A R C H I V E S O F M E T A L L U R G Y A N D M A T E R I A L S

Volume 56 2011 Issue 4

DOI: 10.2478/v10172-011-0127-4

H. BERNARD∗, A. LISIŃSKA-CZEKAJ∗, J. DZIK∗, K. OSIŃSKA∗, D. CZEKAJ∗

FABRICATION, STRUCTURAL AND AC IMPEDANCE STUDIES OF LAYER–STRUCTURED Bi4Ti3O12 CERAMICS

WYTWARZANIE, STRUKTURA ORAZ BADANIA IMPEDANCYJNE WARSTWOWEJ CERAMIKI Bi4Ti3O12

In the present research bismuth titanate Bi4Ti3O12 (BiT) ceramics was synthesized by the standard solid-state reactionmethod from the mixture of oxides, followed by free sintering at temperature T=1000C. BiT ceramics was studied in termsof its chemical composition (EDS), crystalline structure (X-ray), microstructure (SEM) and dielectric properties (ac techniqueof complex impedance spectroscopy) over a range of frequency ( f =100Hz to f =1MHz) and temperature (T=200-500C).Experimental results confirmed the phase formation. It was found that BiT ceramics crystallized in orthorhombic symme-try, best described with Fmmm space group and the following elementary cell parameters: a=5.409(6)A, b=5.449(2)A andc=32.816(2)A. It was also found that BiT ceramics exhibited the plate–like microstructure and stoichiometric chemical compo-sition. Impedance spectroscopy measurements showed contribution of three overlapping relaxation processes (three semicirclesin the complex impedance diagrams were observed) ascribed to bulk, grain boundary and electrode/interface polarizationphenomena. Impedance data were fitted to the corresponding equivalent circuit using the complex nonlinear least squares(CNLS) method. The ac conductivity for grains, grain boundaries and electrode processes was calculated from CNLS fit ofthe impedance data and thus the activation energy of ac conductivity (σAC) and relaxation (τ) was calculated for the threerevealed components of the impedance spectra from the slopes of σAC and τ versus 1000/T plots (semi log scale) in the rangeof ∆T=200–500C.

Keywords: bismuth titanate, Aurivillius phase, impedance spectroscopy

W prezentowanej pracy ceramikę tytanianu bizmutu Bi4Ti3O12 (BiT) syntezowano metodą reakcji w fazie stałej z miesza-niny prostych tlenków w T=1000C. Zbadano skład chemiczny (EDS), strukturę krystaliczną (X-ray), mikrostrukturę (SEM)oraz właściwości dielektryczne metodą spektroskopii impedancyjnej w zakresie częstotliwości od f =100Hz do f =1MHz wtemperaturze T=200-500C. Analizując obraz SEM można zauważyć, że płytkopodobne ziarna tworzą liczne, rozgałęzionepręty. Wyniki eksperymentalne potwierdziły wytworzenie się fazy krystalicznej BiT. Ceramika BiT krystalizuje w symetriirombowej, opisywana jest grupą przestrzenną Fmmm, charakteryzuje się parametrami komórki: a=5.409(6)A, b=5.449(2)A,c=32.816(2)A.

Pomiary spektroskopii impendancyjnej wykazały udział trzech nakładających się procesów relaksacyjnych (zaobserwowanotrzy półkola przedstawione w diagramach impedancji) przypisanych wnętrzu ziaren, granicy ziaren oraz zjawisk polaryzacji nagranicy powierzchni elektroda/próbka.

Dane impedancyjne przybliżono odpowiednimi schematami zastępczymi przy użyciu metody złożonych, nieliniowych,najmniejszych kwadratów (CNLS). Wyznaczono przewodnictwo zmiennoprądowe dla ziaren, granic ziaren i procesów elektro-dowych oraz energię aktywacji przewodnictwa Ec

a i relaksacji Eta w zakresie temperatur ∆T=200–500C.

1. Introduction

The family of oxides having bismuth oxide layers al-ternating with perovskite structure layers, known as per-ovskite layered structures, were discovered by Aurivillius[1-3]. Since the discovery, more than fifty compoundshave been found to adopt this type of structure andmany have shown to exhibit displacive ferroelectric tran-sitions. The crystal structure of bismuth layer-structured

ferroelectrics (BLSFs) may be described as a stackingof bismuth oxide layers (Bi2O2)2+ and perovskite lay-ers (Am−1BmO3m−1)2− along the pseudotetragonal c axis.Here A can be mono-, di- or trivalent ions or a mixtureof them, where B represents Ti4+, Nb5+, Mo6+, W6+,Fe3+. In the (Am−1BmO3m−1)2− units, B ions are enclosedby oxygen octahedral, which are linked through cornersforming O-B-O linear chains. A ions occupy the spacesin the framework of BO6octahedral. The perovskite slab

∗ UNIVERSITY OF SILESIA, DEPARTMENT OF MATERIALS SCIENCE, 41-200 SOSNOWIEC, 2 ŚNIEŻNA STR., POLAND

1138

contains (m) oxygen regular octahedrons along its thick-ness. Adjacent perovskite-like layers are displaced in re-lation to each other for a0

/√2 in [110] direction, where

a0 is the prototype lattice parameter (regular perovskitecell) [4-6].

Fig. 1. Structure of Am−1Bi2MmO3m+3 Aurivillius phase for m=3 [16]

Archetypal members of the Aurivillius familyare therefore Bi2WO6 (m=1), SrBi2Ta2O9 (m=2) andBi4Ti3O12 (m=3). As shown in Fig. 1 for the case ofBi4Ti3O12 (BiT) the structure can be envisaged as lay-ers of corner-connected octahedrally coordinated ions,separated by (Bi2O2)2+ layers [4]. The layers with theoctahedrally coordinated ions i.e., (Bi2Ti3O10)2− layers,look like slabs of the perovskite structure. That is to saythat Ti ions are enclosed by oxygen octahedral, which arelinked through corners forming O-Ti-O linear chains. Biions occupy the spaces in the framework of TiO6 oc-tahedral. The height of the perovskite-type layer sand-wiched between Bi2O2 layers in Bi4Ti3O12 is equal to

six O-Ti-O distances or approximately to m=3 ABO3perovskite units [7].

It has been shown [7, 8] that when the perovskiteblock is an even number of octahedral thick, the sym-metry imposes a restriction on the polarization direction,confining it to the a−b plane. In contrast, when the per-ovskite block is an odd number of octahedral thick, itis possible to develop a component of the polarizationalong the c-axis (nearly perpendicular to the layers). Ithas important implications for the use of the bismuthlayer perovskites structure. Since there are comparativelyfew allowed directions for the spontaneous polarization,the remanent polarization is rather small for many filmorientations.

Bismuth titanate, Bi4Ti3O12 (BiT; m=3) is one ofthe best studied member of the BLSFs, and wide varietyof solid solutions based on BiT have successfully beensynthesized [e.g. 9, 10]. It is a well-known candidatefor high temperature piezoelectric applications, memorystorage, and optical display devices due to its high Curietemperature (T∼675C), a good resistance vs tempera-ture, low dielectric dissipation, a relatively high dielec-tric constant ε≈200 and an anisotropic nature originatingfrom its layered structure [11, 12]. Excellent electricaland even electro-optic properties could be achieved inferroelectric films of BiT [13].

Electrical and physical properties of bismuth ti-tanate layer-structured ferroelectrics can be formed and/or changed due to their large possibilities of changing(their) chemical composition (e.g. by ionic substitutionsin A and B positions) and crystalline structure (due todisplacement of ions from their regular positions, twistof octahedral chains or displacement of ions with mutualtwist of octahedrons).

In the present study the authors have focused on fab-rication of Bi4Ti3O12 ceramics by mixed oxide methodfollowed by free sintering and characterizing it from thepoint of view of its thermal behavior, crystalline struc-ture, microstructure, chemical composition and ac im-pedance behavior at low frequency ( f =100Hz–1MHz)and in the temperature range up to T=500C.

2. Experimental

The synthesis of Bi4Ti3O12 ceramic powders wasperformed according to the formula:

2Bi2O3 + 3TiO2 → Bi4Ti3O12 (1)

from the stoichiometric mixture of bismuth oxide Bi2O3(99.9% ALDRICH) and titanium oxide TiO2 (99.9%POCH) by the standard mixed oxide method (MOM).

1139

The consolidation of the powders was reached byfree sintering method at T=1000C. The soaking timeof the sintering process was t=2h.

Simultaneous thermal analysis (STA), was used toinvestigate synthesis effects in the stoichiometric mix-ture of powders in which both thermal analysis (DTA)and mass change effects (TG) are measured concurrent-ly on the same sample. The measurements were ob-tained with Netzsch STA409 thermal analyzer which isa combined DTA/TG/DTG system. The heating rate wasκ=10C/min.

Microstructure of the ceramic samples was ob-served by scanning electron microscopy with HITACHIS-4700–type equipment. The stoichiometry of Bi4Ti3O12ceramics was studied with the EDS chemical composi-tion analysis system. The X-ray patterns were recordedin a 2Θ range from 10 to 105 with the X’Pert Prodiffractometer in the Θ-2Θ mode at room temperatureusing CoKα radiation, detector scan step ∆2Θ=0.01 anda counting time t=7s. The analysis of the X-ray diffrac-tion patterns of the ceramic powders was carried outusing the X’Pert HighScore Plus software (PANalyticalB.V). X’Pert HighScore Plus program has enabled phaseanalysis, using the latest available ICSD and ICDD data-base. The phase analysis was also conducted using free-ware databases IUCr/COD/AMCSD using the Match!-computer program (Crystal Impact, Inc.). Microstruc-ture and crystalline structure studies were carried out atroom temperature.

For impedance measurements, sintered sampleswere polished and silver paste was deposited on bothsides. The ac impedance behavior of Bi4Ti3O12 ce-

ramics were studied with the impedance analyzer ofHP4192A type. All impedance data were fitted to appro-priate equivalent circuits using ZView Software (Scrib-ner Associates, Inc.). The validity of the CNLS (com-plex non-linear least squares method [14]) fitting proce-dure was estimated according to the following methods:χ-squared and the weighted sum of squares, referred toas χ2 and WSS [15]. In the present study we chose a“modulus data weighting” [16].

3. Results and discussion

The results of the simultaneous thermal analysis ofthe stoichiometric mixture of starting oxides, i.e. Bi2O3and TiO2 that were used for synthesis of Bi4Ti3O12 ce-ramics are given in Fig.2. One can see from Fig. 2that the mass loss starts at about T=250C and fin-ishes at about T=500C. The total value of the massloss measured at T=500C is rather small and amountsto ∆m=0.29%. No DTA peak corresponds to this masschange effect.

Taking into account the results of thermal analysisof pure titanium oxide and bismuth oxide [4] one cansay that pure TiO2 titanium oxide does not exhibit anythermal events being heated to temperature T=1300Cwhile pure Bi2O3 suffers phase transitions (exemplifiedas peaks on DTA curve) from the monoclinic δ-phaseto the high temperature cubic phase at approximatelyT=730C. The δ-phase is stable up to its melting point atapproximately T=825C [4]. No such effects are visibleon thermal analysis curves of the stoichiometric mixtureof oxides given in Fig. 2.

Fig. 2. Thermal analysis data of stoichiometric mixture of powders, viz. 2Bi2O3 + 3TiO2 leading to formation of Bi4Ti3O12 compound

1140

According to the experimental data on mechanismof formation of Bi4Ti3O12 [17] its thermal behavior andcrystal structure [18], Bi4Ti3O12 formation occurs by themechanism of successive rearrangements of α-Bi2O3 in-to Bi12TiO20 (initiated by incorporation of titanium diox-ide into Bi2O3 structure) and Bi12TiO20 into Bi4Ti3O12(transport processes) [17]. Formation of an intermediateBi12TiO20 phase, determined from X-ray analysis and re-ported in [17], starts at about T=300C reaches its max-imum value at about T=550C. With a further increasein temperature up to T=800C the amount of Bi12TiO20phase monotonously decays to zero. Within the tempera-ture range ∆T=500–800C formation of Bi4Ti3O12 phasetakes place [17]. Taking into account the above resultsthe mass loss observed in Fig. 2 can be assigned to theformation of an intermediate Bi12TiO20 phase whereasformation of Bi4Ti3O12 occurs without a mass loss.

The sharp endothermic peak on DTA curve atT=1210C (onset at T=1193C; Fig. 2) is in a goodagreement with both differential thermal analysis anddifferential scanning calorimetry data performed for syn-thesized Bi4Ti3O12 compound. One can state that thepeak corresponds to the solidus temperature of the syn-thesized materials [18, 19].

Fig. 3. shows the SEM images and EDS spectrumof BiT ceramic samples fabricated by free sintering atT=1000C. One can see from Fig. 3 that microstruc-ture of BiT ceramics consists of unusually straight andorthogonally branched rods and well packed platelets.The grains exhibited a plate-like shape and were aboutd=0.1µm wide and l=0.2µm long. It is usually thought[1] that the plate-like grains are useful as materials formaking grain-oriented ferroelectric ceramics. Stoichio-metric ratio of the main metallic components was calcu-lated for simple oxides Bi2O3 and TiO2 on the base ofEDS measurements and the results of such calculationare presented in Tab.1 where both theoretical and exper-imental content of elements for Bi4Ti3O12 ceramics canbe compared. It can be seen from Tab. 1 that BiT ce-ramics fabricated according to our technology exhibitedstoichiometric composition with an accuracy typical forthe measuring method applied.

TABLE 1Theoretical and experimental content of elements (calculation for

simple oxides) for Bi4Ti3O12 ceramic

FormulaTheoretic contents

of oxides [%]Content of oxidesfrom EDS [%]

Bi2O3 79.55 79.58

TiO2 20.45 20.42

Fig. 3. SEM micrograph and analysis spectrum EDS of Bi4Ti3O12 ceramic prepared by free sintering at T=1000C

1141

Fig. 4. X-ray diffraction pattern of Bi4Ti3O12 ceramic powder at room temperature

The room temperature X-ray diffraction pattern ofBi4Ti3O12 ceramic powder is shown in Fig.4. The struc-ture of Bi4Ti3O12 was found as orthorhombic (spacegroup Fmmm) with the following unit cell parameters:a=5.409(6)A, b=5.449(2)A and c=32.816(2)A. A moredetailed description of the X-ray analysis performed forthe fabricated ceramics can be found elsewhere [20].

Impedance spectroscopy (IS) was utilized for char-acterization of BiT ceramics. It is a powerful method ofcharacterizing many of the electrical properties of mate-rials and their interfaces with electronically conductingelectrodes. A great strength of impedance spectroscopylies in the fact that, with the appropriate data analysis, itis often possible to characterize the different electricallyactive regions in a material by demonstrating their exis-tence and by measuring their individual electric proper-ties [21, 22].

In the present study the complex impedance formal-ism expressed as:

Z∗ = Z′ − jZ

′′= RS − j

ωCS(2)

was used. Z ′ and Z ′′ are the real and imaginary compo-nents of impedance, ω=2π f – angular frequency, RS andCS are series components of equivalent circuit compo-nents, i.e. resistance and capacitance, respectively.

The idealized plot in the complex plane (i.e. Z ′′ vs.Z ′) that describes a polycrystalline oxide material usu-ally includes three components with their correspondingrelaxation frequencies. At higher frequencies the com-ponent normally corresponds to the bulk properties, at

intermediate frequency the element corresponds to thegrain boundaries and at low frequency the contributionof electrode processes or processes occurring at elec-trode/interface are represented [21, 22].

Visual analysis of the individual IS data plots giv-en in Fig. 5 (T=200C (a), T=400C (b), T=500C (c))have shown that the experimental points form smoothcurves with three partially overlapping depressed semi-circles (Fig. 5, open circles). To check the quality of theimpedance data, which is essential for a proper CNLSanalysis [14], the Kramers–Kronig (K-K) validation testswere performed.

In this connection it is worth noting that the CNLSmethod centers on minimizing the object function S withrespect to the parameters ak of the model function Z∗(ν,ak):

S =

N∑

i=1

wi

[Z′i − Z

′(νi, ak)

]r +

[Z′′i − Z

′′(νi, ak)

]2, (3)

where Z′i − jZ

′′i represent the measured impedance at fre-

quency νi and wi is the weight factor. However, it issometimes difficult to obtain an acceptable match be-tween measurement and model. The reason can be aninadequate model function or date that is corrupted by asystematic deviation. Therefore it is important to discernbetween “bad” data and an inappropriate model function.

The Kramers-Kroning relation present a very use-ful tool for data validation [23]. The Kramers-Kroningrule states that the imaginary part of the dispersion iscompletely determined by the form of the real part and

1142

vice-versa – the real part is determined by imaginarydispersion over the frequency range 06ν6∞. These re-lations can be expressed for real and imaginary part ofimpedance or admittance and found in literature [eg. 6,23]. In the course of the present study the computer pro-gram by B.A.Boulamp was used [23, 24] for performingthe K-K validation tests.

Fig. 5. Impedance diagram for Bi4Ti3O12 ceramics at T=200C (a),T=400C (b), T=500C (c), open circles, measured data, crosses,Kramers-Kronig transform test results. The insets show results of theK-K test in the form of the relative differences plot

The results of the K-K validation test are shown asinsets in Fig. 5a,b,c. The resulting “pseudo Chi-squared”value is also shown in the figure.

An important control tool is the distribution of theresiduals versus frequency. It is commonly known [eg.14, 21-24] that for a correct impedance data set, the rel-ative differences between the measured and calculatedvalue for the real part of impedance and the imaginarypart (residuals) should be randomly distributed aroundthe log(ν) axis.

In the present case, as indicated in the insets ofFig. 5, one can see a clear trace which shows devia-tion from K-K behavior. The deviation, however, is quitesmall as the residuals are less than 0.5%. The combina-tion of the measured data (open circles) and its K-Ktransformation (crosses) shown in Fig. 5 proves that theimpedance data of Bi4Ti3O12 ceramics is of a good qual-ity and can be used for the further analysis.

In order to extract as much information as possiblefrom impedance measurements, the experimental spectrawere subjected to CNLS fitting procedure with ZViewcomputer program (Scribner Associates, Inc.). The ele-ments of an equivalent circuit model represent the vari-ous (macroscopic) processes involved in the transport ofmass and charge. Equivalent circuit used for impedancespectroscopy data simulation in the present study wascomposed of resistors R and constant phase elements,CPE (symbol Q in Fig. 6). In solid materials, a distri-bution of relaxation times is usually observed and thecapacitance is replaced by a CPE, which represents thebehavior of the grain interior, grain boundary and elec-trode processes more accurately [25]. The CPE causesdepression of the ideal semicircle, observed on the com-plex plane plots by an angle β. Results of the data simu-lation in a form of estimated parameters of the relaxationprocesses are given in Tab. 2, whereas the combinationof the measured data and its CNLS-fit in a complexZ ′′–Z ′ plane are given in Fig. 7 (T=200C (a), T=400C(b), T=500C (c)). Calculations of were performed onthe base of the equivalent circuit parameters fitting andequations given elsewhere [21, 22].

The quality of the parameter fit of an equivalentcircuit to a set of impedance data can be best seen in fitquality plot (FQ-plot) [14, 21, 22]. The FQ – plot showsfrequency distributions of the residuals defined by:

∆Re,i =Z′i − Z

′(νi, ak)

|Z∗(νi, ak)| , (4)

∆Im,i =Z”

i − Z”(νi, ak)|Z∗(νi, ak)| , (5)

1143

Fig. 6. Schematic representation of the equivalent circuit used in the dispersion analysis of Bi4Ti3O12 ceramics within the temperature range∆T=200–5000C

TABLE 2Estimated values of resistance (Ri), capacitance (Ci), relaxation frequency (νmax,i) and depression angle (βi) calculated from equivalent

electric circuit parameters. The validity of the fitting procedure was estimated by χ2 and WSS

Parameter T=200C T=400C T=500C T=550C

R [Ω] 14.9 136.9 175.8 176.6

R1 [Ω] 1516.3 569.1 188.6 88.4

C1 [F] 1.75×10−10 2.88×10−10 6.49×10−10 1.56×10−10

ν1max [Hz] 0.599194×106 0.970258×106 1.29957×106 1.15456×106

β1 [deg] 23.56 14.79 3.31 -7.16

R2 [Ω] 1867.7 682.7 270.7 177.8

C2 [F] 1.29×10−8 1.17×10−8 0.88×10−8 0.78×10−8

ν2max [Hz] 6600 19881 66866 115379

β2 [deg] 20.23 25.89 27.53 26.11

R3 [Ω] 3539.9 990.8 330.6 192.3

C3 [F] 7.16×10−7 6.20×10−7 6.10×10−7 6.80×10−7

ν3max [Hz] 62 259 789 1217

β3 [deg] 35.52 37.30 49.03 52.67

WSS 1.82×10−4 5.46×10−4 1.55×10−4 3.59×10−9

χ2 1.20×10−6 3.59×10−6 1.05×10−6 2.42×10−11

where Z ′, Z ′′ – real and imaginary components of themeasured impedance, Z∗(ν, ak) – impedance model func-tion, ak – parameter of the model function.

The residual are plotted against frequency (ν) on alog scale (Fig. 8). For a good fit these deviations shouldbe distributed randomly around the frequency axis. Inour experiment a clear sinusoidal type trace is seen inFig.8 (T=200C (a), T=400C (b), T=500C) (c), how-ever, the residuals are less than 0.5%. In this connectionit is worth noting that a good quality of the simulationwas reached during present studies.

From the estimated values of grain, grain bound-ary and material/electrode interface resistances calculat-

ed from the semicircular arcs in the impedance spec-trum by CNLS fitting of experimental impedance spec-tra to the equivalent electric circuit parameters (Tab. 2)the corresponding values of conductivity (σ) at differ-ent temperatures in the measuring range were calculatedaccording to the general equation:

σ =l

S × R, (6)

where: l, S – dimensions, R – resistance.

1144

Fig. 7. Plot of Z ′′ vs. Z ′ for Bi4Ti3O12 at T=200C (a), T=400C (b), T=500C (c). Combination of the measured data (circles) and itsCNLS-fit (line) is given

1145

Fig. 8. FQ-plots for the final parameter set obtained with the dispersion analysis program for Bi4Ti3O12 ceramics at temperature range∆T=200–500C (T=200C (a), T=400C (b), T=500C (c))

1146

Fig. 9. Arrhenius plot of bulk (σB ) grain boundary (σGB) and material/electrode interface (σEL) conductivity for Bi4Ti3O12 ceramics

Fig. 10. Arrhenius plot of bulk (τB) grain boundary (τGB) and material/electrode interface (τEL) relaxation times for Bi4Ti3O12 ceramics

The variation of grain (σB) grain boundary (σGB)and material/electrode interface (σEL ) conductivity wasexpressed as a function of inverse absolute temperature(1000/T) and is shown in Fig. 9. The Arrhenius plot ofthese data made it possible to calculate the activationenergy for the electrical conduction process accordingto the formula:

σ =

(AT

)exp

(−Eca

kT

), (7)

where A is a constant, T the absolute temperature, Eca

the activation energy of conductivity, and k is the Boltz-mann’s constant [26].

1147

From the relaxation times, the activation energiesfor relaxations were calculated using the relation:

τ = τ0 exp(−Et

a

kT

), (8)

where Eta is the activation energy for relaxation. Fig. 10

shows the temperature dependence of relaxation time ina log – 1/T scale.

The results of calculation of activation energies forboth ac conductivity and relaxations for grain (bulk),grain boundary and electrode processes in polycrys-talline Bi4Ti3O12 ceramics are given in Tab. 3.

TABLE 3Activation energy for bulk (EB), grain boundary (EGB) and

electrode processes (EEL) for Bi4Ti3O12 ceramic

Parameter EB [eV] EGB [eV] EEL [eV]activation energyfor conductivity (σ) Ec

a0.60 0.40 0.54

activation energyfor relaxation (τ) Et

a0.07 0.56 0.46

Eca obtained in these samples in the temperature

range ∆T=200-500C is around Eca=0.6eV for grain in-

teriors, Eca=0.4eV for grain boundaries and Ec

a=0.54eVfor ceramic sample/electrode interface.

The activation energy for relaxation is aboutEt

a=0.07eV, Eta=0.56eV and Et

a=0.46eV for bulk, grainboundaries and electrode processes, respectively.

It is worth noting that Eca values of conductivity ob-

served in the present studies are consistent with the acti-vation energies of bismuth layer-structured ferroelectricsin which conductance is determined by oxygen vacancies[26]. It was found that the activation energies for relax-ation processes differ from those obtained from conduc-tivity plots. The activation energy for relaxation ascribedto grain boundaries is higher than the activation energyfor ac conductivity, whereas in case of grain interiorand material/electrode interface the activation energy ofconductivity is higher than the activation energies forrelaxations in the temperature range selected. In case ofbulk compound activation energy for relaxation is aboutten times smaller than activation energy for conductivity.Usually small activation energy is ascribed to instabilityin the systems [27].

Taking into consideration that both relaxation andconduction processes are dependent on defect–vacancyimpurity complexes and that the space charges de-veloped in the sample may also affect the relaxationprocesses–higher activation may be obtained from relax-ation temperature plot in case of grain boundary com-ponent.

4. Conclusions

Bismuth titanate (Bi4Ti3O12) ceramics was synthe-sized by the mixed oxide method. It was found that BiTceramics sintered at T=1000C adopted the orthorhom-bic symmetry of Fmmm space group and the followingelementary cell parameters: a=5.409(6)A, b=5.449(2)Aand c=32.816(2)A. Impedance diagrams of Bi4Ti3O12ceramics show three overlapping semicircles that werewell resolved by CNLS fitting. Each of them correspond-ed to the a particular microstructural region of the ceram-ic material, i.e. grain interior (bulk), grain boundariesand material/electrode interfaces. The proposed equiva-lent electric circuit consisted of a series combination ofthree parallel combinations of the resistance (R) and theconstant-phase element (Q). The experimental responsewas found to be in excellent agreement with the simu-lated one thus indicating that the proposed model gavean adequate representation of the electrical propertiesof the sample. From the estimated values of resistancesthe corresponding values of conductivity were calculat-ed and activation energies of both ac conductivity andrelaxations were calculated.

Acknowledgements

The present research has been supported by Polish Ministry ofEducation and Science from the funds for science in 2008-2011 as aresearch project N N507 446934.

REFERENCES

[1] B. A u r i v i l l i u s, Mixed bismuth oxides with layerlattices. The structure type of CaNb2Bi2O9, Arkiv forKemi 1, 54, 463-480 (1949).

[2] B. A u r i v i l l i u s, Mixed bismuth oxides with layerlattices. Structure of Bi4Ti3O12. Arkiv for Kemi 1, 58,499-512 (1949).

[3] B. A u r i v i l l i u s, Mixed oxides with layer lat-tices. Structure of BaBi4Ti4O15, Arkiv for Kemi 2, 37,519-527 (1950).

[4] A. L i s i ń s k a - C z e k a j, D. C z e k a j, Synthesisof Bi5TiNbWO15 ceramics, Archives of Metallurgy andMaterials 54, 4, 869-874 (2009).

[5] A. L i s i ń s k a - C z e k a j, M. C z a j a, L.K o z i e l s k i, D. C z e k a j, M. P i e c h o w i a k,M. N o w a k o w s k i, K. Z a w i ś l o k, Photolumi-nescence of nanocrystalline bismuth titanate thin filmssynthesized by the sol-gel method, Materials ScienceForum 514-516, 128-132 (2006).

[6] A. L i s i ń s k a - C z e k a j, E. J a r t y c h, M.M a z u r e k, J. D z i k, D. C z e k a j, Dielektrycznei magnetyczne właściwości ceramiki multiferroicznejBi5Ti3FeO15, Materiały Ceramiczne 63, 126-133 (2010).

1148

[7] R. W a s e r (Ed): Nanoelectronics and informatiomtechnology, Wiley-VCH, Weinheim, 2005.

[8] R.E. N e w n h a m, R.W. W o l f e, J.F. D o r r i a n,Structural basis of the bismuth ferroelectricity in titanatefamily, Materials Research Bulletin 6, 1029 (1971).

[9] A. L i s i ń s k a - C z e k a j, D. C z e k a j, Z.S u r o w i a k, Synthesis and dielectric properties ofSrBi3Ti2NbO12 ceramics with layer perovskite structure,Key Engineering Materials 206-213, 1429-1432 (2002).

[10] E. J a r t y c h, M. M a z u r e k, A. L i s i ń s -k a - C z e k a j, D. C z e k a j, Hyperfine interactions insome Aurivillius Bim+1Ti3Fem−3O3m+3 compounds, Jour-nal of Magnetism and Magnetic Materials 322, 51-55(2010).

[11] Z.S. M a c e d o, C.R. F e r r a r i, A.C. H e r n a n -d e s, Impedance spectroscopy of Bi4Ti3O12 ceramicproduced by self-propagating high-temperature synthe-sis technique, Journal of the European Ceramic Society24, 2567-2574 (2004).

[12] A. S n e d d e n, P. L i g h t f o o t, T. D i n g e s, M.I s l a m, Defect and dopant properties of the Aurivilliusphase Bi4Ti3O12, Journal of Solid State Chemistry 177,3660-3665 (2004).

[13] C. P a z d e A r a u j o, J.F. S c o t t, G.W. T a y l o r(Ed): Ferroelectric thin films: synthesis and basic proper-ties, Gordon and Breach Publishers, Amsterdam, 1996.

[14] B. B o u k a m p, A nonlinear least squares fit proce-dure for analysis of immitance data of electrochemicalsystems, Solid State Ionics 20, 31-44 (1986).

[15] P. Z o l t o w s k i, Non-traditional approach to measure-ment models for analysis of impedance spectra, SolidState Ionics 176, 1979-1986 (2005).

[16] M. G u i l l o d o, J. F o u l e t i e r, L. D e s s e -m o n d, P. D e l G a l l o, Electrical properties ofdense Me-doped bismuth vanadate (Me=Cu, Co)pO2-dependent conductivity determined by impedancespectroscopy, Journal of the European Ceramic Society21, 2331-2344 (2001).

[17] M.I. M o r o z o v, L.P. M e z e n t s e v a, V.V.G u s a r o v, Mechanism of formation of Bi4Ti3O12,Russian Journal of General Chemistry 72, 7, 1038-1040(2002).

[18] M. K r z h i z h a n o v s k a y a, S. F i l a t o v, V.G u s a r o v, P. P a u f l e r, R. B u b n o v a, M.

M o r o z o v, D.C. M e y e r, Aurivillius phases inBi4Ti3O12/BiFeO3 system: thermal behaviour and crys-tal structure, Zeitschrift Fur Anorganische Und Allge-meine Chemie 631, 1603-1608 (2005).

[19] N.A. L o m a n o v a, M.I. M o r o z o v, V.L.U g o l k o v, V.V. G u s a r o v, Properties of Au-rivillius phases in the Bi4Ti3O12- BiFeO3 system,Inorganic Materials 42, 2, 189-195 (2006).

[20] H. B e r n a r d, J. D z i k, A. L i s i ń s k a - C z e k a j,K. O s i ń s k a, D. C z e k a j, Zastosowanie metodyMOM do wytwarzania ceramiki Bi4Ti3O12, InżynieriaMateriałowa 178, 6, 1404-1408 (2010).

[21] D. C z e k a j, A. L i s i ń s k a - C z e k a j, T. O r k -i s z, J. O r k i s z, G. S m a l a r z, Impedance spec-troscopic studies of sol-gel derived barium strontiumtitanate thin films, Journal of the European Ceramic So-ciety 30, 465-470 (2010).

[22] B. W o d e c k a - D u ś, D. C z e k a j, Fabrication anddielectric properties of donor doped BaTiO3 ceram-ics, Archives of Metallurgy and Materials 54, 923-933(2009).

[23] B.A. B o u k a m p, Electrochemical impedance spec-troscopy in solid state ionics; Recent advances, SolidState Ionics 169, 1-4, 65-73 (2004).

[24] B.A. B o u k a m p, A linear Kronig–Kramers transfor-mation test for immitance data validation, Journal of theElectrochemical Society 142, 6, 1885-1894 (1995).

[25] A. L a s i a, Electrochemical impedance spectroscopyand its applications. In: Modern Aspects of Elec-trochemistry (ED): B.E.Conway, J.O.M.Bockris andR.E.White, Kluger Academic/Plenum Publishers, NewYork 1999.

[26] E. V e n k a t a R a m a n a, S.V. S u r y a -n a r a y a n a, T. B h i m a S a n k a r a m,AC impedance studies on ferroelectromagneticSrBi5−xLaxTi4FeO18 ceramics, Materials ResearchBulletin 41, 1077-1088 (2006).

[27] R. M a c h a d o, M.G. S t a c h i o t t i, R.L. M i g o n i,A. H u a n o s t a T e r a, First principles determinationof ferroelectric instabilities in Aurivillius compounds,Physical Review B70, 214112-1-214118-8 (2004).

Received: 20 March 2011.