- Previous research 1

Alternative arrangement of plus and minus terminals enable magnetic flux to be cancelled, resulting in low inductance. This concept can be shown in stranded cables.

The problem is DC electrical degradation. DC voltage is applied to the decoupling capacitor for the longest time; therefore, oxygen vacancies can be piled up at the interface between the perovskite-type dielectric and the reverse-biased electrode, resulting in the increase of leakage current.

This slide shows the plots of current density versus mean applied electric field across the (Ba_0.5Sr_0.5)TiO₃ thin film. The bias voltage was applied to a top electrode: so in the positive electric field region in the slide, the bottom electrode was reverse-biased. Electrons are injected from the bottom electrode to the (Ba_0.5Sr_0.5)TiO₃ film. At the bottom-side interface, an oxygen-depleted layer was intentionally inserted. The insertion of oxygen-depleted layer caused the increase of leakage current. When the thickness of oxygen depleted layer exceeded 20 nm, the leakage current became independent of the thickness of oxygen depleted layer, and became independent of temperature. The experimental evidence indicate that the electron injection mechanism changed from Schottky injection to tunneling injection, possibly defect-assisted tunneling.

Besides decoupling applications, there are other applications of thin film capacitor. Tunable capacitor is one of them.

Tunable capacitors are intended for use in RF tunable applications. This slide shows the plots of relative dielectric constant versus applied voltage. As shown in this slide, the voltages at which the relative dielectric constants have their maximum values were shifted toward the positive-bias region. The shift is increased with increasing the thickness of the oxygen-depleted layer in the range of oxygen-depleted layer thickness of 0 nm through 20 nm. This can be attributed to positively charged oxygen vacancies near the bottom electrodes. However the thickness of the oxygen-depleted layer exceeds 20 nm, the peak shift is saturated.

By the way, it is assumed that the voltage shift in the sample without intentionally inserted oxygen depleted layer is due to the native positively charged defects.

The problem on RF application is, a lag time for tuned capacitance to be settled after applying the bias. It takes considerable time for capacitance to be tuned.

Before explaining it, I should talk about the Schottky effect dependence of capacitance. Unless it is a perfect insulator, a dielectric with many defects should be considered as a semiconductor. Therefore, a depletion layer is inevitably formed. At a defect density of 10¹¹-10¹² cm^-2, the contributions of the space charges in the depletion layer and the dielectric charges of (Ba,Sr)TiO₃ are comparable. Under such a condition, the built-in voltage tunes capacitance.

Lag time results from the narrow depletion accompanied with the electron detrapping from the deep level.

This is a schematic which explain the narrow depletion accompanied with the electron detrapping from deep defect levels near the interface between (Ba,Sr)TiO₃ and reverse-biased electrode. There are deep-level donors in a (Ba,Sr)TiO₃ film with oxygen vacancies. The deep-level donors are ionized as a result of the lowering of quasi-Fermi level. Therefore, the density of the ionized donors increases and the depletion layer becomes narrower. Note that the depletion layer of some conventional semiconductors with a low purity also becomes narrower owing to the ionization of the deep-level donors.

Such a detrapping from deep levels causes a several seconds of lag time for tuned capacitance to be settled after applying the bias. It can be a disaster on RF applications.

Powerpoint file. Link