Structural parameters of AgNbO3 at T=298K (Yashima et al., 2011). Number of formula units of AgNbO3 in a unit cell: Z=8. Unit-cell parameters: a = 15.64773(3) Å, b = 5.55199(1) Å, c = 5.60908(1) Å, α=β=γ= 90 deg., Unit-cell volume: V = 487.2940(17) Å3. U (Å2)=Isotropic atomic displacement parameter.
Many ferroelectric oxides possess the ABO3 perovskite structure (Mitchell, 2002), in which the A-site cations are typically larger than the B-site cations and similar in size to the oxygen anion. Figure 1 shows a schematic drawing for this structure, where the A cations are surrounded by 12-anions in the cubo-octahedral coordination and the B cations are surrounded by 6-anions in the octahedral coordination. An ideal perovskite exhibits a cubic space group Pm3m. This structure is centrosymmetric and cannot allow the occurrence of ferroelectricity that is the presence of a switchable spontaneous electric polarization arising from the off-center atomic displacement in the crystal (Jaffe et al., 1971; Lines & Glass, 1977). The instability of ferroelectricity in the perovskite oxides is generally discussed with the Goldschimidt tolerance factor (t) (Goldschmidt, 1926 and Fig. 2),
where r O, r A, and r B are the ionic radii of the O, A, and B ions. For a critical value t=1, the cubic paraelectric phase is stable. This unique case can be found in SrTiO3, which has an ideal cubic perovskite structure at room temperature and doesn’t show ferroelectricity down to the absolute 0 K (Müller & Burkard, 1979). However, ferroelectricity can be induced by the substitution of O18 in this quantum paraelectric system at T<T c~23 K (Itoh et al., 1999). When t>1, since the B-site ion is too small for its site, it can shift off-centeringly, leading to the occurrence of displacive-type ferroelectricity in the crystal. Examples of such materials are BaTiO3 and KNbO3 (Shiozaki et al., 2001). On the other hand, for t<1, the perovskite oxides are in general not ferroelectrics because of different tilts and rotations of BO6 octahedra, which preserve the inversion symmetry. But exceptions may be found in the Bi-based materials, in which large A-site displacement is observed. This large A-site displacement is essentially attributed to the strong hybridization of Bi with oxygen (Baettig et al., 2005). Similar cases are observed in Pb-based materials, which commonly have large Pb displacement in the A-site (Egami et al., 1998) and strong covalent nature due to the unique stereochemistry of Pb (Cohen, 1992; Kuroiwa et al., 2001).
Although BaTiO3- and PbTiO3-based ceramics materials have been widely used in electronic industry (Uchino, 1997; Scott, 2000), there remain some importance issues to be solved. One of such challenges is to seek novel compounds to replace the Pb-based materials, which have a large Pb-content and raises concerns about the environmental pollution (Saito et al.,2004; Rodel et al.,2009 ).
The discovery of extremely large polarization (52 µC/cm2) under high electric field in the AgNbO3 ceramics (Fu et al., 2007) indicates that Ag may be a key element in the designs of lead-free ferroelectric perovskite oxides (Fu et al., 2011a). With the advance of first-principles calculations (Cohen, 1992) and modern techniques of synchrotron radiation (Kuroiwa et al., 2001), we now know that the chemical bonding in the perovskite oxides is not purely ionic as we have though, but also possesses covalent character that plays a crucial role in the occurrence of ferroelectricity in the perovskite oxides (Cohen, 1992; Kuroiwa et al., 2001 ). It is now accepted that it is the strong covalency of Pb with O that allows its large off-center in the A-site. Although Ag doesn’t have lone-pair electrons like Pb, theoretical investigations suggest that there is hybridization between Ag and O in AgNbO3(Kato et al., 2002; Grinberg & Rappe, 2003,2004), resulting in a large off-center of Ag in the A-site of perovskite AgNbO3 (Grinberg & Rappe, 2003,2004). This prediction is supported by the results from X-ray photoelectron spectroscopy, which suggest some covalent characters of the chemical bonds between Ag and O as well as the bonds between Nb and O (Kruczek et al., 2006). Moreover, bond-length analysis also supports such a theoretical prediction. Some of the bond-lengths (~2.43 Å) in the structure (Sciau et al., 2004; Yashima et al. 2011) are significantly less than the sum of Ag+ (1.28 Å) and O2- (1.40 Å) ionic radii (Shannon, 1967). All these facts make us believe that AgNbO3 may be used as a base compound to develop novel ferroelectric materials. Along such a direction, some interesting results have been obtained. It was found that ferroelectricity can be induced through the chemical modification of the AgNbO3 structure by substitution of Li (Fu et al., 2008, 2011a), Na(Arioka, 2009; Fu et al., 2011b), and K (Fu et al., 2009a) for Ag. Large spontaneous polarization and high temperature of para-ferroelectric phase transition were observed in these solid solutions. In the following sections, we review the synthesis, structure, and dielectric, piezoelectric and ferroelectric properties of these solid solutions together with another silver perovskite AgTaO3 (Soon et al.,2009, 2010), whose solid solutions with AgNbO3 are promising for the applications in microwaves devices due to high dielectric constant and low loss (Volkov et al. 1995; Fortin et al., 1996; Petzelt et al., 1999; Valant et al., 2007a).
Both single crystal and ceramics of AgNbO3 are available. Single crystal can be grown by a molten salt method using NaCl or V2O5 as a flux (Łukaszewski et al., 1980; Kania, 1989). Ceramics samples can be prepared through a solid state reaction between Nb2O5 and silver source (Francombe & Lewis, 1958; Reisman & Holtzberg, 1958). Among the silver sources of metallic silver, Ag2O and Ag2CO3, Ag2O is mostly proper to obtain single phase of AgNbO3(Valant et al., 2007b). For silver source of Ag2O, thermogravimetric analysis indicates that phase formation can be reach at a firing temperature range of 1073-1397 K (Fig. 3). The issue frequently encountered in the synthesis of AgNbO3 is the decomposition of metallic silver, which can be easily justified from the color of the formed compounds. Pure powder is yellowish, while grey color of the powder generally indicates the presence of some metallic silver. It has been shown that the most important parameter that influences the phase formation is oxygen diffusion (Valant et al., 2007b). In our experiments to prepare the AgNbO3 ceramics, we first calcined the mixture of Ag2O and Nb2O5 at 1253 K for 6 hours in O2 atmosphere and then sintered the pellet samples for electric measurements at 1323 K for 6 hours in O2 atmosphere (Fu et al., 2007). Insulation of these samples is very excellent, which allows us to apply extremely high electric field to the sample (Breakdown field >220 kV/cm. For comparison, BaTiO3 ceramics has a value of ~50 kV/cm.)
2.2 Electric-field induced ferroelectric phase
Previous measurements on D-E hysteresis loop by Kania et al. (Kania et al., 1984) indicate that there is small spontaneous polarization P s in the ceramics sample of AgNbO3. P s was estimated to be ~0.04 µC/cm2 for an electric field with an amplitude of E=17 kV/cm and a frequency of 60 Hz. Our results obtained at weak field have confirmed Kania’s reports (Fig. 4 and Fu et al., 2007). The presence of spontaneous polarization indicates that AgNbO3 must be ferroelectric at room temperature. The good insulation of our samples allows us to reveal a novel ferroelectric state at higher electric field. As shown in Fig.4, double hysteresis loop is distinguishable under the application of E~120 kV/cm, indicating the appearance of new ferroelectric phase. When E>150 kV/cm, phase transformation is nearly completed and very large polarization was observed. At an electric field of E=220 kV/cm, we obtained a value of 52 µC/cm2 for the ceramics sample. Associating with such structural change, there is very large electromechanical coupling in the crystal. The induced strain was estimated to be 0.16% for the ceramics sample (Fig.5). The D-E loop results unambiguously indicate that the atomic displacements are ordered in a ferri-electric way rather than an anti-ferroelectric way in the crystal at room temperature.
2.3. Room-temperature structure
There are many works attempting to determine the room-temperature structure of AgNbO3 (Francombe & Lewis, 1958; Verwerft et al., 1989; Fabry et al., 2000; Sciau et al., 2004; Levin et al., 2009). However, none of these previous works can provide a non-centrosymmetric structure to reasonably explain the observed spontaneous polarization. Very recently, this longstanding issue has been addressed by R. Sano et al. (Sano et al., 2010). The space group of AgNbO3 has been unambiguously determined to be Pmc2 1 (No. 26) by the convergent-beam electron diffraction (CBED) technique, which is non-centrosymmetric and allows the appearance of ferroelectricity in the crystal (Fig.6).
On the basis of this space group, M. Yashima (Yashima et al., 2011) exactly determined the atom positions (Table 1) in the structure using the neutron and synchrotron powder diffraction techniques. The atomic displacements are schematically shown in Fig.7. In contrast to the reported centrosymmetric Pbcm (Fabry et al. 2000; Sciau et al. 2004; Levin et al. 2009), in which the Ag and Nb atoms exhibit antiparallel displacements along the b-axis, the Pmc2 1 structure shows a ferri-electric ordering of Ag and Nb displacements (Yashima et al., 2011) along the c-axis of Pmc2 1 orthorhombic structure. This polar structure provides a reasonable interpretation for the observed polarization in AgNbO3.
2.4. Dielectric behaviours and phase transitions
Initial works on the phase transitions of AgNbO3 and their influence on the dielectric behaviors were reported by Francombe and Lewis (Francombe & Lewis, 1958) in the late 1950s, which trigger latter intensive interests in this system (Łukaszewski et al., 1983; Kania, 1983, 1998; Kania et al., 1984, 1986; Pisarski & Dmytrow, 1987; Paweczyk, 1987; Hafid et al., 1992; Petzelt et al., 1999; Ratuszna et al., 2003; Sciau et al., 2004). The phase transitions of AgNbO3 were associated with two mechanisms of displacive phase transition: tilting of oxygen octahedra and displacements of particular ions (Sciau et al., 2004). Due to these two mechanisms, a series of structural phase transitions are observed in AgNbO3. The results on dielectric behaviors together with the reported phase transitions are summarized in Fig.8. Briefly speaking, the structures of the room-temperature (Yashima et al., 2011) and the high temperature phases (T> T O1-O2=634 K ) (Sciau et al., 2004) are exactly determined, in contrast, those of low-temperature (T<room temperature) and intermediate phases ( T C FE=345 K <T< T O1-O2) remain to be clarified. In the dielectric curve, we can see a shoulder around 40 K. It is unknown whether this anomaly is related to a phase transition or not. It should be noticed that Shigemi et al. predicted a ground state of R3c rhombohedra phase similar to that of NaNbO3 for AgNbO3 (Shigemi & Wada, 2008). Upon heating, there is an anomaly at T C FE=345 K, above which spontaneous polarization was reported to disappear (Fig.8 (c) and Kania et al., 1984). At the same temperature, anomaly of lattice distortion was observed (Fig.8 (c)and Levin et al., 2009). The dielectric anomaly at T C FE=345 K was attributed to be a ferroelectric phase transition. Upon further heating, there is a small peak at T=453 K, which
|Atom||573 K||645 K||733 K||903 K|
is not so visible. However, it can be easily ascertained in the differential curve or in the cooling curve. This anomaly is nearly unnoticed in the literatures (Łukaszewski et al., 1983; Kania, 1983, 1998; Kania et al., 1986; Pisarski & Dmytrow, 1987; Paweczyk, 1987; Hafid et al., 1992; Ratuszna et al., 2003). The detailed examination of the temperature dependence of the 220o d-spacing (reflection was indexed with orthorhombic structure) determined by Levin et al. (Levin et al., 2009 and Fig.8(c)) reveals anomaly that can be associated with this dielectric peak. These facts suggest that a phase transition possibly occurs at this temperature. Around 540 K, there is a broad and frequency-dependent peak of dielectric constant, which is also associated with an anomaly of 220o d-spacing. However, current structural investigations using x-ray and neutron diffraction do not find any symmetric changes associated with the dielectric anomalies at 456 K and 540 K, and the structure within this intermediate temperature range was assumed to be orthorhombic Pbcm (Sciau et al., 2004). At T=T C AFE=631 K, there is a sharp jump of dielectric constant due to an antiferroelectric phase transition (Pisarski et al., 1987; Sciau et al., 2004). The atomic displacement patterns in the antiferroelectric phase (Pbcm) are shown in Fig.7 (b) using the structural parameters determined by Ph Sciau et al. (Sciau et al., 2004). For T>T C AFE, there are still three phase transitions that are essentially derived by the tilting of oxygen octahedral and have only slight influence on the dielectric constant. For conveniences, the tilting of octahedral (Sciau et al., 2004) described in Glazer’s notation is given in Fig.8 and the structure parameters (Sciau et al., 2004) are relisted in Table 2.
3. (Ag1-x Li x )NbO3 solid solution
Li can be incorporated into the Ag-site of AgNbO3. However, due to the large difference of ionic radius of Li+(0.92Å), and Ag+( 1.28Å)(Shannon, 1976), the solid solution is very limited. Nalbandyan et al.(Nalbandyan et al., 1980), systematically studied the stable and metastable phase equilibrias and showed that solid solution limit is narrow (x~0.02) for the stable phase, but is relatively wide (x~0.12) for the metastable phase(Sakabe et al., 2001; Takeda et al., 2003; Fu et al., 2008, 2011a). With a small substitution of Li for Ag (x>x c =0.05~0.06), a ferroelectric rhombohedra phase is evolved in the solid solution (Nalbandyan et al., 1980; Sakabe et al., 2001; Takeda et al., 2003;Fu et al., 2008, 2011a). In this solid solution, the strong off-center of small Li (Bilc & Singh, 2006) plays an important role in triggering the ferroelectric state with large spontaneous polarization (Fu et al., 2008, 2011a).
Single crystals of (Ag1-x Li x )NbO3 can be obtained by the melt growth process (Fu et al., 2008). Stoichiometric compositions of Ag2O, Li2CO3, and Nb2O5 were mixed and calcined at 1253 K for 6 h in an oxygen atmosphere. The calcined powder was milled, put into an alumina container, and melted at 1423 K for 4 h in an oxygen atmosphere. The melt was cooled to 1323 K to form the crystal at a rate of 4 K/h, followed by furnace cooling down to room temperature. Using this process, single crystals with size of 1–3 cm can be obtained for the (100) p (Hereafter, subscript p indicates pseudocubic structure) growth face. Due to the volatility of lithium at high temperature, the exact chemical composition of the crystal is generally deviated from the starting composition and is required to be determined with methods like inductively coupled plasma spectrometry.
Ceramics samples can be prepared by a solid state reaction approach. Mixtures of Ag2O, Nb2O5, and Li2CO3 were calcined at 1253 K for 6 h in O2 atmosphere, followed by removal of the powder from the furnace to allow a rapid cooling to prevent phase separation. The calcined powder was milled again and pressed to form pellets that were sintered at 1323 K for 6 h in O2 atmosphere, followed by a rapid cooling.
The structural refinements using the powder X-ray diffraction data suggest that (Ag1-x Li x )NbO3 solid solution with x>x c has the space group of R3c (Fu et al., 2011a). Table 3 lists the structural parameters of this model for composition x=0.1. Figure 9 shows a schematic drawing for this structure. In this rhombohedral R3c phase, the spontaneous polarization is essentially due to the atomic displacements of the Ag/Li, Nb, and O atoms along the pseudocubic  direction.
|Ag0.9Li0.1NbO3 (R3c, No.161, T=room temperature)|
3.3. Ferroelectric and piezoelectric properties
Evolution of the polarization state in Ag1-x Li x NbO3 solid solutions is shown in Fig.10. Basically, when x<x c, the solid solutions have the ferrielectric state of pure AgNbO3 with a small spontaneous polarization at zero electric field. In contrast, when x>x c, a normal
ferroelectric state with large value of remanent polarization (P r) is observed. All ceramics samples show P r value comparable to P S of BaTiO3 single crystal (26 µC/cm2) (Shiozaki et al., 2001). Moreover, the polarization in Ag1-x Li x NbO3 solid solution is very stable after switching. Large P r value and ideal bistable polarization state of Ag1-x Li x NbO3 ceramics may be interesting for non-volatile ferroelectric memory applications. Measurements on single crystal samples (Fig.11) indicate that saturation polarization along the <111> p rhombohedra direction (~40 µC/cm2) is greatly larger than that along the <001> p tetragonal direction (~24 µC/cm2) and the ratio between them is (Fu et al., 2008), which is in good agreement with results of structural refinements. This suggests that the polar axis is the <111> p direction of pseudo-cubic structure.
The strain-E loops indicate that there are good electromechanical coupling effects in Ag1-x Li x NbO3 crystals. Although the spontaneous polarization is along the <111> p axis, the <001> p -cut crystal shows larger strain and less hysteresis than the <111> p -cut one (Fig.11 (b) and (c)). These phenomena are very similar to those reported for the relaxor-ferroelectric crystals (Wada et al., 1998). The most significant result exhibited from Ag1-x Li x NbO3 single crystal is its excellent g 33 value that determines the voltage output of the piezoelectric device under the application of an external stress (Fu et al., 2008). The g 33 value together the d 33 value and dielectric constants for the <001> p -cut single crystals are shown in Fig. 12. The high g 33 value is a direct result from the large d 33 constant and the low dielectric constant of the single crystal.
3.4. Dielectric behaviours and proposed phase diagram
Figure 13 shows the dielectric behaviours of the ferroelectric Ag1-x Li x NbO3 solid solutions. For comparison, the temperature dependence of dielectric constant of AgNbO3 is also shown. It can be seen that solid solution with x > x c shows different temperature evolutions of the dielectric constant as compared with AgNbO3. In sharp contrast to the complex successive phase transitions in AgNbO3, ferroelectric Ag1-x Li x NbO3 solid solutions (x>x c) show only two phase transitions in the temperature range of 0-820 K. Polarization measurements suggest that the high temperature phase () is nonpolar, thus it seems
that the higher temperature phase transition is not related to a ferroelectric phase transition. On the basis of the dielectric measurements, the phase diagram of Ag1-x Li x NbO3 solution is summarized in Fig.14, where T, O, R and M represent tetragonal, orthorhombic, rhombohedra, and monoclinic symmetries, respectively. At room temperature, structure transformation from O to R phase at x c dramatically changes the polar nature of Ag1-x Li x NbO3. It is a ferrielectric with small spontaneous polarization in the O phase, but becomes a ferroelectric with large polarization in the R phase.
4. (Ag1-x Na x )NbO3 solid solution
The ionic radius of Na+ (1.18 Å) is comparable to that of Ag+ (1.28 Å) (Shannon, 1976), allowing to prepare the Ag1-x Na x NbO3 solid solution within the whole range of composition (x=0-1) (Kania & Kwapulinski, 1999). Kania et al. previously performed investigation on the dielectric behaviors and the differential thermal analysis for the Ag1-x Na x NbO3 solid solutions, and stated that the solid solution evolves from disordered antiferroelectric AgNbO3 to normal antiferroelectric NaNbO3. As described in section §2.3, we now know that AgNbO3 is not antiferroelectric but rather is ferrielectric at room temperature (Yashima et al., 2011). Moreover, recent reexamination on the polarization behaviors of stoichiometric and non-stoichiometric NaNbO3 polycrystallines indicates that the reported clamping hysteresis loop of NaNbO3 can be interpreted by pining effects while stoichiometric NaNbO3 is intrinsically ferroelectric (Arioka et al., 2010). Therefore, reexamination on this solid solution is necessary. Evolution of the polarization with composition clearly indicates that the solid solution evolves from ferrielectric AgNbO3 to ferroelectric NaNbO3 (Fu et al., 2011b).
Ag1-x Na x NbO3 solid solution was prepared by a solid state reaction approach. Mixtures of Ag2O(99%), Nb2O5 (99.99%), and Na2CO3 (99.99%) were calcined at 1173 K for 4 h in O2 atmosphere. The calcined powder was ground, pressed into pellet with a diameter of 10 mm at thickness of 2 mm, and sintered with the conditions listed in Table 4.
|x=0.4, 0.5, 0.6||1323K||5h||O2|
Figure 15 shows the change in polarization with composition in the Ag1-x Na x NbO3 solid solutions. For a wide range of composition x<0.8, the solid solution possesses the characteristic polarization behaviours of pure AgNbO3: small spontaneous polarization at E=0 but large polarization at E> a critical field. This fact suggests that the solid solution is ferrielectric within this composition range. This is also supported by the temperature dependence of dielectric constant (Fig.16). On the other hand, for the Na-rich composition, particularly, x>0.8, we observed large remanent polarization with value close to the saturation polarization at high field. This result indicates that a normal ferroelectric phase is stable in these compositions. Therefore, polarization measurements show that the Ag1-x Na x NbO3 solid solution evolves from ferrielectric AgNbO3 to ferroelectric NaNbO3 (Fu et al., 2011b).
4.3. Dielectric properties
The dielectric properties of the Ag1-x Na x NbO3 solid solutions are summarized in Fig.16. The change in dielectric behaviours with composition is very similar to that observed in polarization measurements (Fig.15). For x≤0.8, the solid solution shows successive phase transitions similar to pure AgNbO3. In contrast, it has the characteristic phase transition of pure NaNbO3 for x>0.8. The composition dependence of transition temperature derived from the dielectric measurements is shown in Fig.17. Two noticed features may be seen: (a) the thermal hysteresis is extremely large for antiferroelectric phase transition and reaches a value greater than 100 K at x~0.5. Such large thermal hysteresis is rarely observed in normal polar phase transition. (b) It seems that there is a phase boundary at x=xc~0.8 (Fu et al., 2011b), around which structural transformation between ferri- and ferro-electric phases occurs.
5. (Ag1-x K x )NbO3 solid solution
(Ag1-x K x )NbO3 solid solutions are available only for very limited composition (Weirauch & Tennery, 1967; Łukaszewski, 1983; Kania, 2001). Weirauch et al. reported that solid solution of AgNbO3 in KNbO3 was limited to slightly less than 6 mole % and solid solution of KNbO3 in AgNbO3 was limited to less than 0.5 mole % (Weirauch & Tennery, 1967). However, our process indicates that KNbO3 and AgNbO3 can be alloyed with each other within 20 mole % (Fu et al., 2009a). Apparently, the reported solid solution limit is dependent on the process. In our samples, we found that a ferroelectric phase with large spontaneous polarization can be induced by substitution of K for Ag for x>x c1=0.07. This ferroelectric phase shows nearly composition-independent ferroelectric phase transition. On the other hand, the K-rich solid solution (x>0.8) possesses the ferroelectric phase transition sequence of pure KNbO3 and the transition temperature is dependent on the composition.
(Ag1-x K x )NbO3 solid solutions were prepared by a solid state reaction method. Mixture of Ag2O, Nb2O5, and K2CO3 were calcined at 1173 K for 6 h in O2 atmosphere with a slow heating rate of 1 K/min. The calcined powder was milled again, pressed in a 6-mm steel die with a pressure of 10 MPa to form the pellets, which were then preheated at 773 K for 2 h, followed by a sintering at temperature of 1253–1323 K (1323 K for x=0–0.1 and 1.00, 1273 K for x=0.15 and 0.90, and 1253 K for x=0.17 and 0.80, respectively) for 3 h in O2 atmosphere. The atmosphere and heating rate are found to have significant influences on the phase stability of the solid solution.
5.2. Structural change with composition
Figure 18 shows the change in lattice parameters with composition in the Ag1-x K x NbO3 solid solutions. When the amount of substitution is small, the solid solution possesses the orthorhombic structure of pure AgNbO3. In the phase boundary x c1=0.07, structural transformation occurs and the phase changes into a new orthorhombic structure. In this new ferroelectric phase, the lattice constants show linear increase with composition. Interestingly, the orthorhombic distortion angle βis nearly independent with the composition. In the K-rich region (x>x c3=0.8), the solid solution has the orthorhombic structure of pure KNbO3 at room-temperature, which is also ferroelectric. In contrast to nearly unchanged orthorhombic angle βin the Ag-rich orthorhombic phase, βshows monotonous decreases with x in the K-rich phase.
5. 3. Ferroelectric and piezoelectric properties
Figure 19 shows typical results of polarization and strain behaviors observed at room temperature for the Ag1-x K x NbO3 solid solutions. Similar to pure AgNbO3, merely a small spontaneous polarization was observed in samples with x<x c1. However, when x>x c1, a normal D-E loop with large value of remanent polarization P r was observed. A value of P r=20.5 µC/cm2 was obtained for a ceramics sample with x =0.10, which is greatly larger than that observed for BaTiO3 ceramics (Fu et al., 2010). These results show that Ag-rich orthorhombic phase (x c1<x<x c2 ) is really under a ferroelectric state with large polarization. For K-rich region x>x c3, normal D-E loops were also obtained. Associating with the evolution into the ferroelectric phase, butterfly strain curve were observed. The piezoelectric constants determined from the piezo-d 33 meter generally have values of 46–64 pC/N for these ferroelectric samples (Fu et al., 2009a).
5.4. Dielectric behaviours and proposed phase diagram
Associating with the change in structure, dielectric behaviours of the Ag1-x K x NbO3 solid solution also change with composition. As shown in Fig.20, the temperature dependence of dielectric constant can be sorted by three types: (1) AgNbO3-type for x<x c1=0.07, (2) KNbO3-type for K-rich region x>x c3=0.8, and (3) a new type for the intermediate composition x c1<x<x c2. In this intermediate composition, dielectric measurements indicate that there are two phase transitions within the temperature range of 0-750 K. One transition locates at T c2~420 K with thermal hysteresis and shows small change in dielectric constant, which seems to be due to a ferro-to-ferro-electric phase transition. Another phase transition occurs at T c1~525 K. Around this transition, the dielectric constant changes sharply and follows exactly the Curie-Weiss law. The Curie-Weiss constant was estimated to be 1.47 *105 K for x=0.1 sample, which is a typical value for the displacive ferroelectric transition, suggesting that this is a displacive type ferroelectric transition. A phase diagram is proposed in Fig.21, in which PE, FE, and AFE represent the paraelectric, ferroelectric, and antiferroelectric phases, respectively. When carefully comparing the phase transition temperature with the orthorhombic angle β(Fig.18) due to the ferroelectric distortion (For x c1<x<x c2 and x>x c3. In contrast, for x< x c1, β change is basically due to the oxygen-octahedral tilting.), one might find that there is a correlation between β and the temperature of the ferroelectric phase transition.
AgTaO3 is another oxide of the two discovered silver perovskites (Francombe & Lewis, 1958). It is generally accepted that AgTaO3 undergoes a series of phase transitions from rhombohedral phase (T≤685 K) to monoclinic phase (650 K≤T≤703 K) and then to tetragonal phase (685 K<T≤780 K), and finally to cubic phase at T T-C= 780 K upon heating (Francombe & Lewis, 1958; Kania, 1983; Paweczyk, 1987; Kugel et al.,1987; Hafid et al., 1992). However, due to the coexistence regions between rhombohedral and monoclinic, and monoclinic and tetragonal, the actual transition temperatures from rhombohedral to monoclinic T R-M as well as that from monoclinic to tetragonal T M-T still remain uncertain. Furthermore, the ground state and the origins that trigger these phase transitions still remain to be addressed (Wołcyrz & Łukaszewski,1986; Kugel et al.,1987;Soon et al., 2010).
Although single crystal of AgTaO3 is available through a flux method (Łukaszewski et al., 1980; Kania, 1989), it is extremely difficult to prepare its dense ceramics sample (Francombe & Lewis, 1958;Kania,1983) for electrical measurements. Since decomposition of AgTaO3 occurs at 1443±3 K in atmosphere (Valant et al., 2007b), sintering cannot be performed at higher temperatures to obtain dense ceramics. However, this long-standing synthesis difficulty now can be solved by a processing route involving the conventional solid-state reaction and sintering in environment with a high oxygen pressure at ~13 atm (Soon et al., 2010). In this synthesizing route, Ag2O and Ta2O5, first underwent a grind mixing and was calcined at 1273 K for 6 hours. The calcined powder was then pressed into a pellet in 6 mm in diameter. Sintering was carried out by placing the powder compact into a sealed zirconia tube that was connected to a pressure control valve (Fig.22). Prior to the sintering, oxygen gas at ~6.25 atm was filled into the sealed zirconia tube after the evacuation. Upon heating, the pressure of sealed oxygen gas increased and reached ~13 atm when the powder compact was sintered at 1573K for 2 hours. This eventually led to formation of dense polycrystalline AgTaO3.
6.2. Phase formation and dielectric behaviors
X-ray diffraction analyses (Fig.23) suggest the AgTaO3 is rhombohedral with symmetry at room temperature (Wołcyrz & Łukaszewski, 1986). Diffraction patterns obtained at 68.4 K remain unchanged, indicating that such nonpolar phase persists down to low-temperatures. This is also supported by the measurements on the dielectric constants (Fig.24) and heat capacity (Fig.25), in which no anomaly was probed in the low temperatures.
Although several frequency-dependent peaks are seen in the dielectric loss (Fig.24(b)), it can be reasonably attributed to polarization relaxations due to defects (Soon et al., 2010). Interestingly, within the low-temperature region, the dielectric behavior follows the Barrett’s relation (Barrett, 1952) that is characteristic for the quantum paraelectric system (Abel, 1971; Höchli & Boatner,1979; Itoh et al., 1999), suggesting that AgTaO3 may be a quantum paraelectric. On the other hand, two step-like dielectric anomalies corresponding to the phase transitions from monoclinic to tetragonal and tetragonal to cubic were observed at 694 and 780 K, respectively, upon heating the samples (Fig.26). This observation is in agreement with the previous reports (Kania,1983;Kugel et al., 1987). Furthermore, the temperature dependence of 1/ε for AgTaO3 shows non-linear behavior, which is similar to that of KTaO3, obeying the modified form of the Curie-Weiss law ε=ε L+C/(T-T 0) (Rupprecht & Bell, 1964).
7. (Ag1-x Li x )TaO3 solid solution
Similar to the case of AgNbO3, ~12 mole% of Li can be incorporated into the Ag-site of AgTaO3 to form (Ag1-x Li x )TaO3 (Soon et al., 2009). Although the transition temperature is lower than room temperature, ferroelectricity can be induced in this solid solution due to the strong off-centering nature of the small Li ions.
(Ag1-x Li x )TaO3 was prepared by the conventional solid-state reaction with Ag2O, Ta2O5 and Li2CO3, where the powder mixture was calcined at 1273 K for 6 hours. The calcined powder was then pressed into a pellet with 6 mm in diameter. Sintering was carried out by the same high-pressure process used for pure AgTaO3, which eventually led to formation of dense ceramics samples of (Ag1-x Li x )TaO3
7.2. Dielectric behaviours and confirmation of ferroelectric phase
Figures 27& 28 plot the temperature dependence of dielectric constant ε and loss tanδ for (Ag1-x Li x )TaO3 with x≤ 0.12 obtained at frequencies ranging from 1 kHz to 1 MHz, respectively. It can be seen that a dielectric peak was gradually induced by Li+ substitution in AgTaO3. In contrast to the single peak of the dielectric constant, there are two to four peaks of the dielectric loss within the same temperature window. Since the additional loss peaks do not associate with a remarkable change in the dielectric constant, it is very likely due to the defect effects dependent with sample processing. It can be seen that a well-defined peaks has occurred in the ~MHz frequency regions for the substitution at extremely low level, for example, at x=0.02. This indicates the existence of local polarization in the doped crystal (Vugmeister & Glinchuk, 1990; Samara, 2003). This fact again suggests that AgTaO3 is actually under a critical state of the quantum paraelectric. Any slight modification will lead to the appearance of observable polarization in the system. For small substitution, the location of the dielectric anomaly depends on the observed frequency. Figure 29 gives an evaluation on this frequency dependence. In the figure, the temperature-axis is scaled with the peak position of 1 MHz, making it easy to see the change with composition. For x=0.008, the frequency dispersion is very strong, peak position of the dielectric constant shifts about 50% with respect to that of 1 MHz for f= 1 kHz. However, for x=0.035, such change is less than 2%, meaning that ferroelectric phase transition temperature T c is well defined in the sample. Thus, we can infer that a macroscopic ferroelectric phase is evolved below T c in this composition. This was also confirmed by the results shown in Fig.30, in which ferroelectric loop was obtained for T<T c for a sample with x=0.12 that has the same dielectric behaviors to that of x=0.035.
On the basis of the above results, a phase diagram is proposed in Fig.31, in which PE and FE represents the paraelectric and ferroelectric phases, respectively. As mentioned above, since the peak of the dielectric constant is strongly dependent with frequency for 0<x<0.035, the gray zone in the phase diagram may be attributed to dipole-glass phase(Vugmeister & Glinchuk, 1990;Pirc & Blinc, 1999;Samara, 2003) or a phase with nanosized ferroelectric domains (Fisch, 2003; Fu et al., 2009b), which remains to be addressed by further investigations.
8. Concluding remarks
Our recent works reveal that silver perovskites are of great interests from either the view-point of fundamental research or that of application research in the fields of ferroelectric or piezoelectric. Promising ferroelectric and piezoelectric properties have been demonstrated in some compounds such as (Ag,Li)NbO3 and (Ag,K)NbO3 alloys, but further works are required to improve the material performance, to understand the basic physics of the ferroelectricity/piezoelectricity of the materials, and to seek novel promising compounds among the discovered solutions or alloys with other ferroelectric systems. Moreover, integration techniques of thin films are also a direction for the future works when considering the practical applications.
Part of this work was supported by the Collaborative Research Project of Materials and Structures Laboratory of Tokyo Institute of Technology, and Grant-in-Aid for Scientific Research, MEXT, Japan.