Dielectric properties of CaCu3Ti4O12 based multiphased
ceramics
S. Guillemet-Fritsch, T. Lebey, M. Boulos and B. Durand
Centre InterUniversitaire de Recherche et d’Ingénierie des Matériaux (CIRIMAT/LCMIE),
Université Paul Sabatier, Bât. 2R1, 118, Route de Narbonne, 31062 Toulouse Cedex 04,
France
Laboratoire de Génie Electrique de Toulouse (LGET), Université Paul Sabatier, Bât. 3R3,
118, Route de Narbonne, 31062 Toulouse Cedex 04, France
Abstract
A “soft chemistry” method, the coprecipitation, has been used to synthesize the perovskite
CaCu3Ti4O12 (CCT). Three main types of materials were obtained for both powders and
sintered ceramics: a monophased consisting of the pure CCT phase, a biphased
(CCT + CaTiO3), and a three-phased (CCT + CaTiO3 + copper oxide (CuO or Cu2O)). These
ceramics, sintered at low temperature, 1050 °C, present original dielectric properties. The
relative permittivity determined in the temperature range (−150 < T < 250 °C) is significantly
higher than the one reported in the literature. Internal barrier layer capacitor is the probable
mechanism to explain the particular behaviour. Moreover, the presence of a copper oxide
phase beside the perovksite CCT plays an important role for enhancing the dielectric
properties.
Keywords: Powders-chemical preparation; Sintering; Dielectric properties; CaCu3Ti4O12;
Capacitors
1. Introduction
2. Experimental
2.1. Sample preparation
2.2. Characterization
3. Results
3.1. Powders
3.2. Ceramics
4. Discussion
5. Conclusion
Acknowledgements
References
1. Introduction
Many research have been done recently on the synthesis and characterization of the cubic
perovskite CaCu3Ti4O12, or Ca0.25Cu0.75TiO3 ceramic (commonly called CCT), because of its
unusual electrical properties. Subramanian et al.1 have first reported the high dielectric
constant at 1 kHz in ACu3Ti4O12 and ACu3Ti3FeO12 phases (A = trivalent rare earth or Bi).
Permittivity values ( ′) higher than 10,000 have been reported for ceramics1 and 2 and for
single crystals.3
The real part of the permittivity has been even increased from 10,000 to 300,000, as a result
of grain growth.4 Studies of CaCu3Ti4O12 thin films prepared by pulsed laser deposition
confirm that large mean grain size may be responsible for the high dielectric constant.5
Moreover, the high dielectric constant does not depend on the temperature, which makes it
even more attractive for technological applications, even if the dielectric losses, which are in
most cases very high, have also to be taken into account.
The crystal structure of CCT, determined from powder neutron diffraction method6 and 7 is
well described in the space group Im3. The stabilization of the observed structure is attributed
to the strong preference of Cu for the square planar geometry. No phase transition has been
detected between 25 and 1000 °C, neither using powder diffraction7 nor by Raman
spectroscopy studies.8
Though different hypotheses have been given to explain the high permittivity value, its origin
is not fully understood. Subramanian et al.1 and 2 suggested that the polarizability and the
dielectric constant are enhanced by the tension on the Ti
O bonds, but that the transition to
the ferroelectric state is frustrated by the TiO6 octahedra tilt structure that accommodates the
square planar coordination of Cu2+. According to the same authors, the dielectric constant of
the CCT ceramic is also enhanced by its microstructure and is sensitive to the Cu/Ca ratio. A
characteristic gap energy of 28 meV was calculated for the relaxational mode.2 In a more
recent work, the same authors9 indicated that the numerous twins observed in CCT single
crystals may influence the dielectric properties and that the very high dielectric constant of
CaCu3Ti4O12 might be an extrinsic property. Cohen et al.10 also suggested that the large
dielectric constant observed in CCT monocrystals probably arisen from spatial
inhomogeneities of local dielectric response, like twins, Ca ordering and antiphase
boundaries. Hassini et al.,11 using infrared spectroscopy, indicated that the high permittivity
value is not related to phonons but rather to a relaxational motion of Debye-type with a
characteristic relaxation time expected in the gigahertz range. Sinclair et al.4 believed,
according to impedance spectroscopy studies, that the high permittivity is associated to
internal barrier layer capacitor and not to an intrinsic feature of the bulk crystal structure.
They demonstrated that CaCu3Ti4O12 ceramics are electrically heterogeneous and consist of
semiconducting grains with insulating grain boundaries. They also showed the influence on
the effective permittivity of the ceramic grain size, obtained from various dwell time at the
sintering temperature.12 The “giant” effective permittivity value for the ceramics sintered for
24 h has been associated to the presence of, either thin reoxidized grain boundary regions on
the outer surface of the large semiconducting grains, or to a secondary phase at the grain
boundaries which has not been detected by SEM. Homes et al.,13 using infrared measurements
on CaCu3Ti4O12 and CdCu3Ti4O12, agree with the internal barrier layer capacitance effect, as
well as Lunkenheimer et al.,14 that explain the colossal effect by Maxwell–Wagner-type
contributions of depletion layers at the interface between sample and contacts or at grain
boundaries. Optical conductivity measurements on CCT single crystals have provided more
information about the physics explaining the giant dielectric effect in this material.3 The
discrepancy, noticed between low dielectric constant (80) at infrared frequencies and high
constant (105) at lower radio frequencies, indicates a strong absorption due to dipole
relaxation. So, in the last few years, several authors considered that the exceptional dielectric
properties of CCT are due to opposite electrical properties of the bulk and internal boundaries
regions (one is insulating and the other conducting).
Most of the papers reported in the literature mentioned that the CCT materials studied were
single phased (CCT). Only Subramanian et al.2 indicated the presence of a minute impurity of
CuO detected by X-ray diffraction analysis. Kobayashi et al.15 showed a drop of for a factor
102 when introducing impurities (manganese) and explained it by a dipole moment causing
fluctuating ferroelectric domains in CaCu3Ti4O12.
Most of the CCT samples were obtained by solid/solid reactions. Nanocrystalline CCT
powders were also prepared by mechanical alloying.16 Only few “soft chemistry” methods
were used to prepare CCT. The polymeric citrate precursor route was mentioned.17 The
powder grains prepared from this method are smaller than the ones obtained by solid/solid
reaction. The dielectric constant is also higher in the first case, but largely lower than the 104
reported by Subramanian.1 Hassini et al.11 synthesized both CaCu3Ti4O12 and CaTiO3 by an
organic gel-assisted citrate process, described in.18
In this paper, we present new results concerning the elaboration of CCT powders by the
coprecipitation method. The elaboration of the corresponding ceramics and their physicochemical characterization are then described. Finally, the influence of the structure (presence
of additional phases) and the microstructure (grain size) on the dielectric properties of the
ceramics is shown.
2. Experimental
2.1. Sample preparation
The CCT powders were prepared using one of a soft chemistry technique, i.e. the
coprecipitation of oxalate precursors. The starting materials were: TiCl3 (Prolabo, d = 1.20,
%min = 15), CaCl2 (Prolabo, %min = 96) and CuCl2·2H2O (Prolabo, %min = 99). They were
dissolved in water in various proportions and the coprecipitation was performed by addition
of a solution of oxalic acid dissolved in ethanol. TiCl3 was chosen because it was easier to
handle than TiCl4. The full oxidation of Ti3+ in Ti4+ in the solution was ensured by air
bubbling during the reaction. It was evidenced by the visual observation of the color variation
from white to clear blue. The volume of water was taken much lower than the one of ethanol,
in order to decrease the dielectric constant of the precipitation media. Since the nucleation of
the particles is favoured compared to the growth, smaller particles, homogeneous in size and
composition, are likely to be obtained.19 The solution was aged for a couple of hours, and the
precipitate obtained was centrifuged. The precursors were then pyrolized in air at 900 °C for
10 h to obtain the oxides. These powders were mixed with an organic binder, then uniaxially
pressed into disk (diameter: 6 mm and thickness: 1.5 mm) under 250 MPa pressure. Green
ceramics were sintered in static air at 1050 °C for 24 h.
2.2. Characterization
The accurate chemical composition of the oxide powder was determined by analyzing the Ni
and Mn cations, using an induced coupled plasma spectroscopy (ICP AES Thermo-Optec
ARL3580). Three trials were performed for each sample and the average value was
calculated. The oxide powder morphology was observed with a Scanning Electron
Microscope (JEOL JSM-6400). In some cases, backscattered electrons were used to obtain an
image in chemical contrast. The specific surface area was determined using a Micrometrics
accusorb 2100 E. The phase composition was determined by X-ray diffraction analysis: the
data were collected with a SEIFERT XRD-3003-TT diffractometer using the Cu Kα radiation
(λ = 0.15418 nm).
The electrical characterization consists in measuring the effective permittivity * using a
Hewlett Packard HP 4284A precision LCR meter. The domains explored were 20 Hz to
1 MHz for the frequency, ambient to 160 °C for the temperature. A sinusoidal voltage of
magnitude 1 VAC was applied. In some particular cases and thanks to DEA system, the
permittivity is measured between 0.1 Hz and 100 kHz and from −120 to 220 °C. It has been
checked that this change in the experimental set up has no impact on the values obtained.
3. Results
3.1. Powders
The chemical composition, the phase composition, the specific surface area and the crystallite
size of the different powders prepared by pyrolysis of the oxalate precursors at 900 °C for
10 h are reported in Table 1. The X-ray diffraction patterns are given in Fig. 1. With the
different amount of calcium, copper and titanium analysed in the oxide powder, three
different kinds of materials were obtained: almost pure cubic CaCu3Ti4O12 (JCPDS 75-2188)
(a minute impurity of CaTiO3 is observed), named CCT-A; CaCu3Ti4O12 and a significant
amount of the CaTiO3 phase (JCPDS 46-2400) (CCT-B), and a mixture of three phases:
CaCu3Ti4O12, CaTiO3 and copper oxide, CuO (JCPDS 80-1917) (CCT-C series). In the case
of polyphased material, the amount of the second or third phase (CaTiO3 and CuO) was
estimated from the ratio of the relative maximal peak intensity of the two phases CCT/CaTiO3
and CCT/CuO (Table 1). The influence of the initial chemical composition of the oxide
powder (Ca, Cu, Ti content) on the different phases formation is obvious. As expected, the
pure CaCu3Ti4O12 phase is obtained only when the ratio of calcium, copper and titanium are
close to the stoichiometric ones. The CaTiO3 phase appears if an excess of titanium is present,
and at the same time when the copper content slightly decreases. We suggest that it is the
excess of titanium that leads to the precipitation of CaTiO3, even if there is no excess of
calcium. Since both Ca and Ti form a second phase, Cu is then in excess, with respect with
the stoichiometry of CaCu3Ti4O12. Hence the precipitation of the copper oxide CuO is
observed, beside the CCT and CaTiO3 phases, when the titanium content is lower than 4.00
(CCT-C series), and as the copper content varies in the range 2.68–3.24. So we can suggest
that the Ti content mostly controls the phase composition (single or multiphased material) of
the CCT powder. There is no data on the phase diagram of the Ca/Cu/Ti/O system in the
literature. The present study points out that the cubic perovskite CaCu3Ti4O12 is a defined
compound.
Table 1.
Chemical analysis, phase composition, specific surface area and crystallite size of the
CaxCuyTizO12 oxide powders obtained at 900 °C in air
Reference
Chemical
analysis
Phase
composition
Intensity
ratioa
CCT/CaTiO
Intensity
ratio
CCT/CuO
Sw (m2/g)b
dBET (nm)c
21
–
3.0
400
5
–
6.8
200
3
x
Ca
y
Cu
z
Ti
CCT-A
1.2
4
2.7
6
4.0
0
CCT-B
1.8
8
2.1
2
4.0
0
CCT-C1
1.4
0
2.8
8
3.7
2
CCT+
CaTiO3+Cu
O
5
6
2.0
600
CCT-C2
1.6
4
2.6
8
3.6
8
CCT+
CaTiO3+Cu
O
10
11
2.6
450
CCT-C3
1.1
2
3.2
4
3.6
4
CCT+
CaTiO3+Cu
O
15
10
1.6
740
CCTd (+
traces
CaTiO3)
CCT+CaTiO
3
a
Estimated from the relative ratio of the maximal peak intensity of the two phases.
b
Determined from BET measurements.
c
Calculated from specific surface area data.
d
CCT means CaCu3Ti4O12.
Fig. 1. X-ray diffraction pattern of CCT-based powders calcined at 900 °C for 10 h. CCT-A:
monophased Ca0.25Cu0.75TiO3 (*) (a), CCT-B: Ca0.25Cu0.75TiO3 (*) and CaTiO3 (+) (b) and
CCT-C1: Ca0.25Cu0.75TiO3 (*), CaTiO3 (+) and CuO (o) (c).
The specific surface area and the crystallite size deduced from our data do not vary much in
the whole studied composition range (Table 1). Only the CCT-B powder seems to have
smaller grains, for unclear reasons. So the presence of CaTiO3 and CuO beside the CCT phase
does not influence the powder reactivity. SEM observations of the calcined powders (Fig. 2)
indicate that the grains are homogeneous in shape and size, approximately ranging from 0.5 to
1.5 m, for each composition, i.e. whatever the structure of the powder (single or multiphased
materials).
Fig. 2. SEM micrographs of CCT-based powders calcined at 900 °C for 10 h. CCT-A
(monophased) (a), CCT-B (biphased) (b) and CCT-C1 (three-phased) (c).
3.2. Ceramics
The green ceramics were sintered at 1050 °C for 24 h in static air atmosphere. The phase
composition and the density of the sintered ceramics are reported in Table 2. The density of
the ceramics is in the range 4.5–4.8, whatever the composition. The phase composition of the
ceramics is similar to the one of the powders. However, CCT-A ceramic, contrary to CCT-A
powder, is a pure CaCu3Ti4O12. No impurity of CaTiO3 is observed. For one ceramic of the
CCT-C series (CCT-C3), X-ray diffraction analysis (Fig. 3) shows the presence of additional
copper oxide, Cu2O (JCPDS 78-2076). The crystallization of the Cu2O phase, indicating the
presence of Cu+ cations, is not surprising. In fact, the Cu2+ cations are reduced at high
temperature (T ≈ 1000 °C) into Cu+. When the cooling rate is low enough, the copper cations
reoxidize again during the cooling. In the present work, all the ceramics have been cooled at
the same rate, i.e. 150 °C/h. In fact, the Cu2O phase precipitates in the ceramic whose
composition is the richest in copper (CCT-C3).
Table 2.
Density, phase composition and grain size of the CCT ceramics sintered at 1050 °C during
24 h in air
a
Reference
Density
Phase composition
Grain sizea ( m)
CCT-A (monophased)
4.8
CCT
0.7–1
CCT-B (biphased)
4.6
CCT + CaTiO3
1.5–3
CCT-C1 (3 phased)
4.8
CCT + CaTiO3 + CuO
1–50
CCT-C2 (3 phased)
4.5
CCT + CaTiO3 + CuO
15–90
CCT-C3 (3 phased)
4.6
CCT + CaTiO3 + CuO + Cu2O
10–100
Estimated from SEM observations.
Fig. 3. X-ray diffraction pattern of CCT-based ceramics sintered at 1050 °C for 24 h.
monophased CCT: Ca0.25Cu0.75Ti1O3 (a), biphased CCT: Ca0.25Cu0.75Ti1O3 + CaTiO3 (b),
three-phased CCT-C1 Ca0.25Cu0.75Ti1O3 + CaTiO3 + CuO (c), and CCT-C3:
Ca0.25Cu0.75Ti1O3 + CaTiO3 + CuO + Cu2O (d).
The ceramic surfaces exhibit different aspects (Fig. 4). The grain sizes, determined from SEM
observations are reported in Table 2. A significant difference in grain sizes is observed,
ranging from approximately 1 m for CCT-A to a maximum of 100 m for CCT-C3. In fact,
the huge grains are observed only in the ceramics containing the copper oxide phase (either
CuO or Cu2O) (see for example the microstructure of one of the CCT-C series on Fig. 4).
Moreover, those ceramics also show a particular microstructure. On the grain boundaries, a
liquid phase wets the grains. In order to evidence this phenomenon, a three-phased ceramic
(CCT-C1) has been quenched in air after sintering, so as to retain—at ambient temperature the
high temperature microstructure. The SEM observations (in chemical contrast) of the so
obtained ceramic (Fig. 5) clearly show an ex-liquid phase surrounding the CCT grains. We
suspect that the copper oxide transforms into a liquid phase during the sintering treatment and
leads to anomalous grain growth, as it has been reported for BaTiO3.20 The ceramics
containing only the CCT phase or CCT + CaTiO3 have grains whose size ranges from 1 to
3 m, with a narrow size distribution (CCT-A and CCT-B) (Fig. 4a and b).
Fig. 4. SEM micrographs of the surface of sintered CCT-based ceramics sintered at 1050 °C
for 24 h. CCT-A (monophased) (a), CCT-B (biphased) (b), three-phased CCT (C1 and C2) (c)
and CCT-C3 (containing Cu2O) (d).
Fig. 5. SEM micrograph (chemical contrast) of the surface of a sintered CCT-based ceramics
(CCT-C1) sintered at 1050 °C for 24 h and quenched in air.
TEM observations allow detailing the structure of grain boundaries of a three-phased CCT
(CCT-C1; Fig. 6). No defect is observed on the grain boundary: no particular microstructure,
no twins or two-dimensional defect.
Fig. 6. TEM observations of grain boundaries of three-phased CCT (CCT-C1) ceramic
sintered at 1050 °C for 24 h.
A summary of the changes in the relative permittivity, measured at 1 kHz, versus the
temperature (40 ≤ T ≤ 160 °C), is given in Fig. 7 for the different types of materials. Four
different behaviors are observed. They are likely to depend on various characteristics like the
ceramic structure (monophased or multiphased) and/or microstructure (grain size) and/or
processing conditions.
Fig. 7. Changes in the relative part of the permittivity ( ′) (f = 1 kHz) vs. temperature (range
40 ≤ T ≤ 160 °C) for the different samples.
First of all and in the temperature range under study, very high permittivity values are
measured (ranging from few thousands (CCT three phases samples C3) to values higher than
hundreds of thousands (CCT three phased samples C1&C2)). No ferroelectric-like transition
is noticed. Moreover, a quasi-independent behavior of the permittivity with temperature is
observed.
As expected the monophased CCT ceramics present the behavior already reported in the
literature.1, 2, 9 and 13 In this case, the permittivity is slightly higher than 20,000 and does not
depend on the temperature. Biphased CCT ceramics present a linear increase of the
permittivity which reaches a value of 20,000 at 120 °C. For temperature higher than 120 °C
the permittivity is larger than the value obtained for monophased CCT. Last, two different
behaviors are observed for the three-phased CCT materials. For the first type, permittivity
values greater than 105 are observed. This value is one order of magnitude larger than the
value of a monophased CCT material. This result is reproducible as it is shown for samples
C1 and C2. The changes in the permittivity values of these materials are slightly temperature
dependent. For the second type of materials (sample C3) the permittivity value is lower or
equal to 103, with a slight increase in the temperature range 40–60 °C. It is one order of
magnitude lower than the value of a monophased CCT material.
The frequency and the temperature range of the measurement have been extended for the
monophased and the three-phased materials.
Fig. 8 represents the change in the real part of the permittivity ( ′) versus the temperature at
1 kHz for the monophased and three-phased CCT. Fig. 9 gives the change in tgδ (i.e. the
dielectric losses ( ′′/ ′)) at the same frequency (the inlet is a focus in the lowest temperature
range, showing a relaxation like phenomena). It is important to notice that for temperature
values up to 120–150 °C, the losses are lower than 1. For higher temperatures, the losses
become so large that the capacitive nature of the material is questionable.
Fig. 8. Changes in the relative part of the permittivity ( ′) (f = 1 kHz) vs. temperature (range
−120 ≤ T ≤ 220 °C) for CCT-A (monophased) and CCT-C1 (three-phased) samples.
Fig. 9. Changes in tgδ (f = 1 kHz) vs. temperature (range −120 ≤ T ≤ 220 °C) for CCT-A
(monophased) and CCT-C1 (three-phased) samples.
Regarding ′, the ceramics under study do not exactly present the often reported behavior.
The mechanism consists in a switch from the response of the grain boundary (which is a
dielectric) to the response of the bulk regions of the material, the grains being considered as
semi conductive.4 This switch, always associated to a relaxation phenomenon, produces a
change of at least three decades in the permittivity value. In our case and whatever the
material, the permittivity does not tend to the very low value (of few tens or one hundred)
reported for the lowest temperatures. The decrease in the permittivity of a CCT monophased
sample is plotted in Fig. 10 at four different temperatures (−126 °C, −116 °C, −106 °C,
−96 °C). It is obvious that the data show a slight increase over most of the frequency range
tending to at least two different plateau, the first one in the low frequency range (f < 1 Hz)
with ′ near 20,000, the second one in the intermediate frequency range (1 Hz < f < 10 kHz)
with ′ near 10,000. Then a drastic drop is observed in the MHz range. With increasing the
temperature, the magnitude of the permittivity increases and the different observed switches
shift to higher frequencies. The two last phenomena (the plateau around 10,000 and the switch
to lower values in the high frequency range) are consistent with those previously reported.1, 2
and 3
They are attributed to the change of the physical nature of the contribution from the grain
boundary to the grain as depicted in Fig. 10. Last, the existence of a first plateau for the
lowest frequencies is explained by the nature of the interphases. These results will be
discussed in details in the following.
Fig. 10. Changes in the relative part of the permittivity ( ′) vs. frequency, at four temperatures
(−126 °C, −116 °C, −106 °C, −96 °C) for CCT-A (monophased) sample.
The changes of ′ and tgδ versus frequency are plotted at different temperatures (Fig. 11a–f
and Fig. 12a–f, respectively) for the monophased and the three-phased CCT. The sudden
change of tgδ appears at various temperatures, depending on the nature of the material. All
these plots demonstrate the existence of a thermally activated Debye-like relaxation in the low
temperature region at an intermediate frequency which is generally attributed to a grain
boundary response. Assuming an Arrhenius behavior, an activation energy may be determined
according to Eq. (1):
(1)
where fm is the frequency of the maximum, Ea the activation energy, kB the Boltzman
constant,
Fig. 11. Changes in the relative part of the permittivity ( ′) and in tgδ vs. frequency, at four
temperatures, −126 °C (a), −96 °C (b), −61 °C (c), RT (d), 57 °C (e), 157 °C (f) for CCT-A
(monophased) sample.
Fig. 12. Changes in the relative part of the permittivity ( ′) and in tgδ vs. frequency, at four
temperatures, −115 °C (a), −96 °C (b), −60 °C (c), RT (d), 55 °C (e), 155 °C (f) for CCT-C1
(three-phased) sample.
A value of 0.18 eV is obtained for the monophased CCT ceramic and of 0.53 eV for the threephased CCT materials. Regarding this last one, it is in good agreement with the one proposed
by West et al.13 (0.6 eV) or by Capsoni et al29 (0.54–0.73 eV) in their works on pure
CaCu3Ti4O12 ceramics. The value of the activation energy for the monophased CCT sample is
quite different, ranging between the grain barrier one (near 0.6 eV) and the bulk one
(0.09 eV), according to the values determined by these authors.
In summary and, whatever the material under study, (mono, bi or three-phased CCT):
– The global “classical” behavior of CCT samples is observed,
– The lowest value of the permittivity measured is at least one order of magnitude larger than
the value reported in the literature. However, as illustrated in Fig. 10, another drop (associated
to a relaxation phenomena) is probably occurring in a frequency range higher than 1 MHz at
the lowest measurement temperature (−126 °C in this case). This drop probably leads to
another plateau in the permittivity value which could be measured at very low temperature (in
the 10–100 K range). This temperature range is unfortunately not available in our laboratory.
– A very high value (near 400,000) is obtained for the three phases CCT at 200 °C. At this
temperature, the capacitive nature of the material is questionable since the dielectric losses are
very high. However, if we still consider it as a dielectric ceramic, the temperature at which
this value is obtained is much lower than the one reported by Subramanian1 (400 °C) to reach
the same order of magnitude. Note that in1 no indication of the losses is given.
Therefore, the materials under study have properties which seem to depend on their structure
(nature of the different phases). These results and their most likelihood origins are discussed
in the following.
4. Discussion
It is obvious that, whatever the material considered (CCT-A, B or C), its electrical behavior is
in marked contrast with the ferroelectric one resulting from structural distortion since no
structural phase transition is detected. Therefore, this cannot be invoked to explain the
electrical properties.
The three kinds of materials essentially differ from the overall chemical composition (Table
1), leading both to different grain sizes (Table 2), and particularly to different structures
(monophased CCT or multiphased materials). The copper oxide phase plays an important role
in these multiphased ceramics, both from the microstructural properties (grain size) and from
the electrical properties. As a matter of fact, CuO has been used as an additive to lower the
sintering temperature, for example in (Ba0.8Sr0.2)(Ti0.9Zr0.1)O321 material which is used as a
surface barrier layer capacitor. Its melting during the sintering enhances the grain growth. In
these materials, a bimodal grain size distribution with fine grained matrix containing grains
exhibiting an exaggerated growth was observed. On the other hand, the amount of CuO in the
ceramic seems to play an important role. Hence, for BSTZ materials, 0.5 wt.% is insufficient
to improve the dielectric characteristics while more than 1 wt.% is deleterious.21 In our case,
an amount of more than 1 or 2 wt.% of CuO is present in the ceramics. Moreover, the huge
grains observed for the ceramics containing the copper oxide (CCT three-phased series) are
also probably related to the appearance of a liquid phase that “wets” the grains during the
sintering.
Therefore, the role played by the grain size (strongly related to the previous discussion) has to
be emphasized too. The largest the grain size (three-phased CCT) the highest the permittivity
value. Hence, the monophased CCT sample presenting a sharp size distribution of small
grains has the lowest permittivity. The biphased CCT sample (CCT-B) has an “in between”
behavior since its permittivity value increases linearly with temperature. The permittivity is
lower than the one of monophased CCT for T < 120 °C and then becomes higher as the
temperature increases. Last, the largest permittivity values are obtained for the three-phased
ceramics (CCT-C1, and C2), i.e. the ones presenting the largest grain size. Such a behavior
has already been reported in La1.5Sr0.5NiO4 phase22 or in CCT23 and 24. It leads to the same
conclusions, i.e. an increase in the grain size involves an increase in permittivity.
Finally, the particular high frequency behavior (i.e. the drop and the plateau to the low
permittivity value) usually mentionned in the literature for “pure” CCT samples, is not
observed in our monophased samples, even at the lowest temperature investigated. The low
value of the permittivity observed once the drop has occurred is associated to the bulk (i.e. the
grain itself). It is consistent with the behavior observed in other semiconducting perovskite.
On the other hand, it is well known that the relaxation frequency depends on the size of the
relaxing species. Our results tend to prove that the grain size of monophased CCT materials
synthesized using coprecipitation is probably lower than any other CCT material reported in
the literature. Another hypothesis to explain the difference observed is that the internal
structure of the grain itself may be different in our case. However, the sole grain size variation
cannot account for the permittivity value nor for its dependence on the number of phases
present in the material.
The most plausible explanation, even if the discussion is still open, is an extrinsic origin. It is
not the intrinsic structure of the material itself which confers the properties (except for the
ones which could be measured at very high frequency or at very low temperature, that are
related to the grain itself).
The extrinsic explanation rests on the consideration of the interfaces. Two types of interfaces
have to be considered:
– the grain boundaries leading to an internal barrier layer capacitance effect13,
– the electrode interfaces leading to electrode polarization effects.25
This last point is associated to thin depletion layers which may exist at the electrodes/bulk
interfaces, due to the formation of Schottky diodes. Such behavior has been observed in
CaCu3Ti4O12 or Cu2Ta4O12. The grains being conductors or at least semi-conductors, these
depletion layers lead to the “apparent” colossal value of the permittivity. For the samples
under study, the results obtained on CCT-C1 and CCT-C2, whose general characteristics
(grain size, chemical composition, …) and metallization do not differ from sample CCT-C3,
tend to demonstrate that, at least in our case, the electrode polarization effect is not the
relevant mechanism.
The high “apparent” permittivity is therefore probably associated with an internal barrier layer
capacitor: the bulk of the grains would behave like other semiconducting perovskites while
the grain boundaries and/or the shell would be insulating layers. Hence, for the same sample
thickness, the grain size and the nature of the grain boundary are likely to have an impact on
the final electrical properties. Decreasing the number of grain boundaries (by increasing the
grain size) and keeping the other parameters constant, would produce an increase in the
apparent permittivity, according to relation 2:
(2)
where d is the grain size, t the boundary layer thickness and
insulating layer.25 and 26
b
the permittivity of the
For the biphased CCT samples, the existence of CaTiO3 (a low loss dielectric ceramic of
permittivity 17027 and 28) could be considered as responsible at the same time of an increase in
the value of the insulating layer permittivity and of a modification of the conductivity of the
grains. Such an assumption could explain both the linear behavior observed and the changes
in the tgδ versus frequency. The theoretical expression of the losses (tgδ) is reminded:
(3)
where ε″r corresponds to the part of the losses associated to the relaxation phenomena and
σ/ω 0 the part due to a pure conduction mechanism, with σ the conductivity, ω the pulsation
and 0 the vacuum permittivity.
In Fig. 13, the plot of tgδ versus the frequency for different temperatures and the associated
sketch makes clearly appear the possible existence of the sum of these two mechanisms.
Fig. 13. Changes in tgδ value vs. frequency for CCT-C1 (three-phased) sample and proposed
sketch.
Another possible explanation is related to the interface between the grains which is modified
in the case of the multiphased material. This assumption is confirmed in the case of the threephased CCT ceramics. The grain size increases and the presence of CuO oxide seems to play
the same double role as CaTiO3 (doping of the grain boundary, modifying the permittivity
value), but the melting of the copper oxide phase during the sintering treatment probably
favors the wetting of the shell of the grain leading to a multilayered material as suggested
before. Hence, the charge carriers’ concentration at the different interfaces rises with an
increase in Cu amount. More charges accumulate on both sides of the boundary layers leading
to a higher dielectric constant of the samples. An optimum value of the CuO content must
exist.21 It has been found for CCT-C1 and CCT-C2 but not for CCT-C3 sample. It is obvious
that all these assumptions have to be confirmed and works are still under progress.
5. Conclusion
Different powders containing several phases, CaCu3Ti4O12 (major phase), CaTiO3 and CuO
have been obtained via the calcination at 900 °C of the oxalate precursors, synthesized by
coprecipitation. The corresponding ceramics present huge relative permittivity values and
relatively low losses. The presence of additional phases (CaTiO3 and copper oxides, CuO
or/and Cu2O) gives these ceramics a behavior different from the one observed in pure CCT
materials. The relative permittivity is significantly higher than the one reported up to now
(both at low temperature and at 160 °C).
These CCT ceramics, presenting a real part of the permittivity higher than 250,000, are
considered for internal barrier layer capacitor (IBLC) applications. It is even more remarkable
that these high values can be obtained from a single step-processing route, in air, at relatively
low temperature. It implies that reproducible and fine tuning of the dielectric properties of this
material is possible.
Acknowledgements
The authors thank Ch. Calmet for performing the SEM pictures and J.J. Demai for the TEM
observations.
References
M.A. Subramanian, D. Li, N. Duan, B.A. Reisner and A.W. Sleight, High dielectric constant
in ACu3Ti4O12 and ACu3Ti3FeO12, J. Solid State Chem. 151 (2000), pp. 323–325.
A.P. Ramirez, M.A. Subramanian, M. Gardel, G. Blumberg, D. Li, T. Vogt and S.M. Shapiro,
Giant dielectric constant response in a copper-titanate, Solid State Commun. 115 (2000), pp.
217–220.
C.C. Homes, T. Vogt, S.M. Shapiro, S. Wakimoto and A.P. Ramirez, Optical response of high
dielectric constant perovskite related oxide, Science 293 (2001), pp. 673–676.
D.C. Sinclair, T.B. Adams, F.D. Morrison and A.R. West, One-step internal barrier layer
capacitor, Appl. Phys. Lett. 80 (2002) (12), pp. 2153–2155.
Y.L. Zhao, G.W. Pan, Q.B. Ren, Y.G. Cao, L.X. Feng and Z.K. Jiao, High dielectric constant
in CaCu3Ti4O12 thin film prepared by pulsed laser deposition, Thin Solid Films 445 (2003),
pp. 7–13.
B. Bochu, M.N. Deschizeaux, J.C. Joubert, A. Coullomb, J. Chevanas and M. Marezio,
Synthèse et caractérisation d’une série de titanates perovskites isotypes de CaCu3Mn4O12,J.
Solid State Chem. 29 (1979), pp. 291–298.
S.M. Moussa and B.J. Kennedy, Structural studies of the distorted perovskite
Ca0.25Cu0.75TiO3, Mater. Res. Bull. 36 (2001), pp. 2525–2529.
Kolev, N., Bontchev, R. P., Jacobson, A. J., Popov, V. N., Hadjiev, V. G., Litvinchuk, A. P.
and Iliev, M. N., Raman spectroscopy of CaCu3Ti4O12. Phys. Rev. B 2002, 66, 132102-1-4.
M.A. Subramanian and A.W. Sleight, ACu3Ti4O12 and ACu3Ru4O12 perovskites: high
dielectric constants and valence degeneracy, Solid State Sci. 4 (2002), pp. 347–351.
M.H. Cohen, J.B. Neaton, L. He and D. Vanderbilt, Extrinsic models for the dielectric
response of CaCu3Ti4O12, J. Appl. Phys. 94 (2003) (5), pp. 3299–3305.
A. Hassini, M. Gervais, J. Coulon, V.T. Phuoc and F. Gervais, Synthesis of Ca0.25Cu0.75TiO3
and infrared characterization of role played by copper, Mater. Sci. Eng. B87 (2001), pp. 164–
168.
T.A. Adams, D.C. Sinclair and A.R. West, Giant barrier layer capacitance effects in
CaCu3Ti4O12 ceramics, Adv. Mater. 14 (2002) (18), pp. 1321–1323.
C.C. Homes, T. Vogt, S.M. Shapiro, S. Wakimoto, M.A. Subramanian and A.P. Ramirez,
Charge transfer in the dielectric constant materials CaCu3Ti4O12 and CdCu3Ti4O12, Phys. Rev.
B 67 (2003) (092106), pp. 1–4.
P. Lunkenheimer, V. Bobnar, A.V. Pronin, A.I. Ritus, A.A. Volkov and A. Loidl, Origin of
apparent colossal dielectric constants, Phys. Rev. B 66 (2002) (052105), pp. 1–4.
W. Kobayashi and I. Terasaki, Unusual impurity effects on the dielectric properties of
CaCu3−xMnxTi4O12, Physica B 329–333 (2003), pp. 771–772.
A.F.L. Almeida, R.S. De Oliveira, J.C. Goes, J.M. Sasaki, A.G. Souza Filho, J. Mendes Filho
and A.S.B. Sombra, Structural properties of CaCu3Ti4O12 obtained by mechanical alloying,
Mater. Sci. Eng. B96 (2002), pp. 275–283.
P. Jha, P. Arora and A.K. Ganguli, Polymeric citrate precursor route to the synthesis of the
high dielectric constant oxide CaCu3Ti4O12, Mater. Lett. 57 (2003), pp. 2443–2446.
A. Douy and P. Odier, The polyacrylamide gel: a novel route to ceramic and glassy oxide
powders, Mater. Res. Bull. 24 (1989), pp. 1119–1126.
C. Villette, Ph. Tailhades and A. Rousset, Sur l’élaboration par chimie douce de ferrites de
cuivre à forts champs coercitifs, C.R. Acad. Sci. Paris 316 (1993) (II), pp. 1717–1723.
M. Demartin, C. Hérard, C. Carry and J. Lemaître, Dedensification and anomalous grain
growth during sintering of undoped barium titanate, J. Am. Ceram. Soc. 80 (1997) (5), pp.
1079–1084.
C.F. Yang, Improvement of the sintering and dielectric characteristics of surface barrier layer
capacitors by CuO addition, Jpn. J. Appl. Phys. 35 (1996), pp. 1806–1813.
Rivas, J., Rivas-Murias, B, Fondado, A., Mira, J. and Senaris-Rodriguez, M. A., Room
temperature colossal dielectric constant in the charge-ordered two-dimensional nickelate
La1.5Sr0.5NiO4 (unpublished).
A.R. West, T.B. Adams, F.D. Morrison and D.C. Sinclair, Novel high capacitance materials:
BaTiO3: La and CaCu3Ti4O12, J. Eur. Ceram. Soc. 24 (2004), pp. 1439–1448.
J. Li, K. Cho, N. Wu and A. Ignatiev, Correlation between dielectric properties and sintering
temperatures of polycrystalline CaCu3Ti4O12, IEEE Trans. Dielectrics Electr. Insulation 11
(2004) (3), pp. 534–541.
Lunkenheimer, P., Fichtl, R., Ebbinghaus, S. G. and Loidl, A., Non-intrinsic origin of the
colossal dielectric constants in CaCu3Ti4O12 (unpublished).
Wu, J., Nan, C. W., Lin, Y. and Deng Y., Giant dielectric permittivity observed in Li and Ti
doped NiO. Phys. Rev. Lett. 2002, 89, 21, 217601-1-2176014.
K. Wakino, Recent development of dielectric resonator materials and filters in Japan,
Ferroelectrics 91 (1989), pp. 69–86.
V.M. Ferreira, F. Azough, J.L. Baptista and R. Freer, Magnesium titanate microwave
dielectric ceramics, Ferroelectrics 133 (1992), pp. 127–132.
D. Capsoni, M. Bini, V. Massarotti, G. Chiodelli, M. Mozzatic and C. Azzoni, Role of doping
CuO segregation in improving the giant permittivity of CaCu3Ti4O12, J. Solid State Chem 177
(2004), pp. 4494–4500.
Corresponding author. Tel.: +33 5 61 55 62 83; fax: +33 5 61 55 61 63.
Original text : Elsevier.com