Electrophysical Properties of Structures Based on Ferroelectric Thin Films
Improving characteristics of modern micro- and nanoelectronic devices and expanding their functionality are based on the use of new materials. An example of promising research in this area is the work on composing structures based on ferroelectric (FE) thin films that are used in non-volatile memory microcircuits, infrared uncooled arrays in the focal plane, and ultrahigh-frequency circuits [1, 2].
The reliability of devices using heterostructures is directly related to the existence of transition layers with defects. As a rule, the presence of a large number of surface states leads to undesirable effects when designing microelectronic devices. In this paper, the analysis of effects in transition layers of heterostructures with ferroelectric films was based on measurements of their voltage-capacitance characteristics.
THE RESEARCH METHODOLOGY
In order to study electrophysics of FE films, the MIREA has developed a complex of measuring instruments that make it possible to measure temperatures ranging from 10 to 200 °C . The complex includes both standard measuring tools and specialized data acquisition and processing boards integrated within the LabVIEW frame. The complex, in particular, allows one to:
• determine the pyroelectric coefficient value by low-frequency temperature wave method (the modulation period for temperature is set within 0.1–50s, the amplitude is set within 1–5 °C, the current sensitivity limit is 10−12A, the relative error of pyroelectric coefficient measurements is not more than 20 %);
• detect ferroelectric hysteresis with extraction of the polarization switching charge by subtracting the charge of the differential capacita nce of the “empty” non-ferroelectric capacitor (voltage range: –200…+200V, scanning frequency: 0.01–10kHz, reference capacitance: 10.100nF, polarization measurement error: less than 10 %);
• determine the residual polarization value by means of switching currents when a sample is exposed to a sequence of two pairs of hetero-polar voltage pulses (switching voltage range: –15…+15V, pulse leading-edge time: not more than 0.1µs, maximum frequency: 500kHz, polarization measurement error: not more than 30 %);
• measure dielectric characteristics of films with arbitrarily programmable forms of the voltage scanning applied to a sample (scanning voltage range: up to 200V, 100mV amplitude test signal at frequencies: 0.1, 1, 10 and 1000kHz, error of capacitance and dielectric loss tangent measurements is not more than 0.1 %);
• measure static leakage currents (current sensitivity: 10−14A, measurement error: not more than 20 %) in the same voltage range.
The equipment complex is analogous to RT 66A (manufactured by Radiant Technologies, USA). However, the complex for studying electrophysics of ferroelectric-based structures developed by MIREA is much cheaper and in terms of functionality and technical characteristics is on a par with the US analog.
THE RESULTS AND DISCUSSION
Structures based on lead zirconate-titanate films
Samples of multilayer structures based on lead zirconate-titanate (PZT) films were of the following type: silicon substrate SiO2 (0.5µm) — Ti (0.01µm) — Pt (0.1µm) — PZT (0.52/0.48) — Pt (0.1µm). The samples were prepared by chemical precipitation from a mixture of solutions . The excess amount of lead in the initial components varied from 0 to 50 %, with subsequent heat treatment varying from 500 to 750 °C. The thickness of PZT films depended on the number of cycles of depositing solution with subsequent drying and reached values of 0.2–0.3µm (for 5.6 cycles). Typical dimensions of electrodes were 100×100µm2.
In particular, for the Pt/PbZr0.53Ti0.47O3/Pt/Ti/SiO2/Si type samples, a significant asymmetry in voltage-capacitance characteristics and ferroelectric hysteresis loops for positive and negative scanning voltages has been observed.
A FE structure with electrodes can be presented as a daisy chain consisting of the FE capacitance itself and two successive capacitances of the upper and lower transition layers at the dielectric-metal boundaries.
Upon switching polarization, the voltage-capacitance characteristic is determined by the change of capacitance at the transition layer displaced backward, and the values of voltage drops at the transition layers can be calculated by means of the 1/С2 dependence on the applied voltage; besides, the density of surface states values in transition layers can be estimated. In this research we have been using a model where transition layers are the metal-dielectric structure layers with the Schottky barrier junction. In this case, the following relation  is valid:
where S is the electrode surface area; Nsc is the density of surface states (measured in 1/m); V is the applied voltage; q is the electron charge; ε is the relative permittivity. The parameters of this model were determined by a method based on linear approximation of voltage-capacitance characteristics of the investigated structures for large values of the applied voltage (Fig. 1).
The study has shown that voltage drops at the transition layers strongly depend both on technological modes of preparing layered structures and the excessive lead content in initial solutions, and can vary from tenths of a volt to several volts (–1.99V and +1.25V as shown in the figure).
The FE hysteresis loops recorded by voltage drops at the measuring capacitor or resistance contain both the switching part of the charge and the part attributable to charging the “empty” non-ferroelectric capacitor.
This fact is illustrated by Fig. 2. In this figure, along with the voltage-capacitance feature typical for PZT films, the markers depict the results of differentiating the hysteresis curves for asymmetric scanning of the applied voltage at the starting points of the return stroke, that is, when the charge changes at the measuring capacitor do not contain the switching component.
The method that we have proposed for processing the results of hysteresis measurements is based on an appropriate correction of hysteresis curves obtained from voltage-capacitance measurements, making it possible to refine data as regards such parameters of FE-based structures as residual polarization, saturation polarization, and the coercive field. The method is described in detail in .
The data obtained from voltage-capacitance characteristics and the hysteresis in PZT-based films samples have served as a basis for testing technological modes of their formation. In particular, it has been shown that introducing 10mol% of excess lead compared to stoichiometry into an initial film-forming solution ensures compensation of lead losses during crystallization, while the optimal crystallization temperature for the films was within the 600–650 °C range.
Further studies of PZT films samples by means of electron microscopy, X-ray spectral analysis, and the second harmonic optical microscopy have confirmed our results and shown that deviations from the optimal formation modes cause crystallization in the films of a significant amount of the pyrochlore metastable phase without FE properties.
Structures with polycrystalline FE films included in any measuring circuit are featured by the existence of smooth sections of the loop in the area where the polarization switches, i.e. by stable states with partially switched spontaneous polarization. As can be seen from Fig. 3, the result of integrating the current loop for a typical PZT film sample reveals a larger width of the curve compared to the data from direct measurements of the polarization dependence on voltage.
The current and charge hysteresis curves have been compared, and as a result, estimates for conductivity of the structure have been obtained that in typical cases give the values 1.2 • 10−4Ω−1 from the well-known relation for the current running through a measuring resistor:
where C and G are, respectively, nonlinear capacitance and conductivity of the structure; V is the applied voltage as a result of comparison of the current and charge of the hysteresis loops of the obtained estimate for the conductivity of the structure, giving in the typical cases, values of 1.2 • 10−4Ω−1.
The mechanisms of nonlinear conductivity in FE-based structures can be explained by the polycrystalline state of the films and the existence of depolarizing fields there. Direct measurements of dielectric losses and current-voltage curves for PZT films for low applied voltages (up to 1V) have confirmed the results.
Measurements of pyroelectric characteristics of differently composed films have been carried out both by means of the static method and the more efficient method of the low-frequency temperature modulation. In the latter case, the sample temperature periodically varied in a sinusoidal way with a certain amplitude T0 and frequency f. The corresponding change in charge at the surface of the pyroelectric capacitor has been detected by a short-circuit current.
After recording the temperature change and the corresponding change in the short-circuit current, the pyroelectric coefficient was derived from the amplitude I0 of the current, the temperature amplitude T0, and the phase shift ϕ by the following relation:
where ϕ is the phase shift determined by the polarization sign of a sample and the ratio of the amplitudes of the pyroelectric and the thermally stimulated currents; A is the area of electrodes of the pyroelectric capacitor.
The data on pyroelectric properties of the investigated samples of PZT-based films are summarized in Table 1.
The research’s special feature was experimental verification of the relationship between the polarization state of FE films and the pyroelectric coefficient (pyroelectric hysteresis). According to a thermodynamic model, this relationship must be linear up to higher order coefficients. In order to determine parameters of the model, a sample was preliminarily brought to a given initial state of polarization before each measurement of pyroelectric current, and then was repolarized by several voltage pulses with variable amplitude. The values obtained for the pyroelectric coefficients were compared with the results of measurements of residual polarization. The results are summarized in Fig. 4.
It should be noted that, in contrast to matrix thermal microbolometer-based detectors, pyroelectric detectors do not require subtraction of background and retain their characteristics when subjected to electromagnetic waves and radiation, while dependence of the pyroelectric effect on the degree of polarization can be used to create detectors with tunable sensitivity. In the long term one can foresee devices of the new generation that would have the ability to adjust to external conditions and perform analog processing of signals, in particular, automatic correction of irregular sensitivity.
The model for a multi-element thermal detector of radiation proposed by the authors is a matrix of ferroelectric detectors integrated with an electronic circuit for reading output of signals where series of analog correction of irregular sensitivity are introduced in the design (see Fig. 5).
The correction is performed by iterative varying of the transformation factor of each detector by the applied polarization voltage, so that the signals from all matrix pixels take on the same value when calibrated. The detectors’ sensitivities are adjusted with the accuracy determined by the range of the polarization voltage variation and the number of calibration cycles. In particular, for polarization voltages within the 8.10V range and 100 calibration cycles, irregularity of sensitivity can be reduced to 0.2 %. As a result, the signal-to-spatial noise ratio takes on a value which is sufficient to solve the problems of detecting objects in the infrared range without using additional circuits for digital signal processing which reduces energy consumption as well as the weight and size of thermal-imaging systems.
Heterostructures with high-temperature superconducting and ferroelectric films
In order to obtain c-axis oriented YBa2Cu3O7–x (YBCO) epitaxial films, the pulsed laser (Nd: YAG, 335nm, 10Hz) beam was applied to deposit on a single-crystal SrTiO3 (001) substrate. The calculated thickness of the YBCO film was 400nm. 500nm thick films of PbTiO3 (PTO) above the YBCO epitaxial layer were prepared in the same way .
The qualitative composition of heterostructures obtained in this way was studied by means of the energy-dispersive X-ray spectroscopy (EDX) methods. The Oxford X-Max Oxford Instruments NanoAnalysis (USA) energy dispersive attachment installed on the Helios NanoLab 450F1 FEI Company (the Netherlands) two-beam system was used. In order to generate the characteristic X-ray radiation, the microscope 10–20kV electron beam was used. The energy-dispersive spectrum of the analysis is shown in Fig. 6.
Measurements of thicknesses for the layers of the studied structure were carried out across vertical cross-sections obtained by means of the fine-focused ion beam method with the 30kV Ga+ ion beam in the Helios NanoLab 450F1 installation, and the surface morphology quality was estimated by means of the high resolution secondary electrons SEM image. A typical example of the (PTO+YBCO)/STO structure cross-section is shown in Fig. 7.
X-ray patterns of the samples were obtained by means of the DRON-4 automated X-ray diffraction meter (CoKα radiation); the measurements were taken at basal planes of the films, in the Θ–2Θ geometry. The observed reflections of the diffraction patterns for PT-YBCO films are shown in Fig. 8.
The analysis of the obtained data shows that diffraction patterns correspond to the PbTiO3+YBa2Cu3Oy, (PTO+YBCO) phase composition as well as SrTiO3 (STO) used as a material for the crystal substrate. Only axial reflections of the (001) type from PTO, (001) type from YBCO, and (h00) type from STO appear, which indicates that the films are oriented with their (001) planes along the basal surface and the crystal substrate surface is oriented along the (100) crystallographic plane. The parameters of the PTO, YBCO and STO elementary cells calculated from the positions of 2Θ x-ray reflections are consistent with reference data . Comparatively small widths of x-ray reflections (about 0.35° at half their height) indicate that crystal lattices of the phases that the films were composed of are sufficiently faultless. The width value of the reflections from the phases constituting the films is comparable with the width value of reflections from the substrate bulk single crystal. X-ray diffraction peaks due to oxygen resulting from deficiency of lead, i.e. the pyrochlore phase deficit, were not observed in X-ray diffraction patterns, which indicates good phase purity within the resolution of the X-ray diffraction meter (2 %). Similarly, X-ray diffraction peaks from other phases were not recorded, too, which indicates the absence of any significant chemical reactions at the interface. Formation of the perovskite phase may be favored by the presence of an appropriate structural and chemical template, i.e. the surface of the YBCO lower electrode (001) oriented along the c-axis.
Measurements of polarization switching characteristics have revealed the existence of ferroelectric hysteresis (see Fig. 9). Typical values of residual polarization for typical samples are 30–40µC/cm2.
Measurements of voltage-capacitance characteristics were carried out at temperatures within 20–100 °C, frequencies of 1kHz, 10kHz, 100kHz and 1MHz, and the 0.1V test signal amplitude. In all cases, for (PTO+YBCO)/STO heterostructures the smoothness of voltage-capacitance characteristics was disturbed in the areas of active scanning stroke of applied voltages for electrostatic field strength values close to the coercive ones (Fig. 10), which is not typical for ferroelectric samples of the PZT/Si type.
Accordingly, voltage-capacitance curves had not two but four local maxima. As shown in Fig. 10, this effect takes place at different sample temperatures. It is for the first time that such an appearance of voltage-capacitance curves for ferroelectric structures was observed, which can be explained within the framework of the model of the built-in electric field near boundary areas that leads to an interaction between FE and HTSC structures.
One explanation for such behavior of structures can be emergence of mobile charged surface states at the FE-HTSC interface due to the dynamics of the switching process of the ferroelectric polarization. The effects associated with the existence of such states in single-layer structures are described, in particular, in  where it has been shown that they lead to different positions of the maxima of quasi static voltage-capacitance characteristics and maxima obtained by differentiating FE hysteresis loops. There’s a number of papers analyzing the “negative capacitance effect” in structures with FE-based thin films  caused by appearance of unstable states upon switching polarization. Table 2 gives estimates of the corresponding values of charges derived from the differences between measured values of differential capacitances and values of the capacitances approximated by the “smooth” decay of voltage-capacitance curves similar to that observed for FE-based single-layered structures (see Fig. 2 and dashed lines in Fig. 10). Error of estimate values is ±0.002–0.007uC/cm2.
Such an effect in FE-based heterostructures can be important when creating new devices for microelectronics based on the capacitance dependence on applied voltage. In particular, it should be taken into account when designing MIS field-effect transistors where FE thin layers are used.
The authors believe that this work has revealed the following new statements and results:
When measuring voltage-capacitance and voltage-current characteristics, the methods of scanning modes with arbitrary setting were used, and this has made it possible to obtain new data on dielectric characteristics of structures based on thin ferroelectric films.
Voltage-capacitance characteristics have allowed studying the effects associated with the Schottky barrier formation at the boundary regions in multilayer structures based on FE films. In order to obtain quantitative characteristics of the transition layers, we have suggested a model where the behavior of heterogeneous structures based on ferroelectric films in electric field intensity is determined by the distribution of surface states of defects and built-in charges in transition layers, as well as the dynamics of their behavior. Quantitative estimates of voltage drops and the density of surface states drops in these layers coincide with the estimates made by the dynamic method for measuring capacitance of ferroelectric structures. It has been shown that the density of surface states’ values depend on the technology of preparing layered structures, and the corresponding voltage drops across the intermediate layer vary from tenths of a volt to several volts.
For the (PTO+YBCO)/STO structures, the effects that are atypical for ferroelectric samples were observed when the smoothness of voltage-capacitance characteristics was broken in areas of active scanning stroke of applied voltages for the electrostatic field strength values close to the coercive ones. It has been shown that this effect can be explained by the existence of built-in mobile electric charges at the boundary between the FE and HTSC, and a method of estimating the values of these charges has been proposed.
The work was supported by the Ministry of Education and Science (state assignment, project code: 8.5098.2017/Base part).
1. Vorotilov K. A., Mukhortov V. M., Sigov A. S. Integrated Ferroelectric Devices // Moscow: Energoatomizdat, 2011, 175 p. (In Russian). ISBN 978-5-283-00872-1.
2. Ferroelectrics — Physical Effects, Edited by Mickaël Lallart, ISBN 978-953-307-453-5, 666 pages, Publisher: InTech, Chapters published August 23, 2011 under CC BY-NC-SA 3.0. DOI: 10.5772/942.
3. Pevtsov E. F., Chuyko A. V., Khodorovich V. G. Experimental Studies of Ferroelectrics-based Structures // Current Problems of Piezoelectric Instrumentation: Proc. II Intl. Scientific. Conf., Rostov-on-Don, September 6–10, 2015; SFU — Rostov-on-Don: ed. SFedU, 2015, Vol. 1. — p. 33–38. (In Russian). ISBN 978-5-9275-1650-0.
4. Park B., Hyun S., Noh T., Lee J. Effects of Interfacial Charges on Electrical Asymmetry of Epitaxial Bi4Ti3O12 Thin Film Capacitors // J. Korean Phys. Soc. 1998, Vol. 32, Pt. 4, p. S1405–S1407.
5. Pevtsov E., Sigov A., Pyzhova A., Gorelov A. The Investigations of Ferroelectric Thin Films in Virtual Measuring System // Micro- and Nanoelectronics — 2003. Proceedings of SPIE. — V. 5401. — 2004. P. 520–524.
6. Sreenivas K., Bjðrmander C., Gri¬shin A. M., Rao K. R. Ferroelectric Properties of Epitaxial PbTiO3/YBa2Cu3O7–δ/SrTiO3 thin film heterostructure // Microelectronic Engineering, 1995. — V. 29. — P. 119–121.
7. ICCD database, files 75-1606 (PTO), 86-0477 (YBCO), 84-0444 (STO).
8. Goltsman B. M., Yarmarkin V. K., Lema¬nov V. V. Influence of Mobile Charged Defects on Dielectric Nonlinearity of Ferroelectric PZT Thin Films // Solid State Physics, 2000, Vol. 42, Issue 6, p. 1083–1086. (In Russian).
9. Catalan G., Jiménez D., Gruverman A. Ferroelectrics: Negative Capacitance Detected // Nature Mater. 2015. Vol. 14. — P. 137–139. Doi:10.1038/nmat4195.