A review is given of ceramic and single-crystal thin film ferroelectric oxides, emphasizing perovskite phases, together with some new developments on hafnia films. It is shown that singlecrystal barium titanate films behave as bulk down to at least 77 nm, with no finite size effects, no phase transition temperature shifts, and no dielectric peak broadening or change from first-to second-order transitions, suggesting that the gradient defect model of Bratkovsky and Levanyuk correctly describes such effects as extrinsic in experimental studies of equally thin ceramic thin films. In ceramic barium-strontium titanate (BST) thin films, it is shown that there is also no intrinsic broadening or shifts in phase transitions, with sharp, unshifted, bulk-like transitions observed only as re-entrant upon warming from cryogenic temperatures; this shows that phase transitions in ceramic thin films are dominated by kinetics and not thermodynamics and are definitely not equilibrium measurements. At high fields (41 GV/m), the films exhibit space charge-limited conduction; no variable-range hopping is observed, contrary to recent studies on SrTiO 3 . Some novel, unconventional switching processes are discussed, comparing the ''perimeter effect'' (non-equilibrium, ballistic) with Molotskii's equilibrium model. Theory and experiment are described for [3D] nanotubes, nanorods, and nanoribbons (or micro-ribbons). The layered-structure-perovskitepyrochlore conversion in bismuth titanate is described together with the PbO1TiO 2 phase separation in lead zirconate titanate during electrical breakdown, as are novel HfO 2 precursors that demonstrate enhanced temperature crystallization from the amorphous state and hence commercial advantages for frontend processing.
II. [2D] SystemsIn planar structures, both fine-grained ceramics and single crystals offer intriguing materials science in the 4-100 nm thickness regime. Several authors 2-4 have indicated that thin films might exhibit second-order phase transitions that, in bulk form, are first order; then, the question is exactly what thicknesses constitute ''thin''? Although the International FRAM ''Roadmap'' suggests 5 that FRAM structures must become [3D] by 2007 to accommodate requisite capacitance, at present, all FRAMs are planar-stacked devices. Therefore, we begin with the [2D] systems' materials science.
J ournal