Epitaxial Ferroelectric Thin Films: Potential for New Applications

Abstract

This chapter provides an overview of the versatile applications and properties of epitaxial ferroelectric materials obtained using the pulsed laser deposition technique. These materials can play a significant role in various electronic and sensing applications or energy harvesting. Materials that are ferroelectric and have a perovskite structure (ABO3 type) show spontaneous polarization that can be changed by an electric field, temperature, mechanical stress, or light. Here, we present results obtained on epitaxial ferroelectric thin films with different compositions, lead-based or lead-free, and the correlation with structural quality of the layers and with different electrostatic conditions induced either by the substrate or by the different dopants. Our studies revealed that the utilization of pulsed laser beam deposition technique is suitable for obtaining ultrathin films depositions with thicknesses measuring less than 5 nm. These results allowed us to reveal the impacts caused by polarization orientation on the band structure or the presence of self-doping phenomena. We also found that the conduction type can be modified by introducing 1% Fe and Nb on PbZrTiO₃ (PZT) epitaxial layers. In the last part of this chapter, we report on obtaining of a lead-free epitaxial thin film and its properties in the energy storage field.

1. Introduction

The complexity of electronic devices has increased due to the development of advanced technologies such as 5G, the Internet of Things (IoT), and artificial intelligence. However, due to consumer preferences for portable gadgets and technological advancements, there is a trend toward smaller and more compact electronic devices. This trend requires materials that can maintain or enhance performance while occupying less space. At the same time, as more and more people become concerned about environmental impact and power consumption, there is a demand for materials that can aid in the creation of more efficient electrical equipment. Such materials are the ones that possess a spontaneous electric polarization P, which is reversible and stable over time under external electric fields greater than the coercive field (i.e., ferroelectrics). The paraelectric to ferroelectric transition is known to entail a structural change from high to low symmetry, which causes the atoms to be off-centered and displaced from their symmetric orientations [1]. Ferroelectric thin films are used in many electronic and sensing applications, such as ferroelectric field effect transistors, pyroelectric infrared detectors, and supercapacitors. Epitaxial thin films offer advantages in terms of controlled growth and strain engineering, which increases their ability to be tailored for specific applications. When targeting an application that employs functional elements, it is crucial to examine the mechanisms, both intrinsic and extrinsic, that play a role in stabilizing a clearly defined ferroelectric state. It is also vital to separate and analyze how these mechanisms affect electrical properties. Currently, magnetron sputtering and pulsed laser deposition (PLD) provide good results with respect to high epitaxial quality [2]. But there is still more to be done: to make the targets more chemically pure, to keep the deposition chamber free of contaminants, precisely control the reactive gas level, and to make sure that the substrates are chemically and thermally treated prior to deposition, with uniform terraces.

The choice of substrate is crucial in PLD to obtain high-quality epitaxial thin films with a well-defined crystal orientation. Monocrystalline substrates, such as SrTiO₃ (STO), LaAlO₃ (LAO), and MgO, are commonly used for epitaxial thin films due to their similar lattice structures and low misfit strain [3, 4]. The choice of bottom electrode material has a substantial impact on the ferroelectric properties and total losses, following the influence of the substrate. The bottom electrode can be either a basic metal or a conductive oxide. The use of conducting oxide materials as bottom electrodes enables the creation of a ferroelectric layer with minimum imperfections and strain mismatch [3, 5].

Here, SrRuO₃ (SRO) and LaSrMnO₃ (LSMO) conductive oxides have been used like bottom electrodes for all structures. Even though lead-based ferroelectric materials are widely studied and used in solid-state technology, research into their fundamental properties, such as the conduction type in ferroelectric films or their intrinsic band structure, is still up for debate. Here, we have investigated these fundamental characteristics in PbZrTiO₃ (PZT) epitaxial layers with different thicknesses, and the electrical characterizations were performed on capacitor-like geometry. For this, top SRO/Au electrodes were deposited by PLD and magnetron sputtering, with a shadow mask, defining ferroelectric capacitors of 100 μm² area. We found also that one way to tailor the PZT properties for certain purposes is through doping.

Doping effects must be studied on materials of excellent crystalline quality, ideally approaching single crystal, to avoid masking effects caused by structural defects, such as grain boundaries. PZT single crystals are challenging to synthesize due to PbO volatility during growth. However, high-quality epitaxial layers can be grown using methods, such as pulsed laser deposition (PLD) on suitable single-crystal substrates, such as SrTiO₃ (STO). This advancement enables the study of dopants’ effects on electrical properties in epitaxial PZT layers at doping levels around or below 1% atomic. The following results show the impact of 1% Fe and Nb doping on the electrical properties of epitaxial PZT films.

Simultaneously, due to concerns about lead-based materials, particularly their environmental impact, we investigated alternate materials. One such material is bismuth-based perovskite compounds, which have been found to exhibit properties useful on energy storage, as well as being nontoxic and environmentally friendly. Bi_0.5Na_0.5TiO₃ (BNT) is considered one of the candidates as lead-free material for electrical energy storage due to its high dielectric constant and excellent ferroelectric properties. Until now, a large number of publications on the energy storage properties of lead-free ferroelectrics are related to ceramics or polycrystalline thin films [6, 7, 8, 9]. Here, the lead-free ferroelectric epitaxial thin films have been developed and researched.

2. Pulsed laser deposition

Due to its flexibility, PLD method is now widely used for the deposition of oxide materials, especially for research, and particularly of multicomponent oxides. This technique offers very good translation of stoichiometry from the target to the deposited film and proved to be already scalable for deposition on large area wafers in the future [10, 11]. PLD process also depends on the type of laser used. It must consider the absorption coefficient and reflectivity of the materials that will be deposited. Because these properties vary with wavelength, the laser must be able to operate in the wavelength range where the target material has low reflectivity and high absorption coefficient for efficient deposition. Excimer lasers (XeCl, KrF, and ArF) are widely utilized for the deposition of complicated oxide films because, at the wavelengths used by these lasers, these materials have a high absorption coefficient and low reflectance [12]. The process of PLD involves several key steps. First, a high-intensity laser pulse is directed onto a target material composed of the desired oxide. This laser pulse vaporizes or ablates the target, creating a plasma plume consisting of atoms, ions, and clusters of the target material [13]. The substrate, typically positioned parallel to the target is then exposed to this plasma, allowing the deposition of the ablated material onto its surface, Figure 1.

Figure 1.

Schematic representation of pulsed laser deposition with parallel setup.

The growth dynamics of thin films produced by PLD are highly dependent on various parameters such as target composition, substrate temperature, laser fluence and frequency, the spot size, and the background pressure [14]. Multiple targets are contained in the target holder, four in our system, enabling the deposition of multilayer thin films. In order to prevent the formation of droplets on the surface of the films during the laser deposition process, it is preferable to use a target with high density, purity, and a reduced laser pulse frequency.

Deposition often involves constant translation and rotation of the target support. The substrate heating block, on our system, has the capability to reach temperatures as high as 1000°C. The experimental results showed that there is a critical substrate temperature (Tc) below which the film structure is not completely monocrystalline and the film composition deviates significantly from the stoichiometric one [15, 16]. The optimal frequency of laser pulses for film preparation depends on the ambient gas pressure and the substrate-target distance [17, 18]. The interaction between the plasma and the substrate has a major impact for deposition. A balance between the target heating speed and substrate temperature is required to achieve the desired film composition and properties. Short-duration pulses (10⁻³ s) and energy densities of 10⁶–10⁸ W/cm² allow for the achievement of heating speeds of 10¹⁰ K/s [13, 14] at the target surface. The heat loss of the components arriving on the substrate is determined by the substrate temperature. The rapid cooling of the compound causes it to crystallize at a slower rate. Due to the extremely short distances between atoms on the surface of the substrate, only a small number of them are able to form crystals.

The temperature of the layer, thus, formed increases when the next vapor pulse is deposited. The impact of the impinging flux of ablated particles on this temperature becomes more significant at higher deposition frequencies and should be considered, particularly when considering the formation of materials with low thermal conductivity using PLD [19].

If the substrate temperature is sufficiently high, the material’s velocity at the substrate surface will decrease, hence facilitating the optimal conditions for complete crystallization, nucleation, and coalescence of the atoms [15, 20]. To ensure proper stoichiometry, the rate of arrival of the constituents of one complex compound must be carefully managed, and this rate is also being affected by the gaseous environment. The gas in the deposition chamber affects the crystalline microstructure and orientation by its chemical activity, which is naturally proportional to the pressure [15]. Complex oxide films are typically prepared in oxygen atmosphere at a pressure of several hundred millitorr. The use of high oxygen pressure is employed for the deposition of complex oxide films, but above a certain point, it can have the opposite effect of what you would expect by slowing down the deposition rate due to oxygen’s high dissociation energy barrier, which makes it chemically reactive [4, 16, 21, 22].

3. Termination control for epitaxial thin films

Achieving epitaxial growth (where the crystal structure of the film aligns with that of the substrate) can be challenging if there are significant differences in lattice parameters [23, 24]. The lattice constant (spacing between atoms) of the thin film may not perfectly match that of the substrate. This can result in strain and defects in the film, affecting its structural properties. Prior to epitaxial growth, the substrate surface needs to be carefully prepared to achieve the desired termination. Substrate termination can impact the nucleation process, affecting the initial stages of thin film growth [25, 26, 27]. Controlling nucleation is crucial for achieving a high-quality and uniform epitaxial layer. Techniques, such as chemical etching, annealing, or other surface treatments, are employed to modify the crystal surface and remove contaminants. The examined structures in this chapter were developed using single crystalline STO (001) and Nb: STO (001) substrates supplied by CrysTec GmbH. These substrates had a miscut angle ranging from 0.05° to 0.5°. Prior to achieving higher quality thin films on single crystalline STO and Nb:STO substrates, preliminary substrates preparation were performed. These transformations involve transforming the optically polished surface into a surface with steps and terraces that are highly organized at the atomic level. To achieve this result, the STO substrates required etching in an NH4-HF solution for a duration of 15 seconds to eliminate any Sr. residues. Following that, the substrates were placed under a process of thermal treatment at a high temperature of around 1000°C for a period of about 4 hours. By following this procedure, a surface consisting completely of TiO₂ is achieved, with approximately the same height steps (about 0.4 nm per unit cell), having parallel position to each other, are selected. The miscut angle value has an important effect on the surface preparation process. While both chemical and thermal annealing follow the same procedure way, we found that different substrates miscut produce different terraces; especially those substrates having a smaller miscut, we noticed that adjusting the annealing period improved the terrace quality (Figure 2). Other groups also reported similar behavior on monocrystalline substrates, which have been found to be correlated with the miscut angle value [28, 29]. However, it was challenging to determine the exact times terrace obtaining.

Figure 2.

Atomic force microscopy images from identically cleaned substrates with varying miscut angles.

Mismatches between oxidic thin films and monocrystalline substrates can be minimized by introducing a buffer layer with intermediate properties between the thin film and substrate. Buffer layers also provide an approach for achieving preferential polarization orientation; they can contain free charges induced by point defects that form during the deposition process. The presence of these defects acts for compensating the depolarization field and maintain the upward polarization direction [30, 31]. In this chapter, conductive oxides, such as LSMO and SRO, were utilized as buffer materials and bottom electrodes in order to form structures similar to capacitors. The electric and magnetic characteristics of LSMO are dependent by several parameters, including La, Sr content, deposition temperature, thickness, and the crystalline structure of the layer [32, 33, 34]. The choice of using La₇₀Sr₃₀MnO₃ as the bottom electrode was based on our previous study of the resistivity properties, particularly its dependence on thickness [33].

3.1 Conditions for buffer layer preparation

A KrF laser (𝜆 = 248 nm) was utilized to ablate Praxair’s commercial SRO and LSMO targets for buffer layer deposition. The repetition rate for LSMO and SRO was 1 and 5 Hz, respectively, with a laser fluence of 2 J/cm². During deposition, the substrate temperature was set to 700°C, with an oxygen atmosphere of 0.133 mbar for SRO and 0.27 mbar for LSMO. The target and substrates were kept at a distance of 60 mm for both types of buffer layers. Further, on these structures, lead-based or lead-free layers with different thicknesses were deposited.

3.2 Lead-based ferroelectric films: ultrathin films

In this section, we describe our investigations related to the preparation of 5 nm thick PZT ultrathin films by using a commercial target from Praxair with a Zr doping x = 0.2. The films were grown on different ferroelectric states, and the 3D band structure was studied using SX-ARPES [35]. By discerning between the effects caused by ferroelectricity and those induced by the substrate on the electronic band structure, we are able to elucidate basic questions concerning the fundamental electronic properties of oxide interfaces and provide further avenues for enhancing the performance of multiferroic systems. The PZT films were grown on LSMO-buffered TiO₂-terminated STO substrates (001) oriented. In this case, the ferroelectric polarization points toward the interface (from herein named DW sample and denoted as P-). Samples grown on TiO₂-terminated, Nb:STO substrate feature an upwards polarization state (P+) and will be designated as the UP sample, with an out-of-plane FE state, mostly single domain. The growth conditions for PZT films were similar to those detailed in our previous work [35].

3.3 Structural characterization

The symmetric 2θ-ω XRD scans, which are displayed in Figure 3a, performed at room temperature on a Rigaku-SmartLab diffractometer in high-resolution setting (X-ray mirror and two bounce Ge (220) monochromator) demonstrate the good crystallographic resemblance of both PZT films. However, DW has somewhat wider peaks than UP, possibly because DW has larger internal microstrains [36]. The PZT films in both DW and UP are fully strained at the in-plane lattice constant of the cubic STO and Nb:STO substrates, as indicated by the RSM near the −103 node of STO (Figure 3b) and Nb:STO, Figure 3c, performed on a Bruker D8 Advance diffractometer (Bruker AXS GmbH Germany) with a copper anode X-ray tube in medium resolution parallel beam setting. PZT exhibits elongated growth along the c-axis in both samples, with the ratio of the in-plane to out-of-plane lattice parameters c/a = 1.073 for UP and 1.081 for DW. The significantly off-centering of the Pb and Ti/Zr cations in the unit cell ensures the stabilization of the FE character of the thin films. The local PFM measurements (Figure 4) on the two samples determined the different out-of-plane FE polarization, pointing downward or outward. The outcome confirms that PZT in the DW sample has indeed the FE polarization orientated inward (P−), whereas PZT in the UP sample has the FE state oriented away from the surface (P+).

Figure 3.

(a) X-ray diffraction. 2θ-ω scan for PZT grown in UP and DW samples; in the inset, the zoomed region of the 001 peak of PZT in linear scale of intensity is given for both samples; (b), (c) reciprocal space mapping (RSM) scans of DW and UP samples are presented in and, respectively [35].

Figure 4.

(a) Piezoresponse phase loops versus applied DC voltage, measured with an AC driving voltage Vac = 1.5 V for DW and (b) UP samples [35].

3.4 k-resolved valence band structure

By navigating in the out-of-plane k_z direction, we first examine the k-space topology of PZT in the UP sample with tetragonal (TG) unit cell. The photon energy h𝜈 in Figure 3a is adjusted at constant k_|| to range from 350 to 520 eV. The resulting iso-energy map, shown in Figure 5a, in the XΓZ plane of the bulk Brillouin zone, locates the valence band maximum (VBM) at a binding energy (BE) of 2.2 eV with respect to the Fermi level in the X point. According to the iso-energy surface obtained from DFT calculations for the tetragonal PZT unit cell, Figure 5c, the square-like symmetry of the PZT in the ab plane accounts for the (k_x − k_y) iso-energy map observed on the UP sample in the XΓM plane at the VBM using hv = 520 eV (Figure 5b).

Figure 5.

(a) Out-of-plane iso-energy (iso-E) maps of PZT recorded by varying hv between 350 and 520 eV while keeping the k_|| in the XΓZ plane of the bulk Brillouin zone; (b) (k_x − _ky) iso-E in the ΓMX plane at the VB maximum—VBM—and (d) in the ZAR plane at 0.5 eV below the VBM. The calculated iso-E surface at (c) VBM and (e) 0.5 eV below VMB [35].

The DFT-predicted pattern centered in the Z point for the tetragonal unit cell, Figure 5e, is also visible on the iso-energy map taken in the ZAR plane 0.5 eV below VBM with hv = 465 eV (Figure 5d). Nevertheless, for the DW sample, the iso-energy estimated in the tetragonal cell as depicted in Figure 6a is inconsistent with the iso-energy map obtained in the ZAR plane 0.5 eV below VB. It reveals four elliptical-shaped features centered in the A points of the k-space in addition to the predicted signature of the hole-like band in the Z point. Their appearance is consistent with a rhombohedral (RH) reconstruction of the PZT unit cell, as shown by the 3D iso-E surface of PZT, Figure 6b. The primary axis of the pocket surrounding the A point, as indicated by the computed iso-E map, is derived from the projection of the RH cell’s WXW direction on the TG-cell’s (001) direction. Consequently, projecting the RH cell’s UXU direction onto the (001) plane of the TG-cell BZ yields the minor axis of the A-centered ellipse.

Figure 6.

(a) Signature of RH reconstruction. In-plane (k_x, k_y) iso-E maps of PZT recorded with hv = 465 eV at 0.5 eV below the VBM in the ZAR plane of the TG cell. The red line indicates the calculated iso-contour at the corresponding energy assuming TG geometry of the unit cell. Gray contours represent the additional signature of the RH-distorted unit cell for DW, and such a signature is absent for the UP sample; (b) the signature of the RH distortion is rendered into the TG unit cell, with the iso-energy surfaces calculated at the same energy, 0.5 eV below VBM, as the experimental maps [35].

3.5 Polarization-dependent band alignment mechanism

We have chosen the two distinct substrates because of their correspondingly larger and smaller work functions (W_f). The migration of negative (positive) charges from PZT at the bottom interface is driven by the material-dependent band bending, Δϕ^DW(UP), defined by the W_f-induced band lineup at the interface with the substrate. We anticipated the stability of distinct, opposing out-of-plane P−(P+) FE states of PZT in conjunction with the accessible positive (in LSMO) and negative (in Nb:STO) charges in the substrate.

The contaminated layer is identified as the cause of improper screening for the depolarizing field at the PZT surface. Based on the examination of the core levels recorded on both UP and DW samples, we will now clarify the mechanism of band alignment. Thus, by opposing the FE-induced field, the residual uncompensated field tends to lower the band bending potential at the surface. Thin films use internal or external mechanisms to make up for the depolarizing field in order to maintain their ferroelectric state. The intrinsic one entails the migration of charge carriers that are already present in the film, producing positive and negative charge sheets at the layer’s opposing surfaces.

These charge carriers are produced when the ideal stoichiometry spontaneously changes during growth, and cations or oxygen vacancies are formed [31]. In the external material, such as the metallic electrode, or within the contaminated layer at the surface with the air to screen the DF, the extrinsic one entails opposite charge accumulation or depletion with respect to the fixed polarization charges from the FE material. As the electric field outside the ferroelectric is strictly zero, the charge modulation effect produced into the joining material is greater than a pure electrostatic picture and should not affect the joining material’s electronic structure. However, it is dependent upon two elements: (i) on the specific band alignment at the interfaces between the FE and the metal or (ii) the FE and the contamination layer, which takes into account both the FE state and the work function difference between the FE and the metallic contact or surface contamination layer [37, 38, 39, 40].

The regions with positive/negative charge buildup at the ferroelectric’s extremities are defined by the band alignment and the potential profile that results across the ferroelectric. It regulates the application of bias, determining whether and how the necessary electrons and/or holes are injected across the interface and how they collect at the ferroelectric’s surface and interfaces via diffusing through the material [37, 38, 39, 41, 42]. Figure 7a shows the mechanism of band alignment based on the FE state of the PZT samples at the surface and at their interface with the substrate. It is derived by comparing the more bulk-sensitive results of the core levels, which extend the probing region down to the substrate interface, with the surface-sensitive ARPES data. This shows the variation of the potential inside the FE film going from the surface toward the interface with the substrate.

Figure 7.

(a) Band bending mechanism from the core levels. Band bending at the PZT surface as a function of the ferroelectric state; (b, c) sketch of the band bending potential along the UP and DW samples Pb 4f and Ti 2p core level spectra of DW (blue) and UP (red). Samples recorded at 12 K and after 20 min of continuous exposure to the X-ray beam [35].

The potential profile when opposing charges build up in the ferroelectric film near the interfaces for depolarizing filed screening is shown by the dashed line. The potentials V(z) inside our ferroelectric films, for both UP and DW samples, are qualitatively traced with full lines based on the experimental results. The ARPES observations reveal a 1.25 eV relative binding energy shift between the UP and DW samples. This shift, along with the band dispersions, is consistent with the modest band bending near the surface and the reported ferroelectric-induced band offset across the first ≈1–2 nm near the surface. In the UP sample Δϕ^UP = ϕ_PZT-ϕ_Nb:STO = 0.4 eV, corresponding to upward band bending, and in the DW sample Δϕ^DW = ϕ_PZT-ϕ_LSMO = −0.3 eV associated with the downward band bending of PZT at the LSMO substrate [35]. A negatively charged LSMO interface stabilizes the ferroelectric state pointing toward the substrate (P−), while V(z) in Figure 7b depicts the positive charging into Nb:STO at the interface to correct the ferroelectric polarization pointing away from the substrate (P+) [35, 37, 38, 39, 42, 43]. A necessary prerequisite for the stabilization of the FE state appears to be the substrate-induced band bending at the bottom interface, which makes it easier for positive and negative charge sheets to accumulate at the surface/bottom interface and stablize the P+/P − FE polarization. Figure 5d shows the emergence of a new component in the Pb 4f spectra at lower binding energies, which suggests that the UP sample’s creation of oxygen vacancies and the clustering of metallic Pb were caused by broken Pb-O bonds [41].

The fact that the DW sample lacks this component suggests that the oxygen vacancies and the corresponding negative charges that develop into the UP sample are necessary to offset the depolarizing field and stabilize the P+ FE state. They build up in the vicinity of the PZT surface to form a negative charge sheet that filters the polarization charges, which are fixed. However, in the DW sample with the P− FE state, the formation of cation vacancies (Pb and Ti) screens the depolarizing field produced by the fixed negative polarization charges at the PZT surface and positive at the bottom contact with LSMO.

4. Lead-based ferroelectric films: doped thin films

In this section, we describe our investigations related to the preparation of doped thin films, with Fe acting as acceptor and Nb acting as donor, and their contribution on understanding conduction type in ferroelectric films. In addition to doping studies, the fabrication of p-n junctions in ferroelectric materials is a topic of interest. Our studies revealed significant differences in electrical properties between Fe- and Nb-doped PZT films, including variations in coercive field, potential barrier height at the electrode interface, charge carrier concentration, and leakage current. These differences are attributed to the alterations introduced by the two dopants in the electronic properties of PZT. Furthermore, a change in doping type was found to affect polarization orientation, with upward polarization dominant in PZT-Nb and slightly downward polarization dominant in PZT-Fe. The establishment of p-n junctions in PZT films involves precise deposition of dopants known to behave as donors or acceptors when substituting Zr or Ti in the PZT lattice. These junctions offer novel device structures and functionalities, revealing quasi-linear current-voltage characteristics and temperature-dependent resistance and elucidating carrier injection mechanisms at electrode interfaces.

The PZT films were deposited using pulsed laser deposition (PLD) from custom-made targets containing precursor metal oxides with a purity of at least 99.99%. Specifically, one target was doped with approximately 1% Fe (referred to as PZT-Fe), while the other was doped with approximately 1% Nb (PZT-Nb). The intended doping concentration was about 10²⁰ cm⁻³ in the targets, although the actual doping level in the films might be lower. The deposition took place on single-crystal STO substrates with (001) orientation. First, a thin layer of SrRuO₃ (SRO), approximately 20 nm thick, was deposited by PLD to serve as the bottom electrode. The growth conditions for PZT films were similar to those detailed in a previous [44, 45]. Both single-doped PZT layers and bilayer p-n junctions were prepared in this study. Two types of bilayers were grown, resulting in the following structures: STO/SRO/PZT-Fe/PZT-Nb and STO/SRO/PZT-Nb/PZT-Fe. The thicknesses of the layers were equal, estimated to be approximately 200 nm and in the bilayer architecture to be 100 nm each. The thickness estimation was based on the number of laser pulses used during deposition.

4.1 Structural characterization

Figure 8 illustrates the outcomes of XRD structural analyses for simple PZT-doped layers, while Figure 9 displays the XRD data for the bilayer structures. In all samples, the structure primarily comprises PZTc-domains, where the c crystallographic axis of the tetragonal lattice is perpendicular to the film surface. Additionally, inclined PZT-domains are observed in the reciprocal space maps (Figure 8b), indicating relaxation through a-domains by forming the a/c structure.

Figure 8.

(a) Diffractograms (2θ-ω scans) of single PZT layer samples, with details around lines 001 and 004 in inserts; (b) RSMs around nodes —103 of pseudocubic structures [45].

Figure 9.

RSMs around nodes—103 of PZT bilayer samples. (Reprint from Ref. [44] copyright 2023 American Chemical Society”).

This phenomenon is specific to PZT films with thicknesses exceeding 100 nm. Two distinct types of tetragonal PZT are discerned, characterized by different values of the lattice constant c perpendicular to the substrate. Most films exhibit a tetragonal structure with a c value close to the bulk (referred to as “relaxed”), while layers situated just above the substrate show a larger c value and in-plane lattice constant a, similar to that of the substrate (termed as “strained”). The proportion of the strained component varies depending on the sample. The analyses based on Figure 8c, reveal intriguing characteristics of the films; in contrast to expectations stemming from the lattice mismatch with the STO substrate, the out-of-plane parameter of the films decreases, while the in-plane parameter increases compared to the target.

Moreover, significant discrepancies are noted among the films, surpassing variations observed among the targets. Specifically, the PZT-Nb film exhibits the largest c lattice constant, whereas PZT-Fe displays the smallest; the un-doped PZT falls between these extremes, mirroring the trends observed in the films. Remarkably, the PZT-Fe film manifests a considerably smaller in-plane lattice constant compared to the others despite sharing the same a lattice constant as PZT-Nb. This observation suggests that Nb doping slightly increases the volume of the unit cell, while Fe doping reduces it relative to un-doped PZT. Such behavior can be rationalized by the inhibitory effect of Nb doping on the formation of oxygen vacancies, whereas Fe doping promotes their generation in the films compared to un-doped PZT.

4.2 Polarization switching at nanoscale

A specific poling map was selected to examine polarization switching at the nanoscale for simple PZT-doped layers, as depicted in Figure 10a. The voltage applied to the PFM tip was incrementally increased from −10 to 10 V along each scan line, as shown in Figure 10b. The written ferroelectric domains are visible in the piezoresponse phase images from Figure 11. In the as-grown un-doped PZT film, the polarization is predominantly upward-oriented (see Figure 11f) similar to previous findings [31]. In the case of as-grown PZT-Nb, the dominant polarization orientation is upward, although some areas exhibit a downward polarization tendency. This deviation from un-doped PZT can be attributed to a reduction in the density of oxygen vacancies, influencing the compensation of the depolarization field during the growth process. This finding correlates with the lower leakage current and larger coercive field observed in PZT-Nb. For PZT-Fe, a mixture of domains with both upward and downward polarization is evident, with a larger area demonstrating downward polarization (estimated to about 80% of the as-grown area before poling, as shown in [45]).

Figure 10.

(a) Poling map applied on a 6 × 6 μm surface for polarization switching; (b) section graph in the poling map image shows the applied voltage between −10 and 10 V [45].

Figure 11.

AFM-PFM measurements on PZT-Nb (a), (b); PZT-Fe (c), (d); and un-doped PZT thin films (e), (f). Surface topography images (a), (c), (e) and piezoresponse phase images (b), (d), (f) after domain writing according to the polling map. The symbols used represent polarization orientation, with a solid circle inside a white circle indicating upward polarization and cross lines in a white circle representing downward polarization [45].

This phenomenon can be attributed to Fe acting as an acceptor, facilitating the formation of oxygen vacancies for local charge compensation. Consequently, the distribution of Fe in the film and other structural defects influence the availability of charges for compensating the depolarization field, thereby favoring either upward or downward polarization orientation. Nonetheless, the transition from Nb (donor) to Fe (acceptor) doping visibly influences the PFM phase signal.

4.3 Electrical properties

The results of standard electrical measurements at room temperature are depicted in Figure 12. These measurements were conducted after establishing the same polarization state in the samples by applying a positive voltage on the top electrode for 10 seconds. Un-doped PZT exhibits intermediate polarization between PZT-Fe (lowest) and PZT-Nb (highest), likely due to differences in leakage current and lattice constants. Coercive field for un-doped PZT lies between PZT-Fe (lower) and PZT-Nb (higher), indicating varying ease of polarization switching attributed to leakage current. Internal electric field, though larger in doped samples, is insufficient alone to dictate polarization direction with its orientation consistent across films due to similar substrate conditions. Doped PZT films exhibit slightly higher dielectric constants, attributed to increased structural defects capable of responding to applied voltage signals. In paper [45], it can be also seen that the height of potential barriers at electrodes differs significantly, with un-doped PZT between PZT-Fe (lowest) and PZT-Nb (highest). Also, the effective density of charge carriers shows minimal variation among the three layers, with Fe doping promoting oxygen vacancies formation and Nb doping inhibiting it. Fe introduces an acceptor level near the valence band, while Nb introduces a donor level near the conduction band (details in [45]).

Figure 12.

The electrical characterization of the simple layers of PZT-Fe, PZT-Nb compared with the un-doped PZT. (a) Polarization-voltage loop, (b) capacitance-voltage characteristics, and (c) current-voltage characteristics presented in absolute value and in logarithmic scale [45].

PZT-Nb displays n-type behavior, while PZT-Fe shows slight p-type behavior. Similar effective densities of free carriers correspond to remnant polarization values, indicating polarization’s control over these densities, regardless of doping.

Self-doping, likely via oxygen vacancies, regulates carrier densities, with consistent contributions from SRO electrodes. Differences in barrier height stem from Fermi level positioning: un-doped PZT has n-type behavior due to oxygen vacancies, placing the Fermi level near the conduction band. PZT-Nb′s Fermi level is influenced by donor levels, leading to higher potential barriers. In PZT-Fe, the Fermi level is near the valence band, resulting in low barriers. Variations in barrier height affect leakage current magnitude, with PZT-Fe exhibiting significantly higher leakage than PZT-Nb. Modest Fe and Nb doping induces notable differences in macroscopic properties of epitaxial PZT films. These differences arise from distinct electronic properties induced by Fe and Nb doping. The effects of p-type and n-type doping can be discerned through specialized poling procedures, offering potential for ferroelectric p-n homojunctions.

Hysteresis measurements reveal distinct hysteresis loops for both bilayers, PZT-Nb/PZT-Fe and PZT-Fe/PZT-Nb, with notable differences in loop shape, magnitude, remnant polarizations, and coercive voltages (Figure 13). When PZT-Fe is the first deposited layer, higher polarization and coercive voltage values are observed compared to when PZT-Nb is the first layer. This variance can be attributed to differing electrostatic conditions at the interface with the bottom SRO electrode. As presented above and in [45], the estimated potential barrier is larger for the PZT-Nb/SRO interface than for the PZT-Fe/SRO interface, approximately 0.3 eV compared to 0.1 eV. This distinction can affect the compensation of the depolarization field during both the growth process and polarization switching. A larger potential barrier impedes the flow of charges involved in compensating the depolarization field, resulting in lower polarization values. The I-V characteristics obtained and presented in Figure 13b display an unexpectedly symmetric linear shape despite the differing potential barriers at the two interfaces with the bottom and top SRO electrodes. This contradicts the anticipated presence of Schottky-like contacts at the electrode interfaces in bilayer structures. The unusual shape of the I-V characteristics may be attributed to two factors: (i) the coexistence of two types of carriers (electrons and holes) in the p-n homojunction and (ii) complex charge compensation processes occurring near the contacts and at the PZT-Nb/PZT-Fe interface. Additionally, it is noted that the PZT-Fe/PZT-Nb/SRO/STO structure exhibits lower resistance than the PZT-Nb/PZT-Fe/SRO/STO structure [44].

Figure 13.

(a) The hysteresis measurements for the bilayered structures. (b) The current-voltage characteristics for the bilayered structures (Reprint from Ref. [44] copyright 2023 American Chemical Society).

The slope of the linear I-V characteristic represents the electrical conductance or the inverse of resistance. Resistance values were extracted from I-V measurements at different temperatures, and the observed decrease in resistance at 300 K is approximately four orders of magnitude, indicating semiconductor behavior as presented in [44]. Arrhenius plots were employed to extract an activation energy for conduction in the two bilayer structures. The obtained values fall within the range of 0.13–0.17 eV, which lies between the estimated potential barrier heights at the SRO electrodes (around 0.1 eV at the PZT-Fe/SRO interface and 0.3 eV at the PZT-Nb/SRO interface, as estimated for single-phase capacitor structures).

5. Lead-free thin films

In this section, we present our investigations related to the preparation of lead-free ferroelectric thin films and their properties on energy storage domain. The hysteresis polarization loops analysis is employed in this study to deliver useful information on the energy storage characteristics of lead-free ferroelectric capacitors based on 0.92(Na_0.5Bi_0.5)TiO₃-0.08BaTiO₃ (BNT-BT). The total stored charge and efficiency measured on a different range of temperatures, from 100 to 400 K, indicating an excellent thermal stability. Using an in-house obtained BNT-BT target, an epitaxial thin film was prepared for this investigation on a STO (001) substrate with SRO bottom electrode. More information regarding target preparation is provided in [46]; for the SRO bottom electrode deposition, we always use the same recipe, see above. The oxygen atmosphere, the substrate temperature, and post-annealing conditions were chosen to promote the growth of high-quality BNT-BT film with good crystallinity and stoichiometry. The thin film deposition occurs in an oxygen environment at a pressure of 0.2 mbar. The target is ablated using a laser fluence of 1 J/cm² with a pulse frequency of 10 Hz. The substrate where the thin film is placed is heated to around 600°C. This temperature is used to promote optimum crystallization of the deposited material in addition to preventing the formation of Bi and Na vacancies caused by their volatility. After deposition, the film was gradually cooled on high oxygen pressure to room temperature.

5.1 Structural characterization

XRD measurements were performed using a Rigaku-SmartLab X-ray diffractometer (Rigaku Corporation, Tokyo, Japan) with a conventional Cu anode X-ray tube, powered at 40 kV and 40 mA. The symmetric XRD scans corrected for the single-crystal surface miscut, 2θ ω scans, were performed with a Ge (220) monochromator in the incident beam and a HyPix detector in 0D mode, in the 2θ range 10–57°. The reciprocal space mapping (RSM) images were taken with a HyPix 3000 camera in 2D mode, around χ = 0 (symmetric to the sample surface), within 2θ = 35–51°, and χ = ±15°. The BNT-BT film is stabilized in tetragonal structure with highly c axis texture, as deduced from the 2ϴ-ω scan shown in Figure 14a. The SRO layer’s layer fringes are visible in the inset of Figure 14a, indicating the atomic scale level smoothness of its sides. The film thickness calculated from the layer fringes’ period is about 20 nm. Reciprocal space maps (Figure 14c) show two distinct types of c-domains with varying lattice constant c-perpendicular values on the substrate. The layers immediately above the substrate have a larger c value and an in-plane lattice constant a, close to the substrate one (named here “strained”), while the majority of the films have a tetragonal structure with a c value close to the bulk one (named here “relaxed”) [47]. The strained layer should be related to the thin layer visible in HRTEM images at the SRO-BNT-BT interface (Figure 15a). The in-plane orientation of the films has been analyzed performing azimuth scans in asymmetric geometry measurement locating tilt planes to the film surface.

Figure 14.

(a) XRD 2ϴ-ω scan; (b) phi-scans recorded in asymmetric geometry for the STO substrate, the SRO, and film and BNT films; (c) RSMs around nodes {103} of pseudocubic structures.

Figure 15.

(a) Cross-section TEM image at low-magnification showing BNT-BT/SRO/STO film, inset: SAED pattern corresponding to the TEM image; (b) HRTEM image at the BNT-BT/SRO and SRO/STO interfaces; (c) periodical defects observed in STEM image analyzed by EELS-SI.

The {103} planes were used for STO and SRO, and the {113} plane family for the assumed tetragonal BNT film and were scanned over a whole rotation around the surface normal (Figure 15b). The relative positions of the phi-scan maxima of the three structures indicate that the square edge of BNT is parallel to the face diagonal of SRO or STO: BNT [100] // SRO [110] //STO [110]. This in-plain orientation is determined by the almost perfect matching of the periods in these directions: aBNT ≈ aSTO.

Cross-section transmission electron microscopy (TEM) specimen of BNT-BT has been prepared from the as-deposited sample by mechanical polishing down to ca. 30 μm, followed by ion milling in a Gatan PIPS machine at 4 kV accelerating voltage and 7° incidence angle. Low-voltage (2 kV) milling was used as the final ion polishing stage in order to reduce the amorphous surface layer enveloping the specimen. TEM observations have been performed using a probe-corrected analytical high-resolution JEM ARM 200F electron microscope operated at 200 kV. The conventional TEM image in Figure 15 a gives a general cross-section view of the BNT-BT/SRO/STO film showing the growth morphology and the layer’s distribution on the (100) STO substrate. The SRO layer is visible as a narrow band of uniform contrast and thickness (20 nm) on the top of the STO substrate. The BNT-BT layer has a thickness of 250 nm with a surface roughness below 5 nm, as displayed by the cross-section image. A strong diffraction contrast can be observed across the BNT-BT layer either parallel or perpendicular to the growth direction, attesting the elastic strain due to various growth defects. The epitaxial growth of the BNT-BT and SRO layers onto the STO substrate is demonstrated by the selected-area electron diffraction (SAED) pattern in the inset of Figure 15a, where the STO substrate is in [100] zone axis orientation. The diffraction pattern (DP) contains contributions from the substrate, the SRO electrode, and the BNT-BT layer. A single family of spots is visible, the 1.3% lattice mismatch being too small to produce evident spot splitting, especially for the low-index spots. The observable deformation of the diffraction spots, which can be noticed as streaks parallel to the (001) STO and (010) STO reciprocal vectors, is most probably related to the high density of defects in BNT-BT oriented perpendicular to the interface (next to the interface) or parallel to it (away from the interface). The diffraction spots attributed to BNT-BT have been indexed according to the tetragonal structure of BNT-BT in agreement with the phase diagram for this stoichiometry at room temperature [47]. Figure 15c presents the cross-sectional HAADF-STEM image at low magnification of the BNT-BT/SRO/STO heterostructure where the spectrum image for EFLS was selected, followed by a relative composition maps in false colors for Ti L and Ba M extracted from EELS-SI data cube and RGB composed image by overlapping the Ti and Ba elemental maps. The Aurivillius phase was also identified, and their structure is built by alternating layers of [Bi₂O₂]2+ and pseudo-perovskite blocks, with perovskite layers that are n (5 in our case) octahedral layers in thickness [48, 49]. In TEM images recorded on the BNT-BT thin film, the presence of different types of defects such as Aurivillius phase, Barium precipitates, grain boundaries, and dislocations was observed.

5.2 Stored/released energies and the efficiency of BNT-BT

The physical properties of a BNT-BT capacitor, such as the area of the electrodes, thickness, dielectric constant, and leakage, determine the amount of charge that it can store and, thus, its capacity. In Figure 16, a PUND hysteresis measure at 300 K is presented. The maximum polarization at the maximum voltage applied is 45 μC/cm². But, the remnant polarization is 10 μC/cm². This represents the capacitor based on BNT-BT thin film that presents a significant back-switching of polarization when the voltage drops to zero. This property can be used for the application of storage and fast discharging of electrostatic energy in supercapacitors. Figure 16c represents the modification of polarization-voltage characteristics obtained for second consecutive positive voltage pulse for temperature from 100 to 400 K. As can be seen in Figure 16b and d, the total polarization, represented here as the area under the switching peak from Figure 16b, has a small decreasing with temperature, but the back-switching component is increasing with temperature.

Figure 16.

(a) Polarization-voltage characteristics obtained with PUND measurement. The back loops represent the results from the first and the third pulse, while the red loops represent the results from the second and the fourth pulse, (b) the current-voltage characteristics obtained during the first pulse for different voltages, (c) the polarization-voltage loop measured obtained during the second positive pulse for different temperature, and (d) the current-voltage characteristics obtained after the second positive pulse for different temperatures.

Thus, the polarization measured by the second consecutive positive pulse increases with temperature as is explicitly represented in Figure 16c in polarization-voltage loops or in Figure 16d by current-voltage loops.

From characteristics presented in Figure 16, the stored and released density of energy and also the efficiency are deduced using the formulas:

$$E_{charge}=\underset{0}{\overset{Pmax}{\int }}EdP$$E1

$$E_{discharge}=\underset{Pmax}{\overset{Prem}{\int }}EdP$$E2

$$\eta =\frac{E_{discharge}}{E_{charge}}$$E3

The stored and released energies and also the efficiency for temperatures between 100 and 400 K are represented in Figure 17a, for the first pulse, and also in Figure 17b, for only the second pulse, which represents the back-switching. The total stored charge is between 25 and 22 J/cm³ for the first pulse and around 22 and 17 J/cm³ for the second pulse. The discharged energy presents a small increasing with temperature between 10 and 11 J/cm³ for both first and second pulse. Thus, efficiency also presents a small increasing by increasing the temperature. Around 60% efficiency is obtained for the second pulse and smaller value of 45% for the first pulse. In comparison to earlier publications, the E charge and E discharge values achieved in this work are higher than those found on bulk ceramics or polycrystalline films [6, 7, 8, 9]. We also found that the characteristics of epitaxial BNT-BT thin films remain constant across a wide temperature range from 100 to 350 K.

Figure 17.

The stability of the stored and released energies and also the effiencies for temperatures between 100 and 400 K; (a) specific for the first pulse; (b) specific for the second pulse.

6. Conclusions

This chapter presents the results found in epitaxial ferroelectric materials, either lead-based or lead-free, obtained by the pulsed laser deposition technique. An overview of the deposition and growth strategies employed to achieve high-quality epitaxial ferroelectric structures with well-defined interfaces is provided initially. Developing lead-based epitaxial thin films with variable thickness and their role in clarifying the fundamental issues of ferroelectric thin films are presented. In the final section of this chapter, we describe the growth of a lead-free epitaxial thin film, followed by its characteristics in the energy storage domain.

Acknowledgments

This research was funded by Core Program, component project PN23080202 and PN23080303, funded by Romanian Ministry of Research, Innovation and Digitization. The fee for open access publication was supported from the project 35PFE/2021, funded by the Romanian Ministry of Research, Innovation and Digitization. D. Popescu acknowledges the received funding from the PN-III-P1-1.1-TE-2021-0136/TE-2021-1053/RaBit.

Conflict of interest

The authors declare no conflict of interest.