Improving the functional properties of (K0.5Na0.5)NbO3 piezoceramics by acceptor doping

This is the post-print (i.e. final draft post-refereeing) of the publication. The final publication is available at Elsevier via http://dx.doi.org/10.1016/j.jeurceramsoc.2014.08.033 Improving the functional properties of (K0.5Na0.5)NbO3 piezoceramics by acceptor doping X.Vendrell1*, J.E. García2, X.Bril3, D.A. Ochoa2, L.Mestres1, G.Dezanneau3 1Departament de Química Inorgànica, Universitat de Barcelona, 08028 Barcelona, Spain 2Department of Applied Physics, Universitat Politècnica de Catalunya, 08034 Barcelona, Spain 3 Lab. SPMS, Ecole Centrale Paris, Grande voie des vignes 92295 Chatenay-Malabry Cedex, France Abstract ZrO2 and TiO2 modified lead-free (K0.5Na0.5)NbO3 (KNN) piezoelectric ceramics are prepared by conventional solid-state reaction. The effect of acceptor doping on structural and functional properties are investigated. A decrease in the Curie temperature and an increase in the dielectric constant values are observed when doping. More interestingly, an increase in the coercive field Ec and remanent polarization Pr is observed. The piezoelectric properties are greatly increased when doping with small concentrations dopants. ZrO2 doped ceramic exhibits good piezoelectric properties with piezoelectric coefficient d33=134 pC/N and electromechanical coupling factor kp=35%. It is verified that nonlinearity is significantly reduced. Thus, the creation of complex defects capable of pinning the domain wall motion is enhanced with doping, probably due to by the formation of oxygen vacancies. These results strongly suggest that compositional engineering using low concentrations of acceptor doping is a good means of improving the functional properties of KNN lead-free piezoceramic system. Keywords: Lead-free piezoceramics, Acceptor doping, Piezoelectric properties, (K,Na)NbO3 *Dr. Xavier Vendrell xavier.vendrel[email protected] Telf: +34 934 021 270 Fax: +34 934 907 725 2 1. Introduction The most widely used piezoelectric ceramics are Pb(Ti,Zr)O3 (PZT)-based materials, on account of their high piezoelectric response, large-scale production capability and the tailoring of their properties through composition. Due to the high toxicity of lead, a wide range of strict regulations concerning environmental preservation are increasingly being introduced worldwide. Many governments have therefore established legislation regarding waste electric equipment (WEEE), restrictions on hazardous substances (RoHS), and end-of-life vehicles (ELV) to introduce directives regarding environmental pollutants[1]. The search for alternative lead-free piezoelectric materials is currently focused on modified bismuth titanates, alkaline niobates (KNN) and other systems in which a morphotropic phase boundary (MPB) occurs[2–4]. Among the available lead-free piezoelectric materials under study, much attention has been paid over the last few years to K0.5Na0.5NbO3 (KNN)-based ceramics as a result of the breakthrough made by Saito et al.,[5] who obtained high d33 (~400pC/N) in the Li-Ta-Sb modified KNN. However, the major drawback of KNN ceramics is the need for special handling of the starting powders, sensitivity of properties to nonstoichiometry, and especially a complex densification process[6]. As regards PZT, sintering aids such CuO, SnO2, ZnO or MnO2 may improve sinterability and modify the dielectric and piezoelectric behaviour of the KNN materials[7–11]. Compositional modification by doping is a very active research line for obtaining piezoceramics with enhanced properties. The (K,Na)NbO3-LiTaO3-LiSbO3, particularly the composition (K0.44Na0.52Li0.04)(Nb0.86Ta0.10Sb0.04)O3, is probably the most workable lead-free piezoelectric system known to date. However, its properties are not suitable for all end use, e.g. for power devices where piezoceramics with low losses and stable properties are required. In this perspective, good results are expected by means of hardener substitutions, such as those that occur in other perovskites[12], although some structural and electrical aspects remain controversial as regards the role of dopants in the KNN system. It has recently been shown that Cu-doped KNN-modified compounds may exhibit typical characteristics of hard behaviour[13–16]. Hardener ions in a perovskite compound (e.g. Cu2+ 3 ions replace Nb5+ ions in Cu-doped KNN) are most often acceptors, the introduction of which leads to the creation of oxygen vacancies, thereby forming the so-called complex defects[17]. These defects operate as pinning centres by hampering the motion of the domain walls. This domain wall pinning effect is responsible for the reduction in dielectric losses and stabilization of properties in ferroelectrics[18]. The goal of this study is thus to test the influence of Zr4+ and Ti4+ acceptor-doping in (K0.5Na0.5)NbO3 on the structural, dielectric, piezoelectric and nonlinear properties. The addition of ZrO2 and TiO2 is expected to improve functional properties of KNN ceramics for power application, i.e. reduce the losses and increase the properties stability. 2. Experimental The (K0.5Na0.5)(Nb1-xMx)O3- compositions with x = 0.0 and 0.005 being M=Zr or Ti, hereafter abbreviated as KNN-M, were synthesized by conventional solid-state reaction. The raw materials of analytical grade used in this study were K2CO3 (99%), Na2CO3 (99.5%), Nb2O5 (99.9%), ZrO2 (99%) and TiO2 (99%). After separate milling, the powders were weighed and mixed by ball milling using ZrO2 balls in absolute ethanol medium for 3 h, then dried and calcined twice at 700 ºC for 2 h. The calcined powders were then milled again and cold-isostatically pressed at 750 MPa into 7 mm diameter pellets and sintered in air, without a binder, at 1125 ºC for 2 h. Care was taken to ensure that a high alkaline element pressure is maintained during the process by surrounding the pellets with powder of the same composition, and the pellets were deposited on Pt foils to avoid reaction with alumina boats. All as-sintered ceramics showed relative densities over 95% measured by Archimedes’ method. X-ray diffractograms were recorded at room temperature (RT) on a Bruker D2 PHASER equipped with a XFlash detector. A two-axis diffractometer in Bragg-Brentano geometry with Cu Kα1,2 radiation was used for the in situ XRD characterization. The control of temperature was provided by a furnace from 30 ºC to 500 ºC. The cell parameters and their evolution with temperature were refined by a LeBail fitting procedure as implemented in the Fullprof suite[19]. The Raman scattering spectra were obtained on a Labram (Horiba) spectrometer with a He-Ne excitation wavelength of 632.8 nm, 4 coupled to a Linkam sample holder, with variable temperature from 30 ºC to 500 ºC. DSC measurements were performed on a Seiko apparatus covering the 100 - 1000 K temperature range. Microstructure was evaluated on polished and thermally etched samples (1000 ºC for 5 min for pure KNN and 980 ºC for 5 min for doped KNN) using a Field Emission Scanning Electroc Microscope, FE-SEM (JEOL JSM-7001F). The micrographs were performed on the polished and thermally etched surfaces of the samples coated with graphite. The voltage conditions were set at 20 kV and the work distance was established at 12 mm. In order to analyze the composition of the material, an energy dispersive spectroscopy (EDS) analysis was performed using an OXFORD X-MaxN EDS detector provided with INCA Energy software. A pure Co sample was used as a reference material. The average grain size was determined from the SEM images using an image processing and analysis software (ImageJ). The grain size was expressed as Feret’s diameter using more than 400 grains in each measurement. Electrical characterization was carried out on ceramic discs with sputtered goldchromium electrodes on both parallel surfaces. The temperature dependence of the dielectric permittivity was measured on unpoled samples using an impedance analyzer (HP 4192A) in a frequency range of 100 Hz to 1 MHz and the temperature range of 30 ºC to 600 ºC. The measurements were done during the heating ramp at a rate of 2 ºC/min. The permittivity dependence with a sub-switching ac electric field was measured at 1 kHz and at RT, by means of a capacitance. The hysteresis cycles were recorded using a modified Sawyer-Tower circuit, at 1Hz and at RT. The samples were poled in silicon oil at 80 ºC for 30 minutes under a 30 kV/cm DC electric field. Subsequently, the longitudinal piezoelectric coefficient was measured using a piezo-d33 meter (YE2730A, APC International) at RT. The piezoelectric constant d31 and the electromechanical coupling factor kp were determined at RT by the resonance/antiresonance method on the basis of the IEEE standards. 3. Results and discussion Figure 1(a) shows the XRD patterns of the pure KNN and KNN-M ceramics measured at RT. For all ceramics, a pure perovskite phase indexed in the Amm2 orthorhombic space group is observed. In contrast to the results obtained by Ramajo et al. [20], no parasitic phases are observed at low 5 concentration of dopant. Our results suggest that taking special care of the reagents and of the synthesis method plays an important role to avoid the formation of parasitic phases. The cell parameters vary by less than 0.5 %, indicating that the structure does not change significantly upon doping with ZrO2 or TiO2. Since the Ti4+ and Zr4+ ionic radii, 0.61 Å and 0.72 Å, respectively, in a six-fold environment[21], are close to that of Nb5+ ion (0.64 Å), dopant atoms should essentially locate on the B-site, which by charge compensation gives rise to the creation of oxygen vacancies. The Raman spectra of the sintered samples (Figure 1(b)) present the typical vibrations corresponding to a perovskite phase, associated with the BO6 octahedra[22]. No extra signals are detected; thus, no parasitic phases are observed, as confirmed by XRD. The vibrations of the BO6, A1g (ν1) and F2g (ν5) are relatively strong scatterings in systems similar to KNN, because of a near-perfect equilateral octahedral symmetry. Furthermore, the stretching modes (ν1, ν5) shift to lower frequencies. This effect may be attributed to a weakening of the bond strength, probably caused by the oxygen vacancies. Figure 1: XRD patterns (a) and Raman spectra (b) of the pure KNN and KNN-M ceramics measured at room temperature. The transition temperatures from orthorhombic to tetragonal phase and from tetragonal to cubic phase are observed at around 200 and 420 ºC, respectively, in pure KNN ceramics [23]. These transitions are confirmed here by different techniques. The analysis of temperature-dependent Raman spectra, XRD patterns and DSC data allows the two transitions for KNN and KNN-M ceramics to be 6 identified without ambiguity, as shown in Figure 2. All techniques show practically the same transitions temperatures, although the DSC data show lower transition temperatures due to the dynamic acquisition process. When adding 0.5 % of Ti4+ and Zr4+-ions on KNN B-site, we observe a decrease of the cubic-to-tetragonal transition temperature of 45 and 25 K, respectively, which is similar to what was previously found in acceptor-doped KNN [24–26]. The tetragonal to orthorhombic temperature transitions decrease only slightly, this effect being higher when doping with TiO2. Figure 2: Temperature-composition evolution of the pure KNN and KNN-M ceramics, based on DSC, Raman spectroscopy and XRD measured from 30ºC to 500ºC. The variation of the microstructures and the grain size distributions of the KNN and KNN-M ceramics is shown in Figure 3 (a)-(f). All the samples exhibit the typical morphology of the alkaline niobate ceramics. Although the pure KNN ceramic micrograph is over-etched, the surface reveals a bimodal microstructure with cube-shape grains ranging from about 500 nm to a few micrometers, with an average grain size of ~3.9 ± 2.3 μm (Figure 3(a,d)). The micrograph and the grain size distributions of the pure KNN also show the presence of some large abnormal grains. When doping with Zr4+ or Ti4+ ceramics show a decrease of the grain size and an increase of the grain size distribution uniformity, Figures 3(b,e) and (c,f). The average grain size of the KNN-Zr ceramics is ~2.1 ± 1.2 μm, and for the KNN-Ti ceramics is slightly lower ~1.7 ± 1.1 μm. As can be observed in Figures 3(d,e,f) the grain size distributions shifts toward smaller grain sizes when doping and causes the grain size distribution 7 to become much narrower. Therefore, the addition of acceptor dopants inhibits both the grain growth and the formation of abnormally grown grains, as already observed [20,27]. The limited grain growth observed when doping could be related with a non-uniform distribution of the Zr4+ or Ti4+ dopants between the grain and the grain boundary. Higher concentration of the dopant in the grain boundary region could hinder the grain growth, as observed for doped BaTiO3 ceramics [28,29]. The TEM studies presented by Malic et al. confirmed the presence of ZrO2 inclusions boundary region [30] that could avoid the grain growth. In our case, there is also a decrease in the grain growth when doping, although the presence of ZrO2 inclusions could not be observed in our case. Probably, it can be attributed to the different synthesis method used by Malic et al. were the dopants are added in the powder mixture after the solid state synthesis, thus, in our case the dopants are incorporated in the lattice instead of forming inclusions in the boundary region. Figure 3: Microstructure and grain size distributions of polished and thermally etched surfaces of pure KNN (a,d), KNN-Zr (b,e) and KNN-Ti (c,f). 8 No evidence of secondary phases was observed by BSE or EDS analysis, as confirmed by XRD and Raman spectroscopy. The EDS analysis of the matrix reveals that the atomic percentages of elements do not differ significantly from the nominal composition, as reported in Table I. It should be noted that Zr and Ti elements could not be detected due to the low content of dopants. Finally, the matrix grains present a Na/K ratio around 1, which is close to the nominal composition, confirming that the low synthesis temperature and the special care taken when sintering avoid the evaporation of Na or K elements. Figure 4: Temperature dependence of the real (a) and imaginary (b) parts of the relative permittivity of pure KNN and KNN-M ceramics sintered at 1125 ºC for 2 h (at 10 kHz). Table I: Elemental composition of KNN and KNN-M ceramics determined by EDS analysis. This table represents the atomic percentages of elements. Na K Nb M Na/K KNN 10.18±0.32 10.24±0.41 20.44±0.27 - 0.99 KNN-Ti 10.04±0.18 9.98±0.34 20.12±0.16 - 1.01 KNN-Zr 10.09±0.34 10.18±0.27 20.34±0.21 - 0.99 KNN Nominal Composition 10 10 20 - 1 9 Figure 4 shows the temperature dependence of the real and imaginary parts of the relative permittivity of pure KNN and KNN-M ceramics, measured at 10 kHz from RT to 450 ºC. Permittivity versus temperature shows two anomalies, one at around 200 ºC associated with the orthorhombic to tetragonal phase transition, and the other at higher temperatures, at which a clear maximum of the permittivity is shown for all ceramics and is associated with the tetragonal ferroelectric to cubic paraelectric phase transition. In doped samples, the orthorhombic-to-tetragonal (O-T) phase transition shifts slightly toward lower temperatures, while the decrease in temperature of the tetragonal-to-cubic (T-C) phase transition is much more significant. The temperature change of O-T and T-C phase transitions for TiO2 or ZrO2 doped samples may also be attributed to the acceptor doping effect [24,30]. As already shown in Figure 4, for pure KNN a typical normal ferroelectric to paraelectric phase transitions is observed showing a narrow phase transition. Meanwhile, the ferroelectric to paraelectric phase transitions broadens on doping, suggesting the appearance of diffuse phase transition. This behaviour may be induced in many ways, such as by microscopic compositional fluctuation, by the merging of micro-domains into macro-domains, or via a coupling of the order parameter and local disorder mode through local strain [31]. The diffuse phase transition is observed for both TiO2 and ZrO2 doped KNN ceramics. Thus, one probable cause of this behaviour is local fluctuations induced by the incorporation of Ti4+ or Zr4+ ions, whose valence is different from the Nb5+ one, into the crystalline lattice of the perovskite. Figure 5: P-E hysteresis loops of the pure KNN and KNN-M ceramics measured at room temperature and at 1 Hz.