Strontium Bismuth Titanate is a very promising material for high temperature piezoelectric applications, its elevated ferroelectric phase transition (530°C), linear piezoelectric properties under low field and relatively low room temperature conductivity (compared to others Bismuth Titanates) make it very attractive for precision sensors. However, under severe conditions (low frequency, high field, high temperature or low oxygen partial pressure) some of those advantages disappear. Piezoelectric response is dominated by charge drift in general becomes unstable. Above all, at high temperature and low oxygen partial pressure, a large conductivity increase reduces the piezoelectric efficiency of the material, in this work, electrical conductivity, piezoelectric properties and dielectric permittivity of SrBIT ceramic have been investigated in conditions of high temperature, low oxygen partial pressure, low to moderate driving field and frequency. As a result of this research, a better understanding of SrBIT properties was achieved. Thanks to a careful study of SrBIT processing as a bulk ceramic, a reproducible route was established. Many basic mechanisms leading to both SrBIT crystallization and sintering have been identified. It was demonstrated that a detailed knowledge of the exact processing conditions is required in order to achieve high quality material. DC conductivity measurements were carried out as a function of temperature, oxygen partial pressure and dopant concentration. It was found that the apparent activation energy for conduction for undoped SrBIT was 1 eV between 140 and 220°C and 1.5 eV between 450 and 700°C. Decrease of the activation energy in the lower temperature range has been discussed considering grain boundary conductivity, as a transition from electronic to ionic conduction or as consequence of small polarons conduction. It was shown that lower activation energy resulted from Manganese doping (0.5 eV), this was interpreted as either growing influence of grain boundary conductivity as dopant concentration increases or as shallow hole trap formation. DC conductivity measurements and acceptor/donor doping experiments demonstrated p-type conductivity in the low temperature range (up to 220°C) as donor doping decreases conductivity, while oxygen partial pressure controlled measurements indicated n-type conductivity at higher temperature (above 700°C). An electrical impedance analysis was performed with several equivalent circuits. The aim of these models was to simulate the impedance of SrBIT. The best approximation was found with a distributed element of Havriliak-Negami type. However, as the physical justification for this circuit was not clear, the investigation of the grain, grain boundary and electrode impedances was performed with multiple discrete parallel RC elements. With temperature, grain size and oxygen activity variations, the identification of three separate contributions as grain, grain boundary and electrode was realized. The anisotropy of conductivity and permittivity was demonstrated with textured material and both DC and AC analysis. With the master curves built for the electrical modulus, it was found that the impedance probably related to Bismuth oxide layers produces an additional high frequency are. From this finding and by comparing characteristic relaxation frequencies of undoped, 2 mol.% Mn and 4 mol.% Nb SrBIT, it was determined that conductivity is higher in the ab plane direction than in the c direction within both perovskite units and Bismuth oxide layers. With conductivity measurements under controlled oxygen partial pressure, it was found that an acceptor-based (intrinsic or extrinsic) model could be used to describe the electrical conductivity under controlled oxygen partial pressure of both undoped and 2 mol.% Mn doped SrBIT. However, as neither a pO2-1/6 region (intrinsic oxygen vacancies compensated by electrons) for undoped SrBIT nor a pO2-1/4 region (oxygen vacancies compensated by singly ionized acceptors) for Mn doped SrBIT were seen, it was concluded that the acceptorcontrolled model is not sufficient for a complete description of SrBIT. For this reason and in order to include the low oxygen partial pressure behavior of undoped SrBIT, a donor-based (intrinsic) model was also considered. The source of intrinsic donors would be in that case Bi3+ cations sitting on Sr2+ sites in the perovskite sublattice. Considering Bismuth vacancies as the negative compensating species, both pO2-1/4 and pO2-independent regions could be predicted with the model. However, even if the donor-controlled model seems to better match conductivity measurements in the full PO2) range, rejecting the acceptor-based model would be an error. It is actually not demonstrated that in SrBIT the concentration of exchanged Bismuth cations is always (all temperature, pressure) larger than the natural acceptor impurity concentration. It is very likely that the cation exchange is dependent of the oxygen partial pressure. It is also not proved as suggested in the literature that direct compensation between exchanged Strontium and Bismuth exists, reducing the net donor-excess. With the acceptor-controlled model, the mass-action constants for reduction and for ionization of intrinsic carriers across the band gap were determined. The band gap of SrBIT was estimated to be 3.5 eV. The ionic conductivity of SrBIT was determined at high temperature with measurements under controlled oxygen partial pressure. It was found that the electrical conductivity of SrBIT is probably mixed (electronic and ionic) as the estimation of the transference number provided quite large values (t=0.8 at 800°C). From electronic and ionic conductivity data, mixed conduction can actually be predicted in a large temperature range (above room temperature). The piezoelectric measurements using direct effect demonstrated that it is actually possible to unlock piezoelectrically active ferroelectric domain walls and create non-linear piezoelectric properties in undoped SrBIT. This occurs above a threshold elastic field, which is thermally activated. With a piezoelectric composite, it was demonstrated that the electromechanical coupling between two different phases creates a piezoelectric relaxation. This one could be positive or negative depending on the respective properties of each composite's component. It was shown experimentally that a small temperature change is sufficient to transform a positive relaxation into a negative one. While these experiments did not provide a detailed microstructural explanation for the piezoelectric relaxation observed in 2 mol.% Mn doped SrBIT, they gave a first insight into an original phenomenological approach. Microstructure and piezoelectric properties were related with the calculation of a piezoelectric relaxation composite made of two textured samples. This demonstrated that a piezoelectric relaxation may occur just because of anisotropy. Microstructural reproduction of this coupling is actually an important feature of Aurivillius phases.
EPFL2000
