Abstract

Ferroelectric materials are very attractive for many technological applications. In the scenario of conventional CMOS technology, they are very suitable to fabricate nonvolatile memories (FeRAM) or as the gate stack of a transistor to realize a FeFET. Conventional ferroelectric materials, like PZT or SBT, exhibit very high permittivity and low coercive field and very thick films must be grown to obtain wide enough memory windows. Because of that, new ferroelectric materials must be investigated.Applications compatible with CMOS technology require very thin layers of dielectrics. Moreover, it is difficult to obtain new insulating materials with low conductivity in the early stages of their technological development. Therefore, ferroelectric properties can be masked by high leakage currents. This can result in the abandonment of these materials before the growing conditions are optimized. In addition, the conventional measurements of ferroelectric properties use techniques in which the electric field is varied with relatively high frequency signals that result in displacement currents that represent an additional term which is added to leakage currents and further conceals ferroelectric behavior. Traditional Sawyer-Tower techniques do not allow the detection of ferroelectricity when leakage and/or displacement currents are excessively important and even experimental values of polarization may be overestimated when affected by charge terms corresponding to the parasitic phenomena mentioned above.Basically, the double swing quiescent-current (DSQC) consists on measuring the current of the sample while varying the applied voltage as follows: a) To obtain the positive remanent polarization, we apply two identical voltage sweeps from zero to values high enough to polarize the sample. The swing rate is chosen slow enough to consider the measure process as a succession of quasi-stationary states. On the first sweep the current has three contributions: leakage, displacement and polarization itself. Since the material has already been polarized in the first sweep, the component due to ferroelectric polarization is no present in the second one. The difference between the curves of the first and second sweeps provide the curve due to the polarization current. b) Once the material is positively polarized, we repeat the same process but applying negative voltage sweeps to negatively polarize the material.To check the validity of the technique, we have performed previous tests with conventional ferroelectric materials. In Figures 1 and 2 we represent the current-voltage and charge-voltage characteristics obtained. We verify that the obtained results match the specifications for these materials.Afterwards, we have applied the DSQC method to characterize the ferroelectric behavior of amorphous SiO2-Nb2O5 nanolaminates and mixture films. The films were grown by atomic layer deposition. The films were grown at 300 oC from Nb(OC2H5)5, Si2(NHC2H5)6, and O3 to thicknesses ranging from 13 to 130 nm. The niobium to silicon atomic ratio was varied in the range of 0.11-7.20.In Figure 3 we observe a complete voltage current cycle. Starting from 0 V, the voltage increases linearly to a value of 12 volts and then decreases to -12 volts. Finally, the voltage returns to zero. A swing rate of 5 mV/s has been used to obtain this curve. In this figure you can observe the two peaks corresponding to the positive and negative polarization. It is also noted that the current is not zero when the sample is already polarized due to the existence of leakage and displacement currents.To remove these components two sweeps must be performed in each direction. In Figures 4 and 5 we present the result of applying double sweep for positive and negative voltages, respectively. The difference between the two sweeps provides the ferroelectric polarization current. We note that this current returns to zero once the sample has been polarized.In summary, we demonstrate that double swing quiescent-current technique allows to detect ferroelectric behavior even in highly conductive materials and obtain reliable values of the remaining polarization that are not affected by the parasitic effects of leakage and displacement currents. Figure 1

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call