We report on a measurement procedure to separate ferroelectric switching current and dielectric displacement current from the leakage current in leaky ferroelectric thin-film capacitor structures. The ac current response is determined for two adjacent frequencies. Taking advantage of the different frequency dependencies of the ferroelectric switching current, dielectric displacement current and ohmic current, the hysteresis loop is calculated without performing a static leakage current measurement, which causes a high dc field stress to the sample. 1 Key material parameters for memory applications are the remnant polarization P R and the coercive voltage V C or the coercive field E C . Since the remnant polarization directly correlates with the amount of charge that can be detected during the read operation of the memory cell, high values are desired for memory applications. The coercive voltage is closely linked with the operating voltage of the storage element. In an ideal system, V C determines the voltage that is needed to switch the polarization inside the ferroelectric thin film. In real materials, the operating voltage might be twice the coercive voltage, since a voltage higher than V C is required to switch the polarization completely and a low interface layer may reduce the field inside the ferroelectric layer.2 Both values can be obtained from the hysteresis curve ͑P -V characteristic͒ of the ferroelectric capacitor. For a loss-free ferroelectric capacitor, the hysteresis loop can be acquired by an integration over the displacement current passing through the capacitor. In contrast, real ferroelectric capacitors show a leakage current in addition that superposes the displacement current and that can affect the shape of the hysteresis curve.3 An integration of the monitored current may then result in a nonphysical increase of remnant polarization and coercive field. Graphical 4 as well as numerical methods 5 have been reported to eliminate the leakage contribution from the current response. A common method is to monitor and to withdraw the static leakage current from the dynamic current measurement.5 Here, the voltage is changed stepwise and the dc current is recorded. This measurement procedure has to be repeated for each data point, whereby the duration of a voltage step mainly depends on the relaxation time of the transient signal response and on the required resolution of the monitored current. A typical duration for a voltage step is in the order of seconds. Depending on the number of data points, several minutes are required to measure the dc I -V characteristic. In some cases, the sample cannot withstand the continuous dc field stress, and the characteristic of the capacitor will change during the static leakage current measurement, e.g., due to resistance degradation. Since this tendency increases with reduced sample thickness, the characterization of ultrathin ferroelectric films is a tough business. In view of the continuous trend to miniaturization, an increase of leakage current originating ...

Direct hysteresis measurements on single submicron structure sizes were performed on epitaxial ferroelectric Pb(Zr,Ti)O3 thin films grown on SrTiO3 with La0.5Sr0.5CoO3 (LSCO) electrodes. The samples were fabricated by focused-ion-beam milling resulting in pad sizes down to 200 nm×200 nm. The influence of parasitic capacitance of the measurement setup was eliminated by applying an enhanced compensation procedure. No size effects were observed in capacitors milled down to 400 nm×400 nm. Thus, a published increase of Pmax1 for decreasing pad size can be explained by the parasitic influence of the setup. Finally, the inaccuracy of increasing coercive voltage due to the coating of the cantilever of the atomic force microscope is discussed.

No abstract

No abstract

A two-dimensional numerical model that predicts the reliability of multilayer capacitors (MLCs) during soldering and bending is presented. The Weibull parameters used in the model are based on measurements of soldered MLC devices. The preheating and soldering temperatures have a dominant impact on the failure probability, in comparison to the thickness of the nickel layer, the soft solder geometry, and the number of inner electrodes. Comparison of calculated and measured reliability of three MLC sizes leads to the assumption that residual stresses due to the manufacturing process or size-related microstructure are important.