Journal of Alloys and Compounds 592 (2014) 140–149

Contents lists available at ScienceDirect

Journal of Alloys and Compounds

journal homepage: www.elsevier .com/locate / ja lcom

Novel monoclinic zirconolite in Bi2O3–CuO–Ta2O5 ternary system: Phaseequilibria, structural and electrical properties

0925-8388/$ - see front matter � 2014 Published by Elsevier B.V.http://dx.doi.org/10.1016/j.jallcom.2013.11.197

⇑ Corresponding author. Tel.: +60 3 8946 7491; fax: +60 3 8943 5380.E-mail address: tankb@science.upm.my (K.B. Tan).

K.B. Tan a,⇑, M.P. Chon a, C.C. Khaw b, Z. Zainal a, Y.H. Taufiq Yap a, P.Y. Tan a

a Department of Chemistry, Faculty of Science, Universiti Putra Malaysia, 43400 Serdang, Selangor, Malaysiab Department of Mechanical and Material Engineering, Faculty of Engineering and Science, Universiti Tunku Abdul Rahman, 53300 Setapak, Kuala Lumpur, Malaysia

a r t i c l e i n f o a b s t r a c t

Article history:Received 2 September 2013Received in revised form 27 November 2013Accepted 27 November 2013Available online 2 December 2013

Keywords:CeramicsSolid state reactionDielectric responseac Impedance spectroscopy

Synthesis of novel monoclinic zirconolite, Bi1.92Cu0.08(Cu0.3Ta0.7)2O7.06 (b-BCT) using solid state reactionhad been finalised at the firing temperature of 900 �C over 24 h. The X–ray diffraction pattern of b-BCTwas fully indexed on a monoclinic symmetry, space group, C2/c with lattice constants, a = 13.1052 (8),b = 7.6749 (5), c = 12.162 (6), a = c = 90� and b = 101.32� (1), respectively. The reaction mechanism studyindicated phase formation was greatly influenced by the reaction between intermediate bismuth tanta-late binary phases and CuO at elevated temperatures. b-BCT was thermally stable up to a temperature of900 �C and contained spherulite grains with sizes ranging from 1 to 14 lm. Electrical properties of thismaterial were characterised over a broad temperature range covering temperatures from 10 K to874 K. At the temperature of 304 K, two semicircles were discernible in complex Cole–Cole plot showingan insulating grain boundary with Cgb = 6.63 � 10�9 F cm�1 and a bulk response capacitance,Cb = 6.74 � 10�12 F cm�1. The Power law frequency-dependent ac conductivity of b-BCT was apparentin three frequency regimes; a low–frequency plateau regime, a high-frequency plateau regime and adispersive regime taking place in the temperature range of 220–576 K. The frequency-dependent acconductivity of b-BCT with increasing temperature was attributed to the thermal activated electricalconduction mechanism within the structure.

� 2014 Published by Elsevier B.V.

1. Introduction

Much research impetus had been focused on a wide spectrum offunctional materials covering energy storage, solid electrolytes,superconductors, glasses and composites. Oxides had often beenrendered for such purposes owing to their exploitable propertiesthat may vary depending on structure, composition, defect and pro-cessing control. Pyrochlore oxide with a unique structure, had beensynonymous with the discovery of an unusual mineral,(Ca, Na)2(Nb, Ta)2(OH, F)7 as early as in 1826 [1,2]. A comprehensivereview concerning the families of pyrochlores could be found in [3].

Despite controversial explanations as to the structuralcharacteristics of pyrochlores, it was generally believed that thesematerials had a general formula of A2B2O6O0 which comprised ofA3þ

2 B4þ2 X7;A

2þ2 B5þ

2 X7 or a required average mixed valency type. Itwas worth highlighting that pyrochlores could crystallise intovarious forms including rhombohedral, hexagonal, triclinic,orthorhombic and the most commonly studied cubic phase [3–6].Pyrochlores had distinct coordination networks: an eight-foldcoordinated A site for large cations and a six-fold coordinated B sitefor small cations within their respective cation–anion bonds. This

gave a strong indication as to the reason why pyrochlore systemswere commonly found to contain heavy elements at the A site,e.g. Bi and Pb, if their ionic sizes were taken into consideration[3,7]. By far, the most extensively studied were, bismuth-basedpyrochlores due to their highly diverse applications. The adventof pyrochlores with better performance was always motivated bytheir wide variety of properties and this brought hundreds of com-positions prepared in various systems, e.g. (Bi1:5M0

0:5)ðM00:5M00

1:5ÞO7

and Bi2ðM02=3M00

4=3Þ2O7 families of pyrochlores with divalent/triva-lent cations, with M0 = Zn, Mg, Ni, Ca, Cu and Fe, and even pentava-lent cations, with M00 = Nb, Ta and Sb [8–16]. Meanwhile,pyrochlores containing trivalent metal cations at the B siteaccounted for Bi2MNbO7 (M3+ = Fe, In and Sc) systems [6,15,17]and, these did not include many other compositions preparedthrough chemical doping and/or different methods [3,18,19].

Cubic pyrochlores (a-phases) in ternary systems, (Bi1.5Zn0.5)(Zn0.5M00

1:5)O7 (M00 = Nb and Ta) had been demonstrated as gooddielectrics for multilayer ceramic capacitors (MLCC) and micro-wave applications [8,20,21]. Attempts to describe the excellentdielectric properties found in these a-phases were made fromthe perspective views of their local structure, non-stoichiometryin composition and defects [6]. The compositional variables andself-compensating occupancy in pyrochlore structures made itextremely difficult to deduce a single satisfactory answer for their

Fig. 1. Indexed X-ray diffraction peaks of Bi1.92Cu0.08(Cu0.3Ta0.7)2O7.06 that crystal-lised in monoclinic structure, space group C2/c. Inset of figure shows schematicpresentation of the expanded zirconolite region with different mol% of Bi2O3, CuOand Ta2O5. Markers shown in the diagram are corresponded to mixture of twophases (open triangle), three phases (open circle) and zirconolite-like phase (closedsymbol).

K.B. Tan et al. / Journal of Alloys and Compounds 592 (2014) 140–149 141

electrical behaviour; however, researchers in this field at least havea fair understanding about this now [6,20,21]. On the other hand,distorted pyrochlores with pseudo-orthorhombic (b-phase) struc-ture in Bi2(Zn2/3M00

4=3)2O7 (M00 = Nb and Ta) systems were reportedto have slightly lower dielectric constants and temperature coeffi-cients of opposite sign if compared to the a-phases [20–23]. How-ever, this b-phase was later confirmed by careful structural studiesas a monoclinic phase, space group, C2/c, No. 15, having a structuresimilar to that of CaZrTi2O7 [24,25]. A detailed structural descrip-tion revealed that the b-phase was of the zirconolite type in whichalternating layers of Bi atoms in parallel to [110] direction were lo-cated between hexagonal tungsten bronze (HTB)-like (001) sheetsof ZnNbO6 octahedra containing Zn in the centres of its sixmembered rings [24]. The slightly off-centre Zn could be distrib-uted between two half occupied (5 + 1) fold coordinated sites nearthese rings and/or found together with the smaller Nb in a 6-foldcoordination environment. Meanwhile, the larger Bi preferentiallyremained at a coordination number of seven or eight [24].

Given the close similarity in terms of ionic radii and electronicshell configurations, Cu substituted a-phases in an analogousBi2O3–CuO–Ta2O5 (BCT) ternary system were successfullysynthesised and characterised [7,11]. Common resemblances wereobserved in our phase diagram studies [11,26–28], as a-phases inboth the Zn- and Cu-analogues could form over a broad subsolidusarea except the latter was found at a slightly lower bismuth region.The ternary phase diagram in the Zn-analogue had used an a-phaseof nominal stoichiometry, (Bi1.5Zn0.5)(Zn0.5Ta1.5)O7 as the keycomposition for the explanations as to subsolidus mechanisms.Meanwhile, the structurally related b-phase, Bi2(Zn1/3Ta2/3)2O7

appeared to have a variable composition that, as yet, had not beenwell characterised [28]. On the contrary, both a- and b-phases inthe Cu-analogue were not single phase materials from which theircompositions had departed slightly from their ideal stoichiometry.With regard to their electrical properties, the a-phase in the Cu-analogue had relatively higher conductivities and lower activationenergies than that of Zn-analogue. But, the a-phase in the Zn ana-logue had demonstrated a higher dielectric constant, lower dielec-tric loss and a relaxor-like behaviour that, again, was not observedin its b-phase [20]. At this juncture, several questions immediatelyarise as to the existence of b-phase in the Cu-analogue and, if so, theelectrical properties of the b-phase could be worthwhile for furtherinvestigations. In order to better understand this phase, the aims ofthis paper are to discuss our findings concerning the synthesis,reaction pathway and ac electrical response of this novel b-phase,Bi1.92Cu0.08(Cu0.3Ta0.7)2O7.06.

2. Experimental procedures

Synthesis of monoclinic zirconolite, Bi1.92Cu0.08(Cu0.3Ta0.7)2O7.06 was carried outby conventional solid state reaction at the firing temperature of 900 �C over 24 h.High purity starting materials, Bi2O3 (99.9%, Acros Organics), CuO (99.7%, AlfaAesar) and Ta2O5 (99.85%, Alfa Aesar) were preheated at the temperatures of300 �C (Bi2O3), 600 �C (CuO, Ta2O5) for 2 h before an approximately 3–4 g stoichi-ometric amount of the mixture was weighed out and homogenised with sufficientacetone in an agate mortar. The sample was fired and preheated in a platinum boatat the temperature of 800 �C over 24 h prior to firing at higher temperatures. Loss ofignition was determined and the phase purity was confirmed by X-ray diffractom-eter (XRD) using an automated Shimadzu Cu Ka radiation at a scan rate of 0.1�/minin the 2h range of 10–70�. The thermal properties of Bi1.92Cu0.08(Cu0.3Ta0.7)2O7.06

within a temperature range of �30–900 �C were examined by means of both ther-mogravimetric (TGA) and differential thermogravimetric (DTA) analysers from Per-kin Elmer. Elemental analyses of the triplicate samples were analysed usinginductively coupled plasma-atomic emission spectroscopy (ICP-AES), Perkin ElmerOptima 2000 DV. The microstructure of the as-fired pellet was studied using scan-ning electron microscopy (SEM), LEO 1455 VPSEM. The sintering condition of the b-BCT pellet for electrical measurement was optimised according to the careful pro-cedure as reported in our previous work [28]. For ac impedance measurement, pel-let of diameter, �8 mm and thickness of �2 mm was coated with platinumelectrodes on both surfaces. The impedance data that were normalised for the sam-ple’s geometric factor were further corrected for stray capacitance of an open jig.

The relative densities of the pellets used for electrical measurement and SEM anal-ysis were greater than 80% of the theoretical density. The electrical properties of b-BCT (above room temperature) were investigated using ac impedance analyser, HP4192 A (Tokyo, Japan) in the frequency range of 5 Hz–13 MHz over a temperaturerange of 304–874 K at an applied voltage of 0.1 V. A thermal equilibrium time of20 min was allowed for each measurement at temperature intervals of 50 �C in boththe heating and cooling cycles. Fixed-frequency low temperature electrical mea-surement (below room temperature) was performed by a precision LCR meter, Agi-lent E4980 A (Wokingham, UK) coupled with Oxford Cryostat (Abingdon, UK)Intelligent Temperature Controller (ITC 503C). The electrical data were collectedat fixed frequencies of 1 kHz, 10 kHz, 100 kHz and 1 MHz in a temperature rangeof 10–304 K with a temperature interval of 10 K, respectively.

3. Results and discussion

3.1. Novel phase in Bi2O3–CuO–Ta2O5 (BCT) ternary system

Composition with high bismuth content is highly firing temper-ature-dependent and careful synthesis control is thereforerequired, especially with the presence of low refractory copperoxide. Sample preparation is performed in a step-wise manner inwhich lower temperatures are used to allow complete reaction ofvolatile Bi2O3. In order to ensure thermal equilibrium, prolongedduration and/or increased firing temperature has been applied sothat, no further phase changes should be observed. Subsequently,the appropriate heat treatment condition for Bi1.92Cu0.08

(Cu0.3Ta0.7)2O7.06 is finalised at the firing temperature of 900 �Cover 24 h.

Careful X-ray analysis has confirmed Bi1.92Cu0.08(Cu0.3Ta0.7)2-

O7.06 as a single phase monoclinic zirconolite similar to thatreported in the literature [27,28]. The composition is found todepart slightly from the Zn analogue of an ideal stochiometry,Bi2(Zn1/3Ta2/3)2O7. The phase compatibility of this structure withadjacent phases and its exact compositional location are depictedin an expanded BCT phase diagram as shown in inset of Fig. 1.All the characteristic diffraction planes are fully indexed (Fig. 1,Table 1) based on monoclinic symmetry, space group C2/c withlattice constants, a = 13.1052 (8), b = 7.6749 (5), c = 12.162 (6),a = c = 90� and b = 101.32� (1), respectively. No significant cellexpansion is observed between the b-phases of Cu and Zn ana-logues, as this could probably due to their comparable ionic radii

Table 1Refined and indexed XRD diffraction planes of BCT monoclinc zirconolite.

h k l 2h (Observed) 2h (Calculated) D(2h)

0 0 2 14.782 14.857 �0.0751 1 1 16.026 16.008 0.018�2 0 2 18.123 18.203 �0.080�1 �1 2 19.027 19.041 �0.014

1 1 2 21.120 21.076 0.0442 0 2 22.235 22.241 �0.006�3 1 1 23.720 23.678 0.042�1 �1 3 24.951 24.969 �0.018�3 1 2 25.835 25.858 �0.023

3 1 1 26.140 26.162 �0.022�2 2 1 27.416 27.348 0.068

4 0 0 27.760 27.769 �0.009�4 0 2 28.885 28.865 0.020

0 0 4 29.952 29.971 �0.019�1 �1 4 31.666 31.671 �0.005�2 �2 3 33.601 33.535 0.066

1 1 4 34.228 34.236 �0.008�5 1 1 36.184 36.221 �0.037

5 1 0 36.881 36.882 �0.071�5 1 2 37.244 37.166 0.078�2 �2 4 38.600 38.636 �0.036�3 �1 5 41.239 41.256 �0.017�6 0 2 41.960 41.928 0.032

6 0 0 42.203 42.194 0.0093 3 1 42.580 42.597 �0.0170 2 5 44.814 44.737 0.077�1 �1 6 46.36 46.368 �0.008�3 �1 6 47.993 48.012 �0.019�4 0 6 48.847 48.926 �0.079

2 4 0 49.499 49.582 �0.083�6 2 3 50.238 50.227 0.011

6 2 1 50.744 50.688 0.0567 1 1 53.257 53.312 �0.055�1 �1 7 54.244 54.238 0.017�6 0 6 56.702 56.659 0.043�6 2 5 57.140 57.141 �0.001�2 �4 4 57.375 57.385 �0.010�1 �3 6 58.147 58.089 0.058�4 4 3 58.475 58.446 0.029

0 2 7 59.398 59.352 0.0464 4 2 59.714 59.722 �0.0083 3 5 60.044 60.066 �0.0220 0 8 62.338 62.282 0.056�1 �3 7 64.927 64.976 �0.049�3 5 2 65.454 65.442 0.012�2 �2 8 66.100 66.151 �0.015

Fig. 2. Room temperature measured X-ray diffraction patterns of BCT monocliniczirconolite at various firing temperatures.

142 K.B. Tan et al. / Journal of Alloys and Compounds 592 (2014) 140–149

[7,24–28]. In contrast to the analogous Bi2O3–ZnO–Ta2O5 system,the problem of having a relatively lower melting point copperoxide, �1300 �C has been circumvented by a slightly lower firingtemperature, i.e. at 900 �C and/or increased content of high refrac-tory tantalum oxide, wherein the Bi:Ta cation ratio is reduced to1.37:1 [28].

At this stage, there is not yet any electron or neutron diffractiondata available, but the structure of Bi1.92Cu0.08(Cu0.3Ta0.7)2O7.06

could be explained qualitatively based on the reported b-phase,Bi2(Zn1/3M00

2=3)2O7 (M00 = Ta or Nb) [24,28]. It is worthwhile men-tioning that transformation of the a-phase into the b-phase couldhave resulted from pyrochlore cell distortion that caused by thelong range coupling effect of Bi 6s2 inert lone pair electrons[9,23]. However, such changes are not observed in our presentstudy and the phase formation of the BCT monoclinic zirconolitephase is solely due to recombination of the formed bismuth binaryphases, BiTaO4 and Bi4Ta2O11. Further reaction with copper oxideat elevated temperatures would lead to the formation of a phasepure sample (more detailed discussion in Section 3.2). As men-tioned earlier, coordination network of the b-phase is rather inter-esting as three different metal cations of varying ionic radii havebeen portrayed using a modified CaZrTi2O7 structure model [24].

We speculate minimal changes in the cation distribution of theb-phase in the Cu analogue due to their close resemblance in struc-tural parameters [24–26,28]. The coordination environments of thebismuth, copper and tantalum cations are assumed to be similar tothose of Zn analogues. However, a slight variation is anticipated asthe increased tantalum content in Bi1.92Cu0.08(Cu0.3Ta0.7)2O7.06 mayhave caused copper to opt for the half occupied (5 + 1)-fold inter-stitial sites formed through six-membered rings of CuTaO6 octahe-dra. Alternatively, the second possibility is for copper to have ahigher coordination number, i.e. seven or eight, similar to that ofBi occupying alternating layers parallel to [110] direction. Thereis little likelihood of a small copper having coordination numberabove six; however, related pyrochlores studies have demon-strated that non-stoichiometry in composition and/or possiblestructural distortion may help to achieve a desirable bond length[6,10]. For better understandings of the b-phase in the Cu analogue,further careful structural investigations are certainly requiredbefore a firm conclusion could be drawn.

3.2. Phase formation and reaction pathway study

Fig. 2 shows the phase evolution study of Bi1.92Cu0.08

(Cu0.3Ta0.7)2O7.06 at different temperatures starting from 600 �Cto 900 �C using qualitative phase analysis. Most constituent oxidesremain unreacted at or below a temperature of 600 �C. An increaseof 50 �C has induced formation of the cubic phase, Bi3TaO7 andfurther heat treatment initiates the emergence of BiTaO4 andBi4Ta2O11 phases at the temperature of 750 �C. Interestingly, thereis a strong affinity for unreactive tantalum oxide to form bismuthbinary phases before it becomes fully diminished below thetemperature of 800 �C. At the temperature of 900 �C, phase pureb-phase is formed through a complete reaction between interme-diate BiTaO4, Bi4Ta2O11 phases and unreacted oxides. The overallformation mechanism of monoclinic zirconolite, Bi1.92Cu0.08 (Cu0.3-

Ta0.7)2O7.06 could be summarised as below:

1:5Bi2O3 þ 0:5Ta2O5 ! Bi3TaO7 600 �C < T < 700 �C ð1Þ

2Bi3TaO7þTa2O5!2BiTaO4þBi4Ta2O11 700 �C6 T 6750 �C ð2Þ

0:5BiTaO4 þ 0:355Bi4Ta2O11 þ 0:68CuOþ 0:095Ta2O5

! Bi1:92Cu0:08ðCu0:3Ta0:7Þ2O7:06 750 �C < T 6 900 �C ð3Þ

Table 2Elemental analysis of the prepared Bi1.92Cu0.08(Cu0.3Ta0.7)2O7.06.

Element Concentration, mg/L Atomic%

Calculated Experimental Calculated Experimental

Bi 39.59 39.87 ± 0.61 17.36 16.61 ± 0.06Cu 4.26 4.23 ± 0.08 6.15 5.79 ± 0.03Ta 25.00 23.72 ± 0.46 12.66 11.41 ± 0.05O 11.15 12.18 ± 1.15 63.83 66.19 ± 0.14

Fig. 4. Surface morphology of BCT monoclinic zirconolite after sintering for 24 h at900 �C.

K.B. Tan et al. / Journal of Alloys and Compounds 592 (2014) 140–149 143

Note that Bi1.92Cu0.08(Cu0.3Ta0.7)2O7.06 is thermally stable as nothermal event is discernible within the studied temperature range(Fig. 3). There is also no sign of any thermal event associated withthe polymorphic transition of the low-symmetry zirconolite phaseinto the high-symmetry cubic pyrochlore structure. Meanwhile,elemental analysis confirms a deviation of less than 5% betweenthe experimental and theoretical values, giving good evidence asto the compositional stoichiometry (Table 2). This again, provesthat the step-wise synthesis is applicable to tackling the problemof volatile oxides. Typical surface morphology of the as-fired pelletwith a density of �83% is shown in Fig. 4. The SEM micrograph re-veals a well-sintered polycrystalline sample with irregular grainsand their grain sizes are found in the range of 1–14 lm. The crys-tallite size estimated by Scherrer method is 31 nm indicating thegrains consist of hundreds of smaller crystallites. In contrast tothe reported a-phases, the b-phase has denser and more spheru-lite-like grains due to its lower crystallisation temperature, whichis caused by the increased bismuth content [11,14].

3.3. Electrical properties

The frequency and temperature dependence of the ac conduc-tivity of Bi1.92Cu0.08

(Cu0.3Ta0.7)2O7.06 are characterised over a broad temperaturerange within impedance frequency regime of below one megahertz(Fig. 5). As illustrated in Fig. 5d, the ac conductivity corresponded topower law frequency dependence (Eq. (4)) showing three differentfrequency regimes: (I) low-frequency plateau regime, (II) dispersiveregime and (III) high-frequency plateau regime, respectively[29,30].

rðxÞ ¼ ro þ Axn ð4Þ

where r(x) and ro refer to angular conductivity and dc conductiv-ity, respectively; Axn is the empirical expression of the transportproperties of the charge carrier with two temperature dependentconstants, 0.6 < n < 1.0 and A.

The complex impedance of an ac electric field, often expressedby Eq. (5), contains both real and imaginary parts that correspondto the components of resistance and reactance; whilst, the recipro-cal form will yield the complex form of admittance (Eq. (6)).

Z� ¼ Z0 � jZ00 ð5Þ

Fig. 3. DTA and TGA thermograms of BCT monoclinic zirconolite.

Y� ¼ Y 0 þ jY 00 ¼ ðZ�Þ�1 ð6Þ

To simplify this, the ac conductivity of pellet sample with a nor-malised geometry factor (thickness/area, cm�1) is considered tohave frequency-independent resistor conductivity (R�1) and thecapacitor conductivity (the admittance, xC) that varies linearlywith increasing frequency [32].

An almost linear frequency-independent, Y0 with low values, inthe order of 10�8–10�6 X�1 cm�1, observed at temperatures below200 K is approximately equal to the dc conduction (Fig. 5a). Hori-zontal lines are anticipated, in particular the jumping relaxationmodel (JRM) suggested a hopping conduction mechanism of chargecarrier within localised sites to another could be hardly perturbedby the applied field in low frequency region [30,31]. In R–C net-work modelling, this had been described as the allowance of perco-lated resistor path to current flow in a low-frequency plateauregime (I) when the ac conductivity of capacitor was sufficientlylow [32]. However, a closer look at the slightly deviated Y0 valuesfrom ideal dc conduction suggests that, there could be contribu-tions from possible more than one relaxation process. As the tem-perature increases, a nonlinear gain in ac conductivity is observedin the dispersive regime (II) by which it is best demonstrated in thetemperature range of 220–576 K (Fig. 5b). Such phenomenon isassociated with the increase of capacitor conductivity in whichthere is less opposition to current due to fewer accumulatedcharges. The charge dissipation with increasing frequency is in facta result of rapid change in the potential difference sign in an alter-nating ac field. For dispersive ac conductivity in regime (II), thecapacitive response is more dominant than the inductive one asthe displacement of charge carriers always lags behind the electricfield and subsequently, this allows the current to reach its maxi-mum earlier [33]. In a higher temperature range of 624–874 K(Fig. 5c), ac conductivity becomes non-frequency dependent; thisagrees reasonably with regime (III) as this could be attributed tothe characteristic of near constant loss (NCL) that found in mostglasses and ceramics [34,35]. In addition, the inductive effect is

Fig. 5. Real admittance, Y0 as a function of frequency: (a) 100–200 K; (b) 220–576 K; (c) 624–874 K with inductive effect of Y0 and (d) graphical representation of typicalfrequency dependence ac conductivity in three frequnecy regimes: (I) low-frequency (dc) plateau regime; (II) dispersive regime and (III) high-frequency plateau regime.

144 K.B. Tan et al. / Journal of Alloys and Compounds 592 (2014) 140–149

accounted for the slight drop in Y0 values at frequencies above100 kHz, especially the temperatures are higher than 724 K.

The log–log curves (Fig. 6) illustrate the variation of imaginarypart of complex impedance, Z00 against the frequency function atvarious temperatures. The emergence of a single peak (Fig. 6a) ordouble peaks (Fig. 6b) at each temperature is indicative of thenumber of relaxation processes occurring in different domains.Beyond the magnitude of the critical frequency, fc, Z00 values dropwith increasing temperature and frequency; a behaviour observedwithin the dispersive regime (II) in Fig. 5d. It is noteworthy that thepeak frequency, fc at the maxima of the imaginary part of compleximpedance, �Z00max (Eq. (7)), could be referred to as the crossover ofone relaxation process to another, in particular the dispersiveY0 value(s) at each fc point should yield a slope of close to one,n = �1, a characteristic of the power law response mentionedearlier in Eq. (4).

�Z00max ¼ R=2 ¼ ð2Y 0Þ�1 ð7Þ

Peak maxima are seen shifting towards lower Z00 values withincreasing temperature, once again, providing a sign of thermallyactivated mechanism that involves charge carriers hopping overpotential barrier. Such frequency-dependent peak maxima alsoconform well to the Arrhenius type equation:

fc ¼ fo expð�Ea=kBTÞ ð8Þ

where fo is the relaxation frequency at infinite temperature, whilstT, kB and Ea are the absolute temperature, Boltzmann’s constant andactivation energy for electrical conduction, respectively.

At this stage, it is clear that two relaxation processes areevidently responsible for the ac conductivity response ofBi1.92Cu0.08(Cu0.3Ta0.7)2O7.06. However, the broader and asymmetri-cal shapes of Z00 curves (visual inspection) at temperatures below150 K suggest a possible distribution of two overlapping relaxationprocesses at around the peak frequency (Fig. 6a). Hence, an imme-diate task arises as to identify the relaxation processes that takeplace in different domains. Ideally, the relaxation time, s of amaterial in the frequency scaled domain could be correlated withresistive and capacitive components through a simple expressionas below:

s ¼ RC ¼ ð2pfcÞ�1 ð9Þ

In reality, it is far more complicated as the relaxation processusually involves a complex interplay of various factors that includethermal and frequency parameters, mobility or travelled distanceof charge carriers within inter- and/or intra- domains as well asthe distribution of potential wells. For short-range charge carrierdiffusion, the relaxation process in bulk domains is anticipated tooccur at higher frequency than that of grain boundary due to theirrelatively closer domains [31]. Consider the fc is of temperature-dependent, the respective relaxation process will shift to higher

Fig. 6. Imaginary part of complex impedance, Z00 at different temperatures: (a)100–190 K; (b) 200–290 K and �300–580 K (inset of figure) as a function offrequency.

Fig. 7. Plots of critical frequency as a function of temperature at 90–280 K and250–624 K (inset of figure).

K.B. Tan et al. / Journal of Alloys and Compounds 592 (2014) 140–149 145

frequencies due to the thermally activated process. At this stage,the peak maxima observed in Fig. 6a are associated with bulkdomains whereas their diminishing peak characteristics at higherfrequencies and temperatures (Fig. 6b) are due to the restrictedmeasuring frequency. Meanwhile, the peak frequency that spurredby the relaxation process in grain boundary domains is onlydiscernible once the temperature reaches above 250 K (Fig. 6band inset). Therefore, the aforementioned overlapping relaxationresponse could be concluded from both domains, in which theslightly deviated (Fig. 5a) behaviour of Y0 is probably due to thecontribution from the high resistive component of the grainboundary.

From Eq. (8), Arrhenius-type plots of critical frequency, fc as afunction of temperature (Fig. 7) yielded straight lines portrayingthree slopes with apparent activation energies of, 0.08 eV, 0.16 eVand 0.37 eV for two different temperature ranges, 90–280 K and250–624 K, respectively. The effect of overlapping relaxationprocesses below temperatures of 150 K is observed again in whichthe required energy for electrical conduction may have been low-ered by the conductivity of the grain boundary domains, Y 0gb. Asthe temperature increases, the ac conductivity of the bulk domains,Y 0bulk becomes more pronounced with an almost two-fold increaseof activation energy (inset of Fig. 7); whilst, higher Ea, of 0.37 eV

(Fig. 7) is required for the resistive grain boundary domains in orderto overcome their energy barrier. According to the classical modelof potential well distribution, the temperature and frequencydependent ac conductivity of Bi1.92Cu0.08(Cu0.3Ta0.7)2O7.06 are ex-pected to have multiple step-wise increases (if the frequency isbroad enough) for the two relaxation processes in both the grainboundary and bulk domains [31].

Note that the slopes of the log–log curves of Y0 against criticalfrequency (insets of Fig. 8) for two temperature ranges show valuesof 0.98 and 0.96 (n = �1), which agreed reasonably well with thepower law dependence of conductivity (Eq. (4)). Therefore, Y0 orr value (Fig. 5) at each critical frequency is then plotted againsttemperature (Fig. 8) using a slightly modified equation of Arrhe-nius conductivity:

r ¼ ro expð�Ea=kBTÞ ð10Þ

The close agreement between activation energies obtained fromthese two methods (Figs. 7 and 8) provides additional evidence forthe thermally activated hopping mechanism proposed earlier.

For further confirmation, complex Cole–Cole or Nyquist plot isused to identify relaxation process within grain and/or grainboundary domains. The complex impedance formalism (Eq. (5))is depicted as Z00’ against Z00 showing a perfect or incomplete semi-circle arc at different temperatures. According to the Debye’s mod-el, a material having a single relaxation process will give rise to anideal semicircle centred on the real axis with an equivalent circuitconsisting of R and C elements that are connected in parallel. At thetemperature of 304 K, two semicircles are discernible for therespective relaxation processes at either a low or high frequencyregime (Fig. 9). The first semicircle at high frequency is corre-sponded to the transport phenomenon in bulk domains (intra-granular) with an associated capacitance of 6.74 � 10�12 F cm�1

(xmaxRC = 1) and resistivity of 2.97� 103 ohm cm that determinedfrom the Z0 intercept, i.e. Z00 = 0. Meanwhile, the second semicircleis attributed to the grain boundary (inter-granular) domains withcapacitance and resistivity values of 6.63 � 10�9 F cm�1 and 3.80� 104 ohm cm, respectively. The orders of capacitance valuesobtained here are fitted well to the typical electrical response indifferent domains [36]; especially higher resistivity in grainboundary domains is usually expected. Hence, the overall equiva-lent electric circuit could be well demonstrated by two pairs of par-allel RC elements that are connected in series; whilst, an inductivecomponent is required for the distinctive feature that observed at

Fig. 8. Plots of the extracted conductivity value (from Fig. 5) at each criticalfrequency as a function of temperature: (a) 90–280 K and (b) 250–624 K. Verifi-cation of the transport property, n using plots of conductivity as a function ofcritical frequency is shown in the respective insets of figure.

Fig. 9. Complex impedance plots measured at the temperature of 304 K and theinset of figure portrays additional one inductive component connected in series toexisting two parallel RC elements at temperature of 774 K.

146 K.B. Tan et al. / Journal of Alloys and Compounds 592 (2014) 140–149

the temperature of 774 K, as the positive Z00 values show substan-tial gain in inductive reactance (inset of Fig. 9).

Electric modulus spectroscopic plots, combining Z00 and M00 onlogarithmic frequency function are useful to probe the electro-ac-tive region within a sample. In particular, the large resistanceand small capacitance are highlighted in respective Z00 and M00 plots[36]. The coincident peaks of M00 and Z00 at frequencies above103 kHz (Fig. 10) are assigned to the bulk domains whereas thelow frequency M00 peak (denoted �) overlapping with the Z00 peakat 102 Hz is associated with the grain boundary domains (insetof Fig. 10). The shoulder peak of M00 (denoted �) at low frequencyhas provided a clear evidence for the significant capacitance differ-ence of several orders between the two domains.

The dielectric response of the b-BCT has been carefully investi-gated in both fixed-frequency and fixed temperature formats,respectively. Dispersive behaviours are observed for the real partof the dielectric constant, e0 in the low frequency regime, i.e. below10 kHz and below 1 kHz as shown in Fig. 11. As the temperatureincreases, the distinctive features of downward step in e0 and thestep-wise increase in Y0 (Fig. 5) are well correlated and this sug-gests a relaxation process of the dipolar degree of freedom [37].High e0 values in the order of P104 in low frequency regime arein fact a result of strong contact contribution that caused by thethin insulating layer in grain boundary domains. In the high fre-quency regime, e0 values remain constant as the polarisation ofthe induced moment could not synchronise with the frequencyof an applied electric field. The relaxor-like, temperature- and fre-quency-dependent e0max as found in the a-BZN and a-BZT ana-logues [20–22,28] are not observed; an indication of the absenceof a paraelectric to ferroelectric phase transition at low tempera-ture. Meanwhile, the e0 of b-BCT at higher frequencies are indepen-dent of temperatures below 220 K showing a bulk permittivity of�70 (inset of Fig. 12a), and then increasing appreciably at temper-atures above �300 K due to the thermally activated dispersive con-ductivity (Fig. 12b).

On the other hand, the key features of the frequency depen-dence and temperature dependence of dielectric loss are illustratedin Figs. 13 and 14. The temperature-independent peak maxima ofdielectric loss, Tm remain constant but shift towards higher fre-quencies with increasing temperature (Fig. 13a); a behaviour typ-ical of the aforementioned nearly constant loss (NCL) in bulkdomains. Meanwhile, the highly resistive grain boundary domainscontribute towards unreasonably high dispersive losses (Fig. 13b)at elevated temperatures when high energy is lost through move-

Fig. 10. Combined spectroscopic plots of Bi1.92Cu0.08(Cu0.3Ta0.7)2O7.06 at the tem-perature of 304 K.

Fig. 11. Dielectric constant, e0 as a function of frequency forBi1.92Cu0.08(Cu0.3Ta0.7)2O7.06 measured at high temperatures and subambient tem-peratures (inset of figure).

Fig. 12. Dielectric constant, e0 as a function of temperature for Bi1.92Cu0.08(Cu0.3-

Ta0.7)2O7.06 at several fixed frequencies: (a) e0 at subambient temperatures with theenlarged scale as shown in the inset of figure and (b) e0 at temperatures above304 K.

Fig. 13. Frequency dependence of dielectric loss, tand of Bi1.92Cu0.08(Cu0.3Ta0.7)2

O7.06 measured in different temperature ranges: (a) subambient temperatures and(b) at temperatures above 304 K.

K.B. Tan et al. / Journal of Alloys and Compounds 592 (2014) 140–149 147

ment of the screening charge against the applied field. Note thatthe energy loss in each hopping transition is correlated throughthe ratio of energy lost, WL to the energy stored, WS, as describedbelow [38]:

WL=WS ¼ ð1� nÞ=n ð11Þ

where n is the restraint of screening imposed by the lattice/domain.Hence, the temperature dependent dielectric loss is found to be

thermally activated giving remarkably high tand values when b-BCT becomes semiconducting due to possible copper oxidationand/or defects at high temperatures (Fig. 14). The high dielectricloss could result from the high leakage current within the electricalconduction of b-BCT. The observation here agrees well with thosereported for b-phases in Zn analogues but with two exceptions;(i) the grain boundary is well-defined as an intrinsic propertyresponsible for the high e0 and tan d and (ii) b-BCT is a semicon-ducting phase.

The electrical response found in this study has clearly indicateda primary structural difference between the pyrochlore and mono-clinic zirconolite phases. However, the worthwhile questions to beconsidered are whether (i) the various electrical plots are useful toexplain the electrical response of b-BCT, (ii) the grain boundary is areal intrinsic property or one that could possibly be eliminated, (iii)the A site in the b-phase is occupied solely by Bi3+ or a minuteamount of copper is allowed to be present and (iv) the electrical re-sponse is associated with the contribution of the A site cation, Bsite cation or oxygen vacancy within the structure. At this juncture,

Fig. 14. Temperature dependence of dielectric loss, tan d of Bi1.92Cu0.08(Cu0.3Ta0.7)2

O7.06 at several fixed frequencies: (a) subambient temperatures and (b) attemperatures above 304 K.

148 K.B. Tan et al. / Journal of Alloys and Compounds 592 (2014) 140–149

there is hardly any firm conclusion could be drawn as to the actualelectrical conduction mechanism of b-BCT. However, the system-atic electrical plots shown here have clearly demonstrated a phys-ical interpretation of the frequency and temperature dependenceof the electrical properties of b-BCT in relation to the interactionbetween the inter- or intra- electro-active regions. Second, thegrain boundary domains appear as an intrinsic characteristic thatcontributes significantly to the electrical response regardless ofthe careful controls applied during sintering. The formation of athin insulating layer is probably attributable to the oxidation pro-cess during synthesis in which the uneven cation distributioncould have yielded electrical imhomogeneity or misalignment ofgrains within the structure. As to questions (iii) and (iv), the highbulk permittivity of b-BCT could probably result from the polarbulk domains that allow small atomic displacement due to thepolarisable nature of the crystal structure. The proposed structuraldiffusion involving A-site divalent cation in a-phase [6,39] doesnot seem applicable to the b-BCT phase, especially since the ques-tionable copper at the A site may exist in a negligibly small amountmaking only a limited electrical contribution. Last but not least, thesemiconducting behaviour of b-BCT is evident at elevated temper-atures and the electrical conduction is expected to be governed bythe hopping electronic transport mechanism.

4. Conclusions

The phase compatibility and formation mechanism of novelmonoclinic zirconolite (b-BCT), Bi1.92Cu0.08(Cu0.3Ta0.7)2O7.06

prepared via solid state powder reaction had been investigatedcarefully. Intermediate bismuth tantalate phases were requiredprior to the formation of thermally stable b-BCT of which its struc-tural characteristics could be well demonstrated on a modifiedmonoclinic zirconolite model. Polycrystalline b-BCT was denseand of irregular grain shapes due to the lower crystallisation tem-perature arising from the high bismuth content. The ac electricalconductivity of b-BCT corresponded well to power-law frequencydependence with distinctive features of conductivity in differentfrequency regimes. The electrical response of b-BCT was well fittedby various electrical plots in which two primary bulk and grainboundary domains were responsible for the electro-active phenom-enon within the structure. Low activation energy was required forthe hopping conduction mechanism by which b-BCT exhibitedsemiconducting behaviour at elevated temperatures.

Acknowledgements

Sincere thanks to Fundamental Research Grant Scheme (FRGS)from the Ministry of Higher Education Malaysia (MOHE) and MPChon’s National Science Fellowship (NSF) from the Ministry ofScience, Technology, and Innovation Malaysia (MOSTI). Theauthors are also grateful for the low-temperature electricalmeasurements at the Department of Materials Science andEngineering, The University of Sheffield, UK.

References

[1] R.V. Gaines, H.C.W. Skinner, E.E. Foord, B. Mason, A. Rosenzweig, Dana’s NewMineralogy, eight ed., Wiley, New York, 1997.

[2] H.V. Gaertner, Neues. Jahrb. Mineral Geol. Palaeontol. 61 (1930) 1–30.[3] M.A. Subramanian, G. Aravamudan, G.V. Subba Rao, Prog. Solid State Chem. 15

(1983) 55–143.[4] A.F. Qasrawi, A. Mergen, J. Alloys Comp. 496 (2010) 87–90.[5] T.A. Vanderah, I. Levin, M.W. Lufaso, Eur. J. Inorg. Chem. 2005 (2005) 2895–

2901.[6] Y. Liu, R.L. Withers, Local structure of relaxor dielectric ceramics, in: C.

Sikalidis (Ed.), Advances in Ceramics–Electric and Magnetic Ceramics,Bioceramics, Ceramics and Environment, In Tech, 2011, pp. 87–102.

[7] R.D. Shannon, Acta Cryst. A32 (1976) 751.[8] X.H. Zhang, W. Ren, P. Shi, X.Q. Wu, X.F. Chen, X. Yao, J. Alloys Comp. 553

(2013) 8–13.[9] X. Wang, H. Wang, X. Yao, J. Am. Ceram. Soc. 80 (1997) 2745–2748.

[10] K.B. Tan, C.C. Khaw, C.K. Lee, Z. Zainal, G.C. Miles, J. Alloys Comp. 508 (2010)457–462.

[11] M.P. Chon, K.B. Tan, C.C. Khaw, Z. Zainal, Y.H. Taufiq Yap, P.Y. Tan, Ceram. Int.38 (2012) 4253–4261.

[12] M. Valant, D. Suvorov, J. Am. Ceram. Soc. 88 (2005) 2540–2543.[13] S.H. Shim, J-W. Yoon, K.B. Shim, J-I. Mutsushita, B.S. Hyun, S.G. Kang, J. Alloys

Comp. 413 (2006) 188–192.[14] P.Y. Tan, K.B. Tan, C.C. Khaw, Z. Zainal, S.K. Chen, M.P. Chon, Ceram. Int. 38

(2012) 5401–5409.[15] D.P. Cann, C.A. Randall, T.R. Shrout, Solid State Commun. 100 (1996) 529–534.[16] A.V. Egorysheva, O.G. Ellert, Yu.V. Maksimov, V.D. Volodin, N.N. Efimov, V.M.

Novotortsev, J. Alloys Comp. 579 (2013) 311–314.[17] M.W. Lufaso, T.A. Vanderah, I.M. Pazos, I. Levin, R.S. Roth, J.C. Nino, V.

Provenzano, P.K. Schenck, J. Solid State Chem. 179 (2006) 3900–3910.[18] T.A. Vanderah, T. Siegrist, M.W. Lufaso, M.C. Yeager, R.S. Roth, J.C. Nino, S.

Yates, Eur. J. Inorg. Chem. 2006 (2006) 4908–4914.[19] T.A. Vanderah, M.W. Lufaso, A.U. Adler, I. Levin, J.C. Nino, V. Provenzano, P.K.

Schenck, J. Solid State Chem. 179 (2006) 3467–3477.[20] J.C. Nino, M.T. Lanagan, C.A. Randall, J. Appl. Phys. 89 (2001) 4512–4516.[21] J.C. Nino, M.T. Lanagan, C.A. Randall, S. Kamba, Appl. Phys. Lett. 81 (2002)

4404–4406.[22] H.J. Youn, T. Sogabe, C.A. Randall, T.R. Shrout, M.T. Lanagan, J. Am. Ceram. Soc.

84 (2001) 2557–2562.[23] M. Valant, P.K. Davies, J. Am. Ceram. Soc. 83 (2000) 147–153.[24] I. Levin, T.G. Amos, J.C. Nino, T.A. Vanderah, I.M. Reaney, C.A. Randall, M.T.

Lanagan, J. Mater. Res. 17 (2002) 1406–1411.[25] J.C. Nino, I.M. Reaney, M.T. Lanagan, C.A. Randall, Mater. Lett. 57 (2002) 414–

419.[26] K.B. Tan, C.K. Lee, Z. Zainal, C.C. Khaw, Y.P. Tan, H. Shaari, Pac. J. Sci. Technol. 9

(2008) 468–479.[27] C.C. Khaw, C.K. Lee, Z. Zainal, G.C. Miles, A.R. West, J. Am. Ceram. Soc. 90 (2007)

2900–2904.[28] C.C. Khaw, K.B. Tan, C.K. Lee, A.R. West, J. Eur. Ceram. Soc. 32 (2012) 671–680.[29] A.K. Jonscher, J. Phys. D: Appl. Phys. 32 (1999) R57–R70.

K.B. Tan et al. / Journal of Alloys and Compounds 592 (2014) 140–149 149

[30] K. Funke, Solid State Chem. 22 (1993) 111–195.[31] W. Li, R.W. Schwartz, Appl. Phys. Lett. 89 (2006) 242906.[32] S. Panteny, R. Stevens, C.R. Brown, Ferroelectrics 319 (2005) 199–208.[33] J.C. Dyre, T.B. Schroder, Rev. Mod. Phys. 72 (2000) 873–892.[34] C. Leon, A. Rivera, A. Varez, J. Sanz, J. Santamaria, K.L. Ngai, Phys. Rev. Lett. 86

(2001) 1279–1282.[35] B. Natesan, N.K. Karan, R.S. Katiyar, Phys. Rev. E. 74 (2006) 042801.

[36] J.T.S. Irvine, D.C. Sinclair, A.R. West, Adv. Mater. 2 (1990) 132–138.[37] V. Bobnar, P. Lunkenheimer, J. Hemberger, A. Loidl, F. Lichtenberg, J. Mannhart,

Phys. Rev. B. 65 (2002) 155115.[38] J.M. Herbert, The properties of dielectrics, in: D.S. Campbell (Ed.), Ceramics

Dielectrics and Capacitors, Gordon and Breach, New York, 1985, pp. 9–62.[39] I. Levin, T.G. Amos, J.C. Nino, T.A. Vanderah, C.A. Randall, M.T. Lanagan, J. Solid

State Chem. 168 (2002) 69–75.