Investigation of piezoelectric 0.65Pb(Mg_1/3Nb_2/3)O₃–0.35PbTiO₃ films in cross section using piezo-response force microscopy

Piezoelectric/ferroelectric thin and thick films are being extensively studied for various applications such as sensors, actuators, ultrasound transducers, energy harvesters, pumps, and power generators.^1–8 The piezoelectric response and ferroelectric domain structure of films can be studied using piezo-response force microscopy (PFM), which is a variant of atomic force microscopy (AFM). In the PFM technique, the electric field is usually applied between the bottom electrode of the sample and a conductive PFM tip, which acts as a movable top electrode and scans across the sample surface. The piezoelectric and ferroelectric responses at the film surface are recorded, and these may be different from the responses inside the films. Readers interested in learning more about the PFM technique can refer to Refs. 9–18.

So far, several papers have investigated piezoelectric films in the cross section using PFM.^19–24 In some of these reports, the electrical voltage was applied between the top and bottom electrodes of the thin-film sample, and the AFM tip was used only to determine the mechanical displacement of the sample. Alternatively, the samples were wedge-polished at a selected angle. In others, the sample was subjected to a preparation procedure similar to that used in transmission electron microscopy, including thinning to electron transparency by ion milling or the preparation of thin lamella using a focused ion beam and surface polishing of the samples with gallium ions. However, such a destructive process can affect the initial ferroelectric domain structure of the sample.

In this investigation, we develop a slightly different PFM-based approach to study the local piezoelectric/ferroelectric behavior, where the PFM tip acts as a moving top electrode during the PFM scanning in a virtual ground mode. The purpose of this experiment was to investigate the interior of the film samples and to determine whether there are differences in the ferroelectric domain structure in the film samples, especially between the regions near the substrate and those far from the substrate. The differences could be due to the internal stresses caused by different boundary conditions during the processing or heat treatment, i.e., because the near-substrate region is clamped while the near-surface region is free on the one side or to other external stimuli such as the electrical poling of the sample.

The relaxor-ferroelectric (1-x)Pb(Mg_1/3Nb_2/3)O₃–xPbTiO₃ (PMN–PT) with a morphotropic phase boundary composition around x ≈ 0.35 (PMN–35PT) has always been the subject of intense research because of its large piezoelectric response,^25–29 high dielectric permittivity and polarization,^27,29 and excellent energy-storage performance.³⁰ As a consequence, screen-printed and aerosol-deposited PMN–35PT films were used as showcase samples in this work.

The mechanochemically synthesized powders were prepared from PbO (99.9%, Aldrich), MgO (99.95%, Alfa Aesar), TiO₂ (99.8%, Alfa Aesar), and Nb₂O₅ (99.9%, Aldrich), in accordance with Ref. 31, and the details of the synthesis procedure are described in supplementary material S1. Screen printing and aerosol processes were chosen for the deposition of the PMN–35PT films, because they rely on different densification mechanisms, which results in PMN–35PT films with different microstructural properties. In the screen-printing process,^2,32,33 the powder is mixed with an organic carrier to form a thick-film paste that is applied to the substrate. The green samples are densified at temperatures typically above 900 °C. The films prepared using the screen-printing process are usually between 10 and 100 μm thick and have a grain size in the micrometer range. In aerosol deposition,^34–36 on the other hand, the powder is not bound and compacted by thermal energy, but by the high kinetic energy at room temperature. These thick films have very high densities and grain sizes in the nanometer range.^34,37–40

In this work, the PMN–35PT screen-printing paste was prepared from the synthesized powders and an organic carrier, i.e., alpha-terpineol-2–2-butoxyethoxyacetate and ethyl cellulose in the ratio 60/25/15. The paste was screen printed on Pt substrates (Pt 999, Zlatarna Celje) and sintered at 950 °C for 2 h. On the other hand, the PMN–35PT powder was aerosol deposited onto stainless-steel substrates (SS no. 304, polished surface A480: No. 8, American Iron and Steel Institute). After aerosol deposition, the samples were annealed at 500 °C for 1 h. Further details of the film preparation can be found in supplementary material S1 and in Refs. 30, 41, and 42. The scanning electron microscope (SEM) microstructures of both film surfaces are shown in Fig. 1. The grains in the aerosol-deposited films are tens to hundreds of nm in size, while the grains of the screen-printed films are a few micrometers in size. This is due to the fact that during the aerosol deposition process, the powder is subjected to high-velocity particle impact on the substrate, which causes the particles to break down into smaller particles that recombine with the substrate to form the ceramic film. This is in contrast to the screen-printed process where the powder is mixed with organic vehicles to form the paste, which is screen-printed to the substrate. In this case, the ceramic powder is not submitted to large mechanical stress during the deposition. Furthermore, in the aerosol deposition process, very high internal stresses are induced during the film growth. In addition, the aerosol-deposited films are extremely densely packed (density ≥ 98% of the theoretical value), while screen-printed films are more porose due to a high organic load that leaves large voids in the film after burnout, resulting in a lower density. Finally, the most important parameter is the thermal budget the films are subjected to after preparation. The aerosol-deposited films were exposed to a temperature of 500 °C for 1 h, while the screen-printed films were exposed to a temperature of 950 °C for 2 h, which allowed the grains to grow.

An atomic force microscope (AFM, MFP-3D and Jupiter, Asylum Research, USA) and Ti/Ir-coated silicon tips (Asyelec 01, Oxford Instruments, Germany) with a high-voltage option were used for the analysis. Normally, the piezoelectric films are scanned with the PFM tip on the top of the sample. However, if we are interested in thickness-dependent properties, PFM scans can also be performed in the cross-sectional view. For this analysis, the films were embedded in epoxy plastic and polished in the cross section. The grinding involved silicon carbide abrasive papers (SiC particles with a size of 5–20 μm) with subsequent polishing on a fabric with diamond paste down to 0.25 μm and final polishing with a colloidal SiO₂ suspension containing particles with a size of only a few tens of nanometers. Distilled water was used to clean the surface of the samples. The water was then removed by blowing the specimens with air. The specimens were removed from the plastic mass, leaving the epoxy layer around it (see supplementary material S1). The specimen was then placed in a specially designed sample holder with the polished cross-sectional surface on top [Fig. 1(c)].

The experiments in cross-sectional views were performed in the virtual grounding mode (V_gnd), as shown in Fig. 1(c). V_gnd was previously used for the study of piezoelectric nanoplates embedded in plastic.⁴³ The conductive AFM tip was placed on the surface of the film in the cross section. The bias voltage was applied to the conducting PFM tip. The schematic experimental setup is shown in Fig. 1(c). PFM imaging was performed in the dual-ac resonance tracking mode (DART) in order to reduce the crosstalk with topography.^44,45 The electric voltage with an amplitude of 5 V or 25 V and a frequency of ∼300 kHz was applied to the screen-printed and aerosol-deposited PMN–35PT thick films, respectively. The PFM amplitude and phase hysteresis loops were measured with the pulsed dc step signal and the superimposed ac drive signal, as described in Ref. 16. Three cycles were measured in the off-electric-field switching spectroscopy mode. The waveform parameters were: the sequence of the rising dc step signal was driven at 20 Hz with a maximum amplitude of 80 V; the frequency of the triangular envelope was 1 Hz; a superimposed sinusoidal ac signal with an amplitude of 10 V and a frequency of ∼350 kHz was used.

Macroscopic piezoelectric measurements were performed on parallel samples from the same batch. For the macroscopic piezoelectric measurements, the gold electrodes (2r = 1.5 mm) were deposited on the top of the PMN–35PT films using a RF magnetron sputtering system (5Pascal, Italy). The macroscopic piezoelectric coefficients d₃₃ of the films were determined with a direct pseudo-static method using a Berlincourt piezometer (Take Control PM10, UK) at a driving mechanical stress frequency of 100 Hz.

The AFM/PFM images of the screen-printed PMN–35PT thick films in the cross section are shown in Fig. 2. The root mean square surface roughness (R_q) of the height images in Figs. 2(a), 2(d), and 2(g) ranges from 6 to 69 nm, depending on magnification, indicating that the surface of the piezoelectric film is nanometer smooth after polishing. The images in the first row [Figs. 2(a)–2(c)] were taken from the entire film thickness, i.e., from the free surface to the substrate, since the film thickness is about 50 μm. From the PFM analysis, it is clear that the film is piezoelectric. The relaxor-ferroelectric domain structure is observed over the entire film thickness and does not change from one side of the sample to the other, although the film was subjected to an average compressive stress of about −170 MPa during the thermal treatment (cooling).^28,41 Larger magnifications reveal an easily visible domain structure inside the micrometer-sized grains [Figs. 2(g)–2(i)]. Wedge-shaped and irregularly shaped domains were observed. A similar domain structure was observed in the bulk PMN–35PT ceramics, as shown in supplementary material S2, in Refs. 46–48 and in other Pb-based solid solutions.^49,50 The in situ electrical poling, in-plane (lateral) DART PFM, and vector PFM experiments of screen-printed PMN–35PT thick films in cross-sectional view are shown in supplementary material S3–S5.

Completely different cross-sectional PFM images are observed for the aerosol-deposited PMN–35PT thick films with a thickness of ∼3 μm (Fig. 3). For these films, the average grain size is less than one micrometer, i.e., in the range of tens of nanometers, and thus, smaller than for the screen-printed films shown in Fig. 2. This is related to the mechanism of aerosol deposition and the thermal treatment of the samples (only 500 °C, i.e., 450 °C less than the thermal treatment of the screen-printed samples). Note that in Figs. 3(a) and 3(b) in the left corner, a piece of the epoxy layer is visible.

To prove that both types of PMN–35PT films are piezoelectric and that the measured PFM signal is connected to the piezoelectricity of the samples,^51–53 the macroscopic piezoelectric response of the films was investigated using the Berlincourt piezometer, as described above. The macroscopic piezoelectric response of the aerosol-deposited films is lower than for the screen-printed films, i.e., the d₃₃ of the screen-printed PMN–35PT on Pt is 140 pC/N, while the d₃₃ of the aerosol-deposited films on stainless steel is only ∼25 pC/N. The differences in the macroscopic piezoelectric properties are most likely related to different microstructures of both films (differences in grain sizes and densities can in PMN–35PT lead to different piezoelectric performance^26,54) differences in stresses (according to the literature, the residual compressive stresses in AD films reach a few hundreds of MPa⁵⁵) and to the order-of-magnitude-smaller thickness of the aerosol-deposited films compared to screen-printed films, which leads to a higher clamping force of the film to the substrate.^56,57 This much lower piezoelectric response of the aerosol-deposited films is most likely the reason for the higher voltage that had to be applied to the aerosol-deposited films than to the screen-printed films during the PFM investigation. Similar to the screen-printed films, a homogeneous relaxor-ferroelectric domain structure is also observed for the aerosol-deposited thick films throughout the film thickness, which does not change from one side of the sample to the other. In this case, however, the domains are smaller, which is due to the submicrometer size of the grains [Figs. 1(b) and 3(e)]. To further prove that the PFM signal in AD films is related to piezoelectricity, the switching spectroscopy experiment was performed, which led to the switching of ferroelectric domains and, thus, to local PFM amplitude and phase hysteresis loops typical for ferroelectrics [Figs. 3(f) and 3(g)].

In summary, the modified PFM setup developed in this work is suitable for the cross-sectional studies of piezoelectric/ferroelectric films. The samples are polished in the cross section using standard metallographic techniques. The samples are placed in a specially designed sample holder with the polished cross-sectional surface facing up. A bias voltage is applied to the conductive AFM tip in the virtual grounding mode. To demonstrate the usefulness of this setup for studying the ferroelectric domain structure of thin films in the cross section, PMN–35PT films prepared by screen printing and aerosol deposition were studied. Both types of films are piezoelectric, but due to the different preparation techniques, the microstructure, namely, the grain size and densities, the thickness and internal stresses and, thus, also the relaxor ferroelectric domain structures differ in these two films. Micrometer-sized, wedge-shaped, and irregularly shaped domains are observed in the screen-printed films, similar to bulk ceramics of the same composition. On the other hand, submicrometer-sized domains are observed in the aerosol-deposited films. However, for both types of films, no change in the piezoelectric response or the domain structures was observed across the film thickness. This method can be applied to other piezoelectric films to investigate the piezoelectric performance and the ferroelectric domain structure across the film thickness.

See the supplementary material for the procedure of powder synthesis, the preparation procedure and properties of screen-printed and aerosol-deposited PMN–35PT films (S1), PFM analysis of bulk PMN–35PT ceramics (S2), in situ electrical poling experiments of screen-printed PMN–35PT thick films in the cross section (S3), in-plane (lateral) DART PFM imaging of screen-printed PMN–35PT films in the cross section (S4), and vector PFM experiments of screen-printed PMN–35PT films in the cross section (S5).

The authors acknowledge the Slovenian Research Agency (Project No. J2-3058, young researcher project PR-08977, research core funding P2-0105) and JSI Director's fund 2017-ULTRACOOL. The authors thank Val Fišinger, Nikola Tutic, Silvo Drnovsek, and Brigita Kmet.