Dielectric and ferroelectric properties of perovskite relaxors.

Abstract

The electrical properties of perovskite relaxors, represented by lead magnesium niobate (PbMg$\sb{1/3}$Nb$\sb{2/3}$O$\sb3$, PMN), are investigated in this thesis. Taking the domain wall movement as a common thread, the thesis focuses on the effects of compositional modification on the domain wall dynamics and hence the electrical behavior of the materials, by referring to the similarities and the differences between relaxors of different compositions, and between relaxors and prototypical perovskite ferroelectrics such as barium titanate. Unlike typical ferroelectrics, relaxors have no permanent polarization at or below the temperature where the permittivity reaches a maximum $(T\sb{max}).$ Our results show that the depolarization temperature and $T\sb{max}$ are correlated, and both depolarization and coercivity of PMN and its modifications can be understood in terms of thermally activated movement of domain walls. A systematic variation of coercivity and depolarization behavior in relaxor materials with their compositions has been addressed. Barkhausen pulses in charging/discharging experiments at a weak field have been taken as further evidence of domain wall activity. Our experimental observations found significant differences in the characteristics of Barkhausen pulses between PMN and BaTiO$\sb3.$ These differences are believed to be due to the different internal fields and domain wall contours in the two materials. Domain wall movement in relaxors is found to be sensitive to the structural imperfections and defects, in particular oxygen vacancies. This sensitivity is demonstrated for Ti-modified PZN single crystals after aged in different atmospheres. The effects of compositional modification on dielectric dispersion have been studied in different Pb(B$\sp\prime\sb{1/3}$Nb$\sb{2/3}$)O$\sb3$ materials modified by A- and B-site substitutions and the results are analyzed using the Vogel-Fulcher equation. We interpret the rate equation as due to the mobility of domain walls which is controlled by kink formation and kink migration, with the sum of kink formation energy and kink migration energy being the activation energy. This result seems to be consistent with our experimental observations, as well as those in the literature.