia 61 (2013) 1489–1498 1491

shown in Fig. 1a. A complete perovskite structure isformed, and no detectable traces of the pyrochlore or otherimpurities are observed as the CoCO3 addition increases.Further observation of the XRD patterns reveals obviouschanges in diffraction peaks near 2h = 45�, indicating thatCo doping induces a structural transition. To determinethe phase evolution, fine scanning was carried out in thediffraction angle range 2h = 43–46�, and the results areshown in Fig. 1b. The diffraction peaks are separated by fit-ting the Gaussian–Lorentz line shape, and the positions ofthe reflections are fixed using the least squares method. Ingeneral, the reflections at 45� can be divided into threepeaks, corresponding to tetragonal (002), rhombohedral(200) and tetragonal (200). As shown in Fig. 1b, as theaddition of CoCO3 increases, the reflection intensity ofrhombohedral (20 0) decreases compared with that oftetragonal (200) and (002), revealing an increasing trendin the tetragonal phase. To quantitatively investigate theeffect of CoCO3 addition on phase transformation, thetetragonal phase content (TP) was calculated using Eq. (1):

TP% ¼ I ð200ÞT þ I ð002ÞT

I ð200ÞT þ I ð002ÞT þ I ð200ÞR� 100% ð1Þ

where I(200)R is the integral intensity of rhombohedral(200) reflection, and I(002)T and I(200)T are the integralintensities of tetragonal (002) and (200) reflections, respec-tively [27]. The results of the calculation using Eq. (1) areshown in Fig. 2. For the undoped specimen, the contentof the tetragonal phase is �55%, and very close amountsof rhombohedral and tetragonal phases imply the presenceof MPB. The tetragonal phase fraction increases abruptlywith the increase in CoCO3 addition. For the CoCO3 addi-tion of 0.2 wt.%, the content of tetragonal phase is near74%. When the addition exceeds 0.2 wt.%, the change inthe content of the tetragonal phase becomes inconspicuous.Moreover, the lattice parameters and tetragonality c/a ofthese compounds were calculated, and the results areshown in Table 1 and Fig. 2. It can be seen from Fig. 2 thatboth curves show similar variation trends, and the turning

Fig. 2. Content of the tetragonal phase and the lattice tetragonality c/a vs.CoCO3 addition.

point appears at the same position of 0.2 wt.% CoCO3

addition. These results suggest that the solubility limit ofCoCO3 in the 0.2PZN–0.8PZT system may be near0.2 wt.%.

Phase transition is related to the evolution of the ferro-electric domain structure, which can be directly visualizedby TEM. Fig. 3a and b shows the bright-field TEM imagesof the undoped and the 1.2 wt.% CoCO3 added 0.2PZN–0.8PZT, respectively. Based on the classic view of thedomain theory, rhombohedral PZT has polarization vec-tors that intersect the domain walls at 71� and 109�, givingwalls lying on the {110}p and {100}p planes [28], whereasthe polarization vectors in the tetragonal PZT intersect thedomain wall at 90�, giving domain walls lying on the{110}p plane. In Fig. 3a, both the lamellar tetragonaldomains and the lenticular rhombohedral domains areclearly observed. However, the rhombohedral domainsare scattered in the tetragonal domain matrix, supportingthe nature of the “two-phase zone” of the MPB structurein the undoped 0.2PZN–0.8PZT [29]. However, inFig. 3b, only lamellar tetragonal domains can be observed,and the corresponding domain size greatly increases com-pared with that of the undoped specimen [30,31], indicatingthat the Co dopant induces phase transition on the tetrag-onal side. These results are in accordance with the previousXRD analysis.

Generally, Co ions enter the B-site of the perovskite lat-tice to replace the Ti4+ and Zr4+ ions, because the radius ofCo2+ (0.665 A) or Co3+ (0.525 A) is close to that of Zr4+

(0.720 A) or Ti4+ (0.605 A) ions. Co ions are reported toexist mainly in the +2 or +3 valence states in oxides orhydroxides [32]. To clarify the exact valence state of Coions in 0.2PZN–0.8PZT ceramics, the XPS spectra of Co2p3/2 were recorded. The Co2p3/2 band were partitionedinto two sub-bands centered at 780.4 and 781.4 eV, usingthe standard Gaussian fitting method. The two decoupledsub-bands are shown in Fig. 4a. To quantitatively investi-gate the valence transformation of Co ions in differentadditions, the relative content of Co3+ was calculatedaccording to the fitting results, as shown in Fig. 4b. As seenfrom the figure, the Co3+ concentration obviously changeswith CoCO3 addition. For the ceramics with 0.2 wt.%CoCO3 addition, the mole ratio of Co3+ ion to Co2+ ionis 92:8, which suggests that the dominant valence state inthe current sample is Co3+. However, further additionresults in a significant decrease in Co3+ concentration.When the addition of CoCO3 is 0.5 wt.%, the Co3+ concen-tration is near 60 mol.%, which is near unity for more addi-tion. From these results, Co ions can be concluded to existin the mixed valence forms of +2 and +3, and their relativeratio depends on doping content. Generally, high-tempera-ture sintering leads to a reduced atmosphere, which isfavorable to the formation of low-valence ions [33]. Com-pared with ions occupying the crystal lattice, dopants exist-ing in the grain boundary are more sensitive to the outersintering atmosphere. In the present work, the solubilitylimit of CoCO3 in the 0.2PZN–0.8PZT matrix is near

Table 1Content of the tetragonal phase and the lattice tetragonality c/a for0.2PZN–0.8PZT ceramics with different CoCO3.

Composition(wt.%)

Latticeparameters

Latticetetragonality (c/a)

Tetragonal phaseTP (%)

a (A) c (A)

0.0 3.9873 4.0156 1.0071 55.460.1 3.9961 4.0333 1.0093 70.030.2 3.9976 4.0392 1.0104 73.870.3 4.0015 4.0427 1.0103 73.950.5 4.0091 4.0512 1.0105 74.680.8 3.9954 4.0366 1.0103 75.441.0 3.9938 4.0345 1.0102 76.331.2 3.9978 4.0394 1.0104 76.081.5 4.0178 4.0471 1.0073 74.44

Fig. 3. Bright-field TEM domain image of 0.2PZN–0.8PZT ceramics withdifferent CoCO3 additions: (a) 0.0 wt.%; (b) 1.2 wt.%. (Rh., rhombohe-dral; Tet., tetragonal).

Fig. 4. (a) XPS spectra of Co 2p3/2 of 0.2PZN–0.8PZT ceramics withdifferent CoCO3 additions and (b) dependence of Co3+ concentration onCoCO3 addition.


0.2 wt.%. Below the solubility limit, the Co ions enter thecrystal lattice, and most of them exit with +3 valence.Compared with Co2+ (0.665 A), Co3+ (0.525 A) has a rela-tively large radius difference from Zr4+ (0.720 A) and Ti4+

(0.605 A) ions. Thus, the related substitution results in alarge distortion of the unit cell, as shown in Fig. 2. Above

the solubility limit, excess Co ions gather in the grainboundaries. The effect of the surroundings reducing theatmosphere causes part of the Co ions to transform intothe low valence of +2, which shows little effect on latticedistortion. In summary, cation radius, which is related toCo valence state, determines the variation in tetragonalityc/a.

Fig. 5 shows the dependence of the apparent density andthe relative density of 0.2PZN–0.8PZT on CoCO3 addi-tion. Clearly, a maximum in the density curve appears.When the doping is <0.2 wt.%, density sharply increasesas doping increases. However, further addition of CoCO3

causes a decrease in ceramic density. The variation trendin the curve indicates that the suitable CoCO3 additionbenefits the densification of ceramics during the sinteringprocess. Fig. 6 shows the SEM micrographs of the ther-mally etched surface of 0.2PZN–0.8PZT ceramics with dif-ferent CoCO3 additions. Grain size remarkably increaseswith CoCO3 addition. As stated earlier, below the solubil-ity limit of 0.2 wt.%, Co ions, mostly with +3 valence, enterthe perovskite lattice to replace the Ti4+ or Zr4+ site.

Fig. 5. Dependence of apparent density and relative density of 0.2PZN–0.8PZT on CoCO3 addition.


Therefore, the imbalance in ion valence leads to the crea-tion of oxygen vacancies, which enhance the transfer ofmass and energy between reactants, thus improving the sin-tering behavior and inducing an increase in grain size.

Fig. 6. SEM micrographs of specimens with different CoCO3 additions: (a) 0.

Interestingly, grain size still shows an increasing trend, withfurther increase in CoCO3 addition above the solubilitylimit. This phenomenon is different from previous observa-tion in other transitional metal ion (i.e., Cr3+ [12], Mn2+

[14]) doped 0.2PZN–0.8PZT systems. In these doped sys-tems, grain growth is inhibited, as the addition of dopantsexceeds the solubility limit. The so-called solid solutionimpurity drag mechanism has been proposed to explainthe inhibitor role of excess doping ions aggregated at thegrain boundary. Thus, in this case, what is the exact mech-anism responsible for grain growth when CoCO3 additionis above the solubility limit? To obtain direct evidence atthe nanoscale, the 0.2PZN–0.8PZT sample doped with1.2 wt.% CoCO3 was studied using TEM. Fig. 7a showsthe bright-field TEM image of triple pockets and grainboundary regions. The triple pockets reveal a slightly con-cave shape, indicating that grain-boundary wettingoccurred. Fig. 7b gives the HRTEM imaging of the inter-face region between PZN–PZT grains. The distinct latticefringe confirms the good crystallinity of the PZN–PZTgrains. The interplanar spacing is �0.401 nm, correspond-

0 wt.%; (b) 0.2 wt.%; (c) 0.5 wt.%; (d) 0.8 wt.%; (e) 1.0 wt.%; (f) 1.5 wt.%.

Fig. 8. Grain boundary movement and solution–reprecipitation betweensmall and large grains.


ing well to the spacing of (100). Moreover, the presence ofan amorphous phase with thickness �2–3 nm can beclearly observed, which is direct evidence of liquid-phasesintering. Fig. 7c and d shows the EDX analysis of the tri-ple pockets and grains, respectively. The Co ions are diffi-cult to detect in the matrix grains. The concentration isassumed below the detection limit of the EDX systembecause of the low doping level in the sample. Compara-tively, a relatively high level of Co concentration wasdetected in the triple pocket regions, combined with excesslead content. Some PbO from the packing PbZrO3 powderpossibly transfer to the PZN–PZT compacts during the sin-tering treatment [34], and the excess Co ions accelerate theformation of the PbO-rich liquid phase. A model fordescribing the grain boundary movement in liquid-phasesintering is illustrated in Fig. 8. The energy of a grainboundary between two grains is known to depend on theirdihedral angle [22]. In the present case, the PbO-rich liquidphase wets and covers the surface of the grains. Grainboundaries with a liquid layer have abundant dihedralangles at the junctions, which can be easier to move com-pared with grain boundaries without a liquid layer. There-fore, the small particles dissolve gradually in the liquid andthen reprecipitate on the larger grains, resulting in largegrain growth based on the liquid-phase sintering mecha-nism [35]. In the final stage of sintering, an amorphous res-idue enriched with cobalt and lead segregates at the grainboundaries and is deposited at the triple grain junctions.

Fig. 7. (a) TEM micrograph of 1.2 wt.% CoCO3-doped 0.2PZN–0.8PZT spegrains; and EDX spectra of (c) the triple junction and (d) the grain. (The detecrespectively.)

3.2. Dielectric, ferroelectric and piezoelectric properties

Fig. 9 shows the temperature dependence of the dielec-tric constant of 0.2PZN–0.8PZT ceramics with differentamounts of CoCO3 addition. All patterns show weak fre-quency dispersions, but the shapes of the curves near theCurie peak present a narrow trend with the addition ofCoCO3, revealing that the system changes from relaxorto normal ferroelectrics.

For a classical ferroelectric, the dielectric constant abovethe Curie temperature follows the Curie–Weiss law:

e ¼ C=ðT� T cÞðT > T cÞ ð2Þ

cimen; (b) HRTEM imaging of the interface region between PZN–PZTted Cu and Fe peaks are caused by the sample holder and the microscope,

Fig. 10. Plots of the inverse dielectric constant as a function oftemperature at 1 kHz for 0.2PZN–0.8PZT ceramics with different CoCO3

additions: (a) 0.0 wt.%; (b) 0.4 wt.%; (c) 0.6 wt.%; (d) 0.8 wt.% (solid line,fitting to the Curie–Weiss law).

Table 2Temperature of the maximum dielectric constant (Tm), temperature of thedielectric constant that follows the Curie–Weiss law (Tcw), and deviation(DTcm) for 0.2PZN–0.8PZT ceramics with different CoCO3 additions at1 kHz.

Samples (wt.%) 0.0 0.4 0.6 0.8

Tm (�C) 321 320 317 313Tcw (�C) 391 383 371 363DTcm (�C) 70 63 54 50


where C is the Curie–Weiss constant, and Tc is the Curie–Weiss temperature.

Fig. 10 shows the plots of the inverse dielectric constantvs. temperature at 1 kHz. The fitting results obtained byEq. (2) are listed in Table 2. The deviation from theCurie–Weiss law can be defined by DTcm:

DT cm ¼ T cw � T m ð3Þwhere Tcw denotes the temperature from which the dielec-tric constant starts to deviate from the Curie–Weiss law,and Tm represents the temperature at which the dielectricconstant reaches a maximum.

The dielectric constants of undoped ceramics obey theCurie–Weiss law at temperatures much higher than Tm,and the calculated DTcm value is 70 �C. However, withthe increase in CoCO3 addition, the DTcm of the specimensdistinctly shows a descending trend. At 0.8 wt.% CoCO3

addition, the DTcm value is 50 �C, implying that the diffusephase transition (DPT) behavior has weakened with theaddition of CoCO3.

For the relaxor ferroelectrics, the reciprocal of thedielectric constant and the temperature obey the Uchinoand Nomura function, a modified Curie–Weiss law [36]:

1=er � 1=emax ¼ ðT� T maxÞc=C ð4Þwhere emax is the maximum value of the dielectric constant,er is the dielectric constant at temperature T, Tmax is thetemperature at the peak of the dielectric constant, C isthe Curie constant, and c is the indicator of the degree ofdiffuseness, taking the values between 1 (normal ferroelec-tric) and 2 (complete DPT).

To further confirm the effects of CoCO3 doping on therelaxor feature of 0.2PZN–0.8PZT ceramics, the plots ofln(1/er � 1/emax) as a function of ln(T � Tmax) were drawnfor the 0.2PZN–0.8PZT specimens with different amountsof CoCO3 additive, as shown in Fig. 11. A linear relation-ship is observed in all specimens. The slope of the fitting

Fig. 9. Temperature dependence of the dielectric constant at 1, 10, 100and 1000 kHz for 0.2PZN–0.8PZT ceramics with different CoCO3

additions: (a) 0.0 wt.%; (b) 0.4 wt.%; (c) 0.6 wt.%; (d) 0.8 wt.%,respectively.

curves is used to determine the c value. With the increasein CoCO3 addition, the c value decreases from 1.94 to1.72, further confirming that the CoCO3 additive can lowerthe DPT behavior of 0.2PZN–0.8PZT.

For a lead-based relaxor, such as Pb(Zn1/3Nb2/3)O3

(PZN), Pb(Mg1/3Nb2/3)O3 (PMN) and Pb(Ni1/3Nb2/3)O3

(PNN), the relaxor feature appears when at least two cat-ions with different valences occupy similar crystallographicB sites of the oxygenic octahedral center, giving rise to ran-dom fields that impede the development of long-rangepolar ordering. For a Co-doped 0.2PZN–0.8PZT system,the doping of Co ions can enhance the positional disorderin the B-site ions of ABO3 perovskite, which is speculatedto enhance DPT behavior. However, this prediction isinconsistent with the observation that the degree of DPTdecreases with the addition of CoCO3. A relaxor featureis directly related to domain configuration in ferroelectrics.Based on the previous XRD and TEM results, Co dopantinduces a gradual phase transition process from MPB tothe tetragonal side, and the corresponding tetragonaldomain size shows an increasing trend. For the MPB com-

Fig. 11. ln(1/er � 1/emax) as a function of ln(T � Tmax) at 1 kHz for0.2PZN–0.8PZT ceramics with different CoCO3 additions (symbols,experimental data; solid line, fitting to Eq. (3)).

Fig. 12. (a) Ferroelectric hysteresis loops of CoCO3-doped 0.2PZN–0.8PZT ceramics measured at room temperature, and (b) dependence ofthe remnant polarization Pr and coercive field Ec at room temperature onCoCO3 addition.


position of the undoped specimen, the domain structureconsists of the rhombohedral and tetragonal phases, whichare superimposed on each other and exhibit large disper-sion in phase transition temperatures Tc in different areas.Thus, the spatial fluctuations in the local Curie tempera-ture lead to the increase in degree of diffuseness [37]. Withthe addition of CoCO3, the tetragonal domain sizeincreases, and the induced merging of micropolar regionsinto macropolar regions is responsible for the transitionfrom relaxor to normal ferroelectric behaviors.

Moreover, as shown in Fig. 9, the dielectric constant er

shows an enhancing trend with the addition of CoCO3,especially in the region near the Curie temperature [15].For the undoped specimen, the emax is merely 6000 at theCurie peak. With increasing CoCO3 content, emax rapidlyincreases. At 0.8 wt.% CoCO3, emax is 14,600, �2.5 timeshigher than that of the undoped specimen. As a “hard”

dopant, Co doping in 0.2PZN–0.8PZT can lead to the cre-ation of oxygen vacancies, owing to the valence differencefrom the B site ion, which tend to form defective dipolesD, producing an inner bias field Ei. In turn, the build-upof Ei reduces polarization rotations and dielectric proper-ties, especially by prohibiting domain wall motion [14,38].However, this assumption is inconsistent with the experi-mental results. To thoroughly clarify the doping mecha-nism, the ferroelectric hysteresis loops of Co-doped0.2PZN–0.8PZT were measured, the results of which areshown in Fig. 12a. As the figure shows, all P–E hysteresisloops are saturated, and the dissymmetrical shape is thetypical feature of the existence of the internal-bias field(Ei). Fig. 12b shows the remanent polarization Pr and coer-cive field Ec as a function of CoCO3 addition at room tem-perature. Interestingly, Pr increases, whereas Ec decreaseswith the increase in CoCO3 addition. Moreover, in theP–E hysteresis loops measured at elevating temperatures,the variation trend of the loops shows features similar to

those measured at room temperature (see Fig. S1 in Sup-porting information). This phenomenon is a departurefrom the hard doping nature of Co ions. It is known thatlarge grain size favors the improvement in dielectric andferroelectric properties because, as the grain size decreases,the grain boundary phases, which are directly related to thevolume of the space charge regions, are enlarged. Accord-ingly, a strong coupling between the grain boundaries andthe domain walls hinders domain switching. This mecha-nism related to grain size is called the size effect. In thiscase, the addition of CoCO3 leads to the remarkablegrowth of the grains (Fig. 6). Thus, the development ofdomains becomes more consummate, neutralizing the harddoping effect and reforming the er of the ceramic sample.

Fig. 13 shows the variation in piezoelectric constant d33

and electromechanical coupling factor kp as a function ofCoCO3 addition. Both kp and d33 show a similar variationwith increasing CoCO3 content. The optimized values of kp

(0.66) and d33 (310 pC N�1) were obtained for 0.8 wt.%CoCO3 content, and they are superior to the reported val-ues (kp = 0.50 and d33 = 278 pC N�1) in a Li2CO3-doped

Fig. 13. Piezoelectric constant d33 and electromechanical coupling factorkp for 0.2PZN–0.8PZT ceramics with different CoCO3 additions.


PZN–PZT system [13]. The evolution of piezoelectricresponses can be explained by the conventional Landau–Devonshire relation:

d33 ¼ 2ere0Q11P r ð5Þwhere e0 is the vacuum permittivity, Q11 is the electrostric-tion coefficient, and Pr is the remanent polarization. Evi-dently, the steep rise in Pr and er contributes to theincrease in d33. Notably, the best piezoelectric propertiesare obtained for 0.8 wt.% CoCO3 content, which deviatesfrom the composition near the solubility limit of0.2 wt.%. This behavior is different from that in previousreports in which super piezoelectric properties are obtainedat the solubility limit in a transitional metal ion-dopedPZN–PZT system [11,12,14,39]. In this work, the combinedeffects of hard doping and microstructure evolution in-duced by CoCO3 addition determine the variation in piezo-electric properties. Further addition above 0.8 wt.% causesa decrease in piezoelectric properties, which may be attrib-uted mainly to the formation of abnormal grain bound-aries, as affirmed by TEM observations.

4. Conclusion

CoCO3-added PZN–PZT ceramics were preparedthrough the conventional solid oxide process. The valencestate and the incorporation site of Co cations significantlyaffect the microstructure and the electrical properties of theceramics. XPS results demonstrate that both Co3+ andCo2+ cations coexist in the system. Below the solubilitylimit of 0.2 wt.% CoCO3, most Co ions exist in the +3valence, which prefer the entry to B sites of the perovskitelattice and the phase transition from MPB to the tetragonalside, resulting in the weakness of the DPT behavior. Abovethe solubility limit, excess Co ions segregate at the grainboundaries, which favor the formation of the Co–Pb-richphase and accelerate grain growth for the liquid sinteringmechanism. Despite the presence of an internal-bias field

and a high concentration of tetragonal phase, CoCO3-doped ceramics exhibit superior piezoelectric propertiescompared with the undoped samples. This result suggeststhat the size effect is stronger than the reverse hardeningeffect of the acceptor dopant. Optimum piezoelectric prop-erties were obtained for the 0.8 wt.%-doped 0.2PZN–0.8PZT ceramic system: d33 = 310 pC N�1 and kp = 0.66.

Acknowledgements

This work was supported by the National Natural Sci-ence Foundation of China (Grant Nos. 51072008 and51172006), the Natural Science Foundation of Beijing(No. 2102006), the Project of New Star of Science andTechnology of Beijing (Grant No. 2007A014) and theFunding Project of IHLB (Grant Nos. PHR201008012and PHR201007101).

Appendix A. Supplementary material

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.actamat.2012.11.026.

References

[1] Cross LE. Ferroelectrics 1994;151:305.[2] Yoon CB, Lee SH, Lee SM, Kim HE. J Am Ceram Soc 2004;87:1663.[3] Lee SH, Yoon CB, Lee SM, Kim HE. J Eur Ceram Soc 2006;26:111.[4] Noheda B, Cox DE, Shirane G, Guo R, Jones B, Cross LE. Phys Rev

B 2001;6301:014103.[5] Zhu ZG, Zheng NZ, Li GR, Yin QR. J Am Ceram Soc 2006;89:717.[6] Chang LM, Hou YD, Zhu MK, Yan H. J Appl Phys

2007;101:034101.[7] Zhang HB, Jiang SL, Zeng YK. Appl Phys Lett 2008;93:192901.[8] Yao X, Chen Z, Cross LE. J Appl Phys 1983;54:3399.[9] Kungl H, Hoffmann MJ. J Appl Phys 2010;107:054111.

[10] Gao F, Wang CJ, Liu XC, Tian CS. Ceram Int 2007;33:1019.[11] Zhu MK, Lu PX, Hou YD, Wang H, Yan H. J Mater Res

2005;20:2670.[12] Hou YD, Lu PX, Zhu MK, Song XM, Tang JL, Wang B, et al. Mater

Sci Eng B – Solid State Mater Adv Technol 2005;116:104.[13] Hou YD, Chang LM, Zhu MK, Song XM, Yan H. J Appl Phys

2007;102:084507.[14] Hou YD, Zhu MK, Gao F, Wang H, Wang B, Yan H, et al. J Am

Ceram Soc 2004;87:847.[15] Yoon SJ, Joshi A, Uchino K. J Am Ceram Soc 1997;80:1035.[16] Hammer M, Hoffmann MJ. J Am Ceram Soc 1998;81:3277.[17] Hu HC, Zhu MK, Xie FY, Lei N, Chen J, Hou YD, et al. J Am

Ceram Soc 2009;92:2039.[18] Lucke K, Detert K. Acta Metall 1957;5:628.[19] Cahn JW. Acta Metall 1962;10:789.[20] Gottstein G, Molodov D. Interface Sci 1998;6:7.[21] Lee YC, Huang YL. J Am Ceram Soc 2009;92:2661.[22] Seo CE, Yoon DY. J Am Ceram Soc 2005;88:963.[23] Zang GZ, Yi XJ, Du J, Wang YF. Mater Lett 2010;64:1394.[24] Lee SM, Yoon CB, Lee SH, Kim HE. J Mater Res 2004;19:2553.[25] Xu Q, Chen M, Chen W, Liu HX, Kim BH, Ahn BK. Acta Mater

2008;56:642.[26] Beck CM, Thomas NW, Thompson I. J Eur Ceram Soc 1998;18:1679.[27] Yamaguchi H. J Am Ceram Soc 1999;82:1459.[28] Randall CA, Barber DJ, Whatmore RW. J Mater Sci 1987;22:925.




[29] Fan HQ, Jie WQ, Tian CS, Zhang LT, Kim HE. Ferroelectrics2002;269:33.

[30] Jin L, He ZB, Damjanovic D. Appl Phys Lett 2009;95:012905.[31] Randall CA, Kim N, Kucera JP, Cao WW, Shrout TR. J Am Ceram

Soc 1998;81:677.[32] Cui ML, Zhu J, Zhong XY, Zhao YG, Duan XF. Appl Phys Lett

2004;85:1698.[33] Kirianov A, Ozaki N, Ohsato H, Kohzu N, Kishi H. Jpn J Appl Phys

2001;40:5619.

[34] Hardtl KH, Rau H. Solid State Commun 1969;7:41.[35] Jones DRH. Philos Mag 1973;27:569.[36] Uchino K, Nomura S. Ferroelectrics 1982;44:55.[37] Wu N, Song X, Hou Y, Zhu M, Wang C, Yan H. Chin Sci Bull

2009;54:1267.[38] Fan HQ, Park GT, Choi JJ, Kim HE. Appl Phys Lett 2001;79:1658.[39] Hou YD, Zhu MK, Wang H, Wang B, Yan H, Tian CS. Mater Sci

Eng B – Solid State Mater Adv Technol 2004;110:27.

Documents

Effect of valence state and incorporation site of cobalt