Confocal Raman Microscopy Can Make a Large Difference: Resolving and Manipulating Ferroelectric Domains for Piezoelectric Engineering

Abstract

Ferroelectric materials exhibit spontaneous and stable polarization, which can usually be reoriented by an applied external electric field. The electrically switchable nature of this polarization is at the core of various ferroelectric devices . The following chapter will review the use of Confocal Raman Microscopy (CRM) for the advanced characterization of Lead-free piezoceramics , emphasizing the relationship between structure and piezoelectric properties. In the first part we give a general introduction on the background of the Potassium-Sodium Niobate (KNN ) Lead-free piezoceramics for elucidating the phase-structure-performance relationships through a classical approach of Raman spectroscopy. In the second part we reveal the role of the formation of secondary phases in KNN piezoceramics by chemical modifications. This phenomenon is one of the major drawbacks identified in KNN-based systems. In the third part we highlight some practical aspects in the study of ferroelectric domains in KNN-based piezoceramics. Special emphasis is placed on coupling confocal Raman spectroscopy and Atomic Force Microscopy (AFM), demonstrating to being a powerful tool in spatially resolving the structure of ferroelectric domains. Finally, we focus on the light-matter coupling and, consequently, in the relationships between phase boundaries and piezoelectric activity, discuss some existing challenges, suggest possible methods for further improving piezoelectricity, and provide some conclusions.

1 Raman Spectroscopy and Confocal Raman Imaging: A Very Useful Technique for the Quick Evaluation of the Structure and the Properties of Lead-Free Piezoceramics

1.1 A Classical Approach of Raman Spectroscopy for In Situ Monitoring of Structural Changes

Polymorphic phase boundaries (PPB) in Lead-free piezoelectric materials has attracted significant interest in recent years, in particular because of the unique properties that can be found in their vicinity. It is very difficult to determine the actual phase compositions and types of phase boundaries in a given material. Significant attention has been paid to exploring the origin of high piezoelectricity in Lead-free perovskite oxides with a PPB, by analyzing the phase constituents using different experimental methods.

The (K$_{0.44}$Na$_{0.52}$Li$_{0.04}$)(Nb$_{0.86}$Ta$_{0.10}$Sb$_{0.06}$)O$_{3}$ (KNL-NTS ) compound is used in this section as a model system to identify the existence of the intermediate phases and thus explain the physical origin for enhanced piezoelectricity around a PPB. KNL-NTS ceramics have been extensively studied as it is one of the most workable Lead-free compositions known to date [1]. The crystallographic structure of the KNL-NTS evolves from a paraelectric cubic (C) phase at high temperatures to a ferroelectric orthorhombic (O) phase at low temperatures, passing through a ferroelectric tetragonal (T) phase. The T to O phase transition gives rise to the PPB, which occurs close to room temperature for this compound (see Fig. 22.1). As a result of the PPB existence, excellent functional properties are achieved in KNL-NTS at room temperature [2]. The PPB temperature can be modulated by chemical modifications, such as the variation of the A/B ratios of the ABO$_{3}$ perovskite structure (see panel (a) of Fig. 22.1, where the unit cell of the perovskite structure is shown, in which two different symmetries coexisting at room temperature in the system are illustrated; orthorhombic (O) and tetragonal (T), respectively).

Fig. 22.1

Fast evaluation of the structure and the properties of Lead-free piezoceramics by Raman spectroscopy: a Unit cell of the perovskite structure with two different symmetries; orthorhombic (O) and tetragonal (T). b XRD diffraction pattern of the structure. c An average typical Raman spectrum of the KNL-NTS ceramic. d Polarization-field response of the for KNL-NTS ceramics with different A/B ratios. e Evolution of the remnant polarization, Pr, and the ionic effective displacement, $\varDelta$z$_{\mathrm{eff}}$

Previous results in KNL-NTS piezoelectric ceramics [3,4,5,6,7] showed that first, the best piezoelectric properties are obtained for the tetragonal symmetry, and second, there is a linear correlation between the tetragonal distortion, c/a ratio, and the piezoelectric properties (see Fig. 22.1e). The amount and distribution of the orthorhombic phase limits the correlation between tetragonality ratio and piezoelectric properties. In addition, obtaining such relationships between the structure and the piezoelectric properties will consequently allow us to understand the parameters leading to the improvement of the properties of these materials. Previous studies showed that not every conventional characterization technique appears suitable for that purpose, yet Raman spectroscopy has already proven to be. In the past, this technique was mainly devoted to fundamental research, but instrumental progress (laser miniaturization, CCD, filters and data processing software improvement) have rendered it a general characterization method. Not only can it provide basic phase identification [8] but also subtle spectral alterations can be detected and used to assess nano-scale structural changes [9] and characterize micromechanical behavior. Some specific features can also be used to study charge transfer [10, 11], film orientation [12] or the size of particles or clusters trapped in nano-cavities [13]. Moreover, in our particular case, Raman spectroscopy is a very sensitive tool to study at a local scale the structural deformation of perovskites, which are induced both by the tilting of the BO$_{6}$ octahedra and by the cationic displacements.

Here, we have selected the (K$_{0.44+x}$Na$_{0.52}$Li$_{0.04}$)(Nb$_{0.86}$Ta$_{0.10}$Sb$_{0.04}$)O$_{3}$ compositions with x = −0.06, 0.00 and 0.04, with the aim of modifying the PPB and consequently the system properties. This way, the ceramics present different A/B ratios in the ABO$_{3}$ perovskite structure, ranging from 0.94 to 1.04. For the sake of clarity, the compositions with A/B = 0.94, 1.00 and 1.04 will be hereafter named respectively as A/B < 1, A/B = 1 and A/B > 1. Three different A/B ratios were studied because previous work demonstrated that these values influence the sinterability of the material [14] and could also produce modifications in the crystalline structure appearing at room temperature [3].

A classical technique to obtain structural information on lead-based piezoceramics with PPB is X-ray diffraction (XRD). The top of Fig. 22.1b represents a sketch of a typical x-ray diffraction equipment. Figure 22.1b shows a typical XRD pattern for a KNL-NTS ceramic, which has a main tetragonal crystal structure with a space group of P4mm. The inset of Fig. 22.1b shows a detail of the XRD diffraction pattern in the 2$\varTheta$ range 44$^\circ$–47$^\circ$, corresponding to (002) and (200) peaks of the perovskite structure. It is well known that the tetragonal symmetry (T) of the perovskite phase can be deconvoluted in two Lorentzian peaks, (002)$_{\mathrm{T}}$, and (200)$_{\mathrm{T}}$. In addition, in this sample two peaks located at $\sim$45.4 and $\sim$45.6$^\circ$ (2$\varTheta$) which are associated with the orthorhombic symmetry (O), are detected. The presence of peaks indicating tetragonal as well as orthorhombic symmetry is evidence for a PPB. In this model case, the peaks associated with tetragonal symmetry are the more relevant ones. In tetragonal P4mm KNL-NTS ceramic, ferroelectricity is mainly due to a relative displacement along the c-axis of the B-cation from its centrosymmetric position in the unit cell and in the oxygen octahedra. In correlation with the B-cation shift, A-ions are also displaced along the fourfold axis of the tetragonal cell. A permanent electric dipole , or ferroelectric polarization , is thus created (blue arrow in Fig. 22.1a). The deviation of the c/a ratio from unity is used as an indication of the presence of the ferroelectric phase. Therefore, we feel that determination of the c/a ratio is a crucial parameter for predicting the functional properties of the ferroelectric ceramics.

In X-ray diffraction, the samples showed an increase of the tetragonality ratio with sintering time and A/B ratios. For the A/B > 1 sample the XRD peak broadening may be caused by the phases’ coexistence as well as a low degree of crystallinity. It therefore may be difficult to fully resolve the tetragonality ratio. In contrast, these disadvantages can be minimized using Raman spectroscopy. Raman spectroscopy is a very sensitive tool to study at a molecular scale the structural deformations of perovskites, which are induced both by the tilting of BO$_{6}$ octahedra and by the cationic displacements. These structural modifications induce significant changes in internal modes associated with the BO$_{6}$ octahedron and thus a modification of the Raman spectra. The main vibrations are associated with the BO$_{6}$ perovskite-octahedron [7, 15]. In that case, the vibrations of the BO$_{6}$ octahedron consist of 1A$_{1\mathrm{g}}$($\nu _{1}$) + 1E$_{\mathrm{g}}$($\nu _{2}$) + 2F$_{1\mathrm{u}}$($\nu _{3}$, $\nu _{4}$) + F$_{2\mathrm{g}}$($\nu _{5}$) + F$_{2\mathrm{u}}$ ($\nu _{6}$) modes. Figure 22.1c shows an average typical Raman spectrum of the Raman modes associated to the BO$_{6}$ octahedron for the KNL-NTS ceramic. Of these vibrations, 1A$_{1\mathrm{g}}$($\nu _{1}$) + 1E$_{\mathrm{g}}$($\nu _{2}$) + 1F$_{1\mathrm{u}}$($\nu _{3}$) are stretching modes and the rest, bending modes. In particular, A$_{1\mathrm{g}}$($\nu _{1}$) and F$_{2\mathrm{g}}$($\nu _{5}$) are detected as relatively strong scattering in systems similar to the one we are studying because of a near-perfect equilateral octahedral symmetry, see insert of the Fig. 22.1c. Raman spectroscopy has been used to evidence the nature of the crystalline symmetry of the different compositions as function of the stoichiometry (A/B ratio) for both the synthesis powders [16, 17] and the sintered samples [3].

Following the above analysis, we concluded that ferroelectricity is mainly due to a relative displacement along the c-axis of B cation from its centrosymmetric position in the unit cell and in the oxygen octahedra. This relative displacement produces a spontaneous and stable polarization, which can usually be reoriented by an applied external electric field. To verify the ferroelectric behavior at the macroscopic level, a common procedure is to assess the electric field-induced polarization hysteresis loops. Figure 22.1d¹ shows room temperature P-E hysteresis loops of the KNL-NTS ceramics for different A/B ratios at 1 (top) and 16 h (bottom) sintering time, respectively. Well-saturated hysteresis loops with a good square shape are clearly obtained in all samples and an increase in remnant polarization, Pr, with the sintering time is observed for each composition. Moreover, in our case the remnant polarization is shown to be proportional to the tetragonality ratio, as evidenced in Fig. 22.1e.²

The most surprising result obtained from Raman spectroscopic investigations is that the evolution of the A1g Raman mode shift is equivalent to the one of the tetragonality ratio calculated from XRD data, as shown in Fig. 22.1e. More important, this fact allows us to gather valuable information about the ionic effective displacement [7, 18], $\varDelta$z$_{\mathrm{eff}}$, of the active ferroelectric ion (top of Fig. 22.1e), because a larger Raman shift of the A$_{1\mathrm{g}}$ mode or a larger tetragonality ratio (c/a) implies a higher value of displacement of the central ion (B$^\text {5+}$) from its equilibrium position, $\varDelta$z$_{\mathrm{eff}}$. In addition, the spontaneous polarization (P$_{\mathrm{s}}$) is directly proportional to piezoelectric d$_{\mathrm{ij}}$ coefficient and therefore a linear relationship between the piezoelectric d$_{\mathrm{ij}}$ coefficient and the tetragonality ratio or the wavenumber of the A$_{\mathrm{1g}}$ mode is also expected, as shown Fig. 22.1e (bottom).

To summarize the first section, Raman spectroscopy can be used to rapidly determine the tetragonality ratio (c/a) of KNN-based ceramics, since Raman spectra can be taken in few seconds, in contrast to several minutes required for an XRD measurement. Moreover, the quantity of sample required for a Raman measurement is significantly smaller than for an XRD measurement, thus being then a very useful technique for material testing also in industrial processes. And last but not least, Raman spectroscopy has allowed us to establish relationships between the ferroelectric properties (Pr), piezoelectric properties (d$_{\mathrm{ij}}$) and the ionic effective displacement ($\varDelta$z$_{\mathrm{eff}}$) of the active ferroelectric ion with the wavenumber of the A$_{\mathrm{1g}}$ mode (Fig. 22.1e). Knowing and understanding these relationships, Raman spectroscopy can be applied to determinate these parameters and be used as a powerful tool for in situ monitoring their changes under different processing conditions.

1.2 Raman Imaging: Can It Make a Significant Difference?

The study of a complex system, such as piezoceramic materials, in general is somewhat problematic since there are two main contributions to the macroscopic properties in such piezoelectric materials: the intrinsic and extrinsic contribution. The intrinsic contribution is associated with the crystalline structure, lattice parameter evolution, phase boundary, etc. The extrinsic contribution is mainly associated with the microstructural aspects (i.e. grains, grain boundaries, domains, domain wall motion, etc.). In order to obtain a best possible understanding of the a real-world-sample, both the intrinsic and the extrinsic contributions have to be taken into account. The intrinsic contributions are mainly the tilting of BO$_{6}$ octahedra and the cationic displacements at microscopic scales. The extrinsic contributions originate mainly from microstructural features. Confocal Raman Microscopy (CRM) allows us to acquire the microstructural features (extrinsic contribution) at a high spatial resolution and the molecular information at excellent spectral resolution (intrinsic contribution). Besides, with the relatively recent advances in confocal Raman imaging, the identification, distribution, and quantification of all crystalline phases within a ceramic material is now possible at resolution scales ranging from centimeters to nanometers.

As previously mentioned, the identities of crystalline phases play a relevant role with respect to the functional properties of the piezoceramics, and therefore, their identification and distribution can be used to devise new strategies in designing new materials with improved properties. Figure 22.2a shows an optical micrograph of the polished surface of a KNN-based ceramic, aligned perpendicular to the Raman laser (more information about the composition and preparation of this material can be found in [19]). The area of $20\times 20\,\upmu$m denotes the selected area where the Raman spectra were collected at a plane located just below the surface of the sample, where the Raman intensity was maximized. The acquisition time for a single Raman spectrum was 500 ms, thus the acquisition of a Raman image consisting of $100\times 100$ pixels (10,000 spectra) required 83 min. Features such as Raman peak intensity, peak width or Raman shift can be extracted from recorded Raman spectra using fitting algorithms, such as Gaussian or Lorentzian fits. These can then be used for comparison and display of the information as for example shown in Fig. 22.2b.

Fig. 22.2

Reprinted with permission from [19] Copyright 2013 AIP Publishing LLC

Fast identification of the phase coexistence of Lead-free piezoceramics by Confocal Raman microscopy: a Optical micrograph of the polished surface of a KNN-based ceramic. b The Raman image of the KNN-based sample exhibiting the domains structure of the ceramic. The colours of the Raman image correspond to the colours of the spectra in (c–e). The regions delimited by bright dotted lines and marked as A, B, C and D represent four different grains. In addition, a secondary phase in yellow is marked with an asterisk ($*$), which will be discussed in the Sect. 22.2. c–e Main Raman spectra associated with the different colours in (b). The inserts show magnified Raman spectra fitted with two Lorentzian peaks, assigned to the E$_{\mathrm{g}}$ ($\nu _{2}$) and A$_{\mathrm{1g}}$ ($\nu _{1}$) Raman modes, respectively.

Raman spectra having the same Raman shift are classified by colours and the colour intensity corresponds to the Raman intensity of the respective peak. The combination of colours results in an image of the ceramic microstructure, which reveals the presence of both ceramic grains (delimited with a bright dotted line for shake or clarity) and the groups of unit cells with the same orientation (and same crystalline phase), which are called ferroelectric domains. From the colour combination, four grains can be observed (marked as A, B, C and D in Fig. 22.2b). These four grains provide different scenarios to study the domain structure (or crystalline phase coexistence) in polycrystalline samples. Grains A and B are characterized with striped domains while grains C and D are irregularly shaped domains. The colours in the image (Fig. 22.2b) correspond to the colours of their spectra: Fig. 22.2c tetragonal phase (blue); Fig. 22.2d mixing of tetragonal and orthorhombic phases (green) and Fig. 22.2e orthorhombic phase (red), confirming the phase coexistence that previously were observed by Raman spectroscopy at macroscopic scales (Fig. 22.1).

As a relevant result, grains C and D show irregular domains with the Raman spectra mainly characteristic of an orthorhombic phase. In addition, grain D shows small isolated regions inside orthorhombic grains, denoted in the literature as watermark domains [20, 21]. These watermark domains exhibit out-of-plane polarization. It is also worth noting further characteristics revealed by the CRM imaging since it has been possible to detect the presence of a secondary phase (marked with ‘$*$’ in Fig. 22.2b), which will be discussed in the following section.

2 The Study of (K,Na)NbO$_{3}$-Based Lead-Free Piezoelectric Ceramics: Identification of the Secondary Phase Location Using Confocal Raman Imaging

Studies from the very early stages and particularly those on polycrystalline (K,Na)NbO$_{3}$-based ceramics have been faced with difficulties in the preparation of the perovskite free of secondary phases. In this section we will reveal the role of the formation of secondary phases in KNN piezoceramics by their chemical modifications. This phenomenon is one of the major drawbacks identified in KNN-based systems. The secondary impurity phase appearance is commonly observed in many of the alkaline niobate based piezoceramics [4,5,6,7, 14, 22,23,24,25]. In particular, the secondary impurity phases are Li- and K-rich phases [4,5,6,7, 23,24,25]. As a consequence, we reported in previous studies that the chemical modifications with metal oxides (M$_{2}$O$_{\mathrm{n}}$) has an important impact on the perovskite structure [26,27,28,29,30,31]. The ionic radii of the M$_{2}$O$_{\mathrm{n}}$ modulate the degree of lattice distortion, whereas the doping concentration rules the formation of the tetragonal tungsten bronze (TTB) phase [31]. In view of the doping levels, we have demonstrated that the inherent solubility of M$_{2}$O$_{\mathrm{n}}$ in the A- or B-sites position of the perovskite is very limited. When the amount of M$_{2}$O$_{\mathrm{n}}$ is higher than x = 0.01, M$^\text {n+}$ ions are supersaturated in the lattice of KNL-NTS, and the excess M$^\text {n+}$ ions enter on the B-sites of the lattice and favor the formation of the TTB secondary phase. This behavior is associated with a compositional segregation of the alkaline elements during the sintering stage [23, 24, 32]. Moreover, the secondary phase on these ceramics is hard to detect, implying that it could be an amorphous phase formed during sintering, with a high solubility in the system that led to its eventual disappearance with sintering time and/or sintering temperature [5, 33]. By X-ray diffraction, amorphous phases and/or crystalline phases with low concentrations (less than 2% of the main phases) appear as background signal without peak resolution, which complicates separation and localization of the components in the mixture. Nonetheless, in Confocal Raman Microscopy, amorphous phases, crystalline phases with a low concentration, and crystalline main phases often exhibit spectra with unique features. This allows signal separation, and hence, it will help to determine the role of each component for the functional properties of the system.

In the present section, W$^\text {6+}$ is selected as the dopant of KNL-NTS ceramics. On the basis of its ionic radii [34], the W$^\text {6+}$ ion (r$_{\text {w}^{6+}}$: 0.60Å for a coordination number CN = 6) falls in the size range of the B-site position (r$_{\text {Nb}^{5+}}$: 0.64Å, r$_{\text {Ta}^{5+}}$: 0.64Å, r$_{\text {Sb}^{5+}}$: 0.60Å CN = 6). Considering its valency, W$^\text {6+}$ can act as a donor-dopant if introduced in the B-site. In order to introduce the W$^\text {6+}$ ion into the B-site of the perovskite lattice, we selected B-site deficiency with a global formula (K$_{0.44}$Na$_{0.52}$Li$_{0.04}$)[(Nb$_{0.86}$Ta$_{0.10}$Sb$_{0.04}$)$_{1-\mathrm{x}}$W$_{5\mathrm{x}/6}$]O$_{3}$, hereafter abbreviated as KNL-(NTS)$_{1-\mathrm{x}}$W$_{5\mathrm{x}/6}$ [35].

To verify that the secondary phase formation plays a relevant role in compositional segregation of the alkaline elements during the sintering process, experiments were performed by atomic force microscope (AFM) and CRM. The ceramic with higher W$^\text {6+}$ contents, x = 0.05, was chosen to determine the secondary phase appearance. Figure 22.3a depicts an optical microscopy image of the sample with x = 0.05, aligned perpendicularly to the AFM cantilevers. The area of 14 $\times$ 14 $\upmu$m (Fig. 22.3a) delimits the range where topographic information was collected by AFM. Figure 22.3b shows a detailed AFM topographic image of two grains corresponding to the secondary phase structure with plate-like shape. The AFM scans along the white arrow of Fig. 22.3b is illustrated in Fig. 22.3c. The grain associated with the secondary phase (i) has a grain size of ${\sim }4$ $\upmu$m and (ii) appears close to grain boundary protrusions (height difference of ${\sim }400$ nm).

Fig. 22.3

Modified and reprinted with permission from [35]. Copyright 2015 Royal Society of Chemistry

Detection of the secondary phase through confocal Raman microscopy: a Optical micrograph of the polished and thermally etched surface of the KNL-(NTS)$_{1-\mathrm{x}}$W$_{5\mathrm{x}/6}$ sintered ceramics with x = 0.05. b AFM image of KNL-(NTS)$_{1-\mathrm{x}}$W$_{5\mathrm{x}/6}$ sintered ceramics with x = 0.05, showing the topography of the secondary phase inside the marked white box of (a). c AFM topography cross section along the white arrow of (b). The white rectangle in the panel (a) shows the positions where the XY Raman image is performed and corresponds with the area of previous AFM analysis. The Raman image shows the secondary phase distribution (blue regions) at the surface by colour code (d) as well as in the depth scan (e). The white arrow in the panel (d) shows the position where the XZ Raman depth scan image is performed. f Experimental Raman spectra corresponding to a KNL-NTS matrix and a minor secondary phase. This spectrum is fitted to the sum of two spectra: spectrum of the matrix (g) and of the secondary phase (h). In addition, the isolate Raman spectrum of the secondary phase has been magnified 10 times due to its low Raman intensity compared to the Raman spectrum of the KNL-NTS ceramic (matrix). The inserts in (f–h) show magnified Raman spectra in the frequency range 450–750 rel. cm${^{-1}}$.

CRM is here combined with AFM in the same experimental setup, thus giving direct correlation between topography and local structure. The selected area is the one previously studied by AFM. As before, Raman spectra having the same Raman shift are classified by colours and the colour intensity corresponds to the Raman intensity of the respective peak. The colour combination results in (i) Raman image of the surface (Fig. 22.3d) and (ii) Raman depth scan image of the cross-section (Fig. 22.3e). From these results we can deduce that there is a clear correlation between the secondary phase region (marked in blue colour) and the protrusion evidenced by AFM. This phenomenon results from the partial retention of the alkali elements (Li, Na, K) induced by the secondary phase formation at high temperatures during the sintering process, which promotes the displacement of the alkaline elements from the main KNL-NTS phase (red region in Fig. 22.3d–e) to the new secondary phase (blue region in Fig. 22.3d–e). As a consequence, we can conclude that the partial retention of the alkaline elements (Li, Na, K) onto the secondary phase, plays an important role in the modulation of functional properties of the system.

Following this section, there are several novel methodological approaches that may deserve further attention. CRM is particularly valuable for delivering information on both the (primary) matrix and the secondary phase. A detailed explanation of how we manage to discriminate the secondary phases, can be followed from Fig. 22.3f–h; the average Raman spectrum of the secondary phase is shown in Fig. 22.3f, which can be indexed on the basis of a phase mixture constituted by a majority of KNL-NTS phase and a minor effect of the secondary phase. The localization of the secondary phase can be observed in Fig. 22.3d, which is displayed as blue regions. As alluded earlier, this secondary phase is hard to detect by corresponding X-ray diffraction pattern. In previous studies concerning the KNL-NTS compounds, we have demonstrated that the secondary phase could be assigned either to K$_{3}$LiNb$_{6}$O$_{17}$ (KLN) or K$_{6}$Nb$_{10.88}$O$_{30}$ (PDF$\sharp$87-1856), both with tetragonal tungsten-bronze type structure (TTB) [5, 26, 31].

We have performed additional analysis to determine the isolated Raman spectrum of the secondary phase. The Raman spectrum discrimination of the secondary phase was calculated from average Raman spectra of KNL-NTS ceramic associated with the secondary phase (Fig. 22.3f). Then, the spectrum is fitted to the sum of two spectra: the first is associated with the matrix Raman spectrum (Fig. 22.3g), as it is well known for the piezoelectric KNL-NTS perovskite systems. The second spectrum is ascribed to the secondary phase with TTB structure (Fig. 22.3h).

To conclude this section, confocal Raman imaging allows the detection of sample characteristics, such as the presence of a secondary phase, which had previously been unobservable using XRD.

3 High Spatial Resolution Structure of (K,Na)NbO$_{3}$ Lead–Free Ferroelectric Domains

Lead zirconate titanate (PZT) based ceramics are currently enjoying wide use in piezoelectric devices despite lead toxicity. Due to growing environmental and human health concerns, the attention to piezoelectric ceramics has been moving to lead-free materials, in particular to potassium-sodium niobate-based lead-free piezoceramics [(K,Na)NbO$_{3}$-based, KNN]. Unfortunately, practical implementations of KNN ceramics for commercial use are still limited by their inferior electrical and electromechanical properties as compared to their conventional Lead Zirconate Titanate (PZT) counterparts.

Almost all recent efforts contributing to property enhancement in KNN ceramics have been concentrating on chemical optimization through doping [36, 37], in addition to control over the sintering process [36, 38] and domain engineering by control of the poling process [39, 40]. For improving of the properties of KNN, the major concern is the increase of the piezoelectric coefficient d$_{33}$, which is determined by the electric charge response to a low external mechanical stress under linear conditions. The low mechanical field applied in the d$_{33}$ measurement inhibited the extrinsic contribution of non-180$^\circ$ domain wall motions so the d$_{33}$ values are dominated by the intrinsic (lattice) piezoelectric responses. Thus, the study of domain walls in piezoelectric materials is of high interest for a better understanding of their properties. Only few papers dealing with the ferroelectric domains in KNN and related compositions, such as single crystals [41, 42] and polycrystalline ceramics [43,44,45], are reported in the literature and they have been performed mainly by using Piezo-Force Microscopy.

Most classical characterization techniques of the ferroelectric domain structure are destructive, since they require the modification of the surface of the ceramic materials. In addition, as it has been mentioned above, the functional properties of the ferroelectric materials depend not only on the structural modification of the lattice distortion (intrinsic contribution) but also on the fine details of their ferroelectric domain configuration (extrinsic contribution) and on their highly ordered architecture at a scale ranging from few nanometers to several microns. Most of these fine details are lost when the surface of the ceramic is modified to be examined by such classical techniques. In order to investigate the structure and distribution of ferroelectric domains, a number of techniques have been commonly applied, among them are scanning probe microscopy (SPM) [43], environmental scanning electron microscopy (ESEM) [46], optical microscopy (OM) [47], transmission electron microscopy (TEM) [48], atomic force microscopy (AFM) [48] and lately, scanning electron microscopy in the backscattered mode (SEM-BS) [44, 49]. In contrast to spectroscopy methods, the above mentioned methods yield no or very limited compositional information and pure knowledge about the topography of the samples is often insufficient for the comprehensive characterization of complex domain structures. Various attempts have been made to combine the high spatial resolution of scanning probe microscopy with chemical information provided by spectroscopic techniques. Methods based on micro Raman spectroscopy give the possibility to study at a local scale the structural deformations of perovskites, which are induced both by the tilting of BO$_{6}$ octahedra and by the cationic displacements, as it is known for piezoelectric PZT and PMN-PT³ perovskite systems [50, 51]. These structural modifications induce significant changes in internal modes associated with the BO$_{6}$ octahedron and thus a modification of the Raman spectra. Raman spectroscopy has been used to demonstrate the correlation of the structure and the piezoelectric properties of the materials, and to calculate the effective ionic displacement causing the piezoelectric polarization [7]. By determining both topography (microstructure) and molecular structure (chemical information) simultaneously, confocal Raman microscopy coupled with atomic force microscopy provide molecular-level information about the relationship of the crystalline structure with the domain formation as well with the polarization (Fig. 22.4). For a correct assessment, we need structural and topographic pieces that are at the core of potential technological applications.

As an example of a correct assessment of the ferroelectric domain structure, the Lead-free (K,Na)NbO$_{3}$ ceramic prepared by microwave–hydrothermal synthesis was chosen, because its grain size is much larger than commonly obtained by other methods, allowing us to perform studies on single crystals prepared by conventional ceramic method. The ferroelectric domain structure of this system in general is of interest because of its potential technological applications [52]. However, it is still much debated how ferroelectric domain structures affect the functional response of the material. Consequently this issue has been of considerable research interest in the last decades.

3.1 Simultaneous Determination of Topographic and Structural Features by CRM Coupled with AFM

Study of the domain structure was performed by Optical Microscopy and Atomic Force Microscope. Figure 22.4a depicts an optical microscopy image of the chemically etched KNN sample aligned perpendicularly to the AFM cantilevers. The domain structure is mainly composed of striped regions, vertically aligned in the image, containing in their interior 180$^\circ$ parallel domains. The domains in adjacent stripes present an angle of about 90$^\circ$, indicating that a 90$^\circ$ domain wall appears between the vertical stripes shown in Fig. 22.4a. The area of $40 \times 40$ $\upmu$m (marked as white box in the Fig. 22.4a) delimits the range where topographic information was collected by AFM. Figure 22.4b shows a detailed AFM topographic image of the domain structure. The topography analyses, Fig. 22.4c, confirm that the 90$^\circ$ domains (stripes) width ranged from 8 to 16 $\upmu$m (90$^\circ$ domain walls are marked by dashed lines). Meanwhile, inside each striped area the domain width is smaller and varies from one strip area to another. Inside the 2–3 area (Fig. 22.4b) the domains are $\sim$450 nm in width and $\sim$32 nm in height, Fig. 22.4d, in contrast with the area 3–4, where domains are $\sim$1.4 $\upmu$m in width and $\sim$100 nm in height, Fig. 22.4e. The domain density is thus different between stripe areas, indicating a relevant difference in the crystallographic orientations.

Fig. 22.4

Modified and reprinted with permission from [52]. Copyright 2012 Royal Society of Chemistry

Determination of both topographic and structure features simultaneously by CRM coupled with AFM: a Optical image of KNN ceramic after chemical etching, which shows a complex domain structure. b AFM image of the same sample showing the domain structure as consequence of the different chemical etching rate of the crystallographic orientations. c–e AFM line scans along the arrows marked in (b). c The AFM scan (green arrow) shows three regions separated by 90$^\circ$ domains walls, with a step height at these 90$^\circ$ domain walls of $\sim$400 nm. d, e Height profiles corresponding to the arrows marked with blue and red colour in the image shown in (b). f Average Raman spectra of adjacent striped domains separated by a 90$^\circ$ domain wall. The inserts show magnified Raman spectra and Lorentzian fits of domain structures in the frequency range between 500 and 700 rel. cm${^{-1}}$. These spectra are fitted to the sum of two Lorentzian peaks, ascribed to the E$_{\mathrm{g}}$ ($\nu _2$) and A$_{\mathrm{1g}}$ ($\nu _1$) Raman modes, respectively. g Raman image of domains structure of the KNN exhibiting clear differences between average spectra of adjacent striped domains separated by a 90$^\circ$ domain wall. The Raman image was derived by integration of the total spectral intensity from 100 to 1000 rel. cm${^{-1}}$.

As we have highlighted in this section, both topographic and structural (chemical) information must be taken into account simultaneously to understand the nature of the micrometric domains in KNN ceramics.

CRM is here combined with AFM in the same apparatus, thus giving direct correlations between topography and local structure. The Raman spectra are collected at a plane located just below the surface of the sample where the Raman intensity is maximized (because CRM is very sensitive to surface topography and large differences in signal intensity could be originated from nanoroughness). The selected area ($60 \times 40$ $\upmu$m) is the one previously studied by optical microscopy (Fig. 22.4a). The acquisition time for a single Raman spectrum was 1.5 s/pixel (Fig. 22.4f). Thus the Raman image consisting of $60 \times 40$ pixels (2400 spectra) required 60 min for the planar-section (Fig. 22.4g). Features such as Raman peak intensity, peak width or Raman shifts are fitted with algorithms in order to compare information and to represent the derived Raman image. The assignments of the observed Raman modes, both symmetry and nature (first and second order), are summarized in the first part of this chapter. Figure 22.4f shows the average Raman spectra obtained in two adjacent strip areas separated by a 90$^\circ$ domain wall. Clear differences can be observed on the shape of both spectra, particularly in the 600 rel. cm${^{-1}}$ region. A detail of the BO$_{6}$ octahedron stretching E$_{\mathrm{g}}$ ($\nu _2$) and A$_{\mathrm{1g}}$ ($\nu _1$) Raman modes of the perovskite is magnified in the respective inserts in Fig. 22.4f and fitted to the sum of two Lorentzian peaks. The A$_{\mathrm{1g}}$ ($\nu _1$) Raman mode located at $\sim$612 rel. cm${^{-1}}$ is a symmetric mode so it is observable for the different crystal orientations and is usually an intense mode, while the E$_{\mathrm{g}}$ ($\nu _2$) Raman mode at $\sim$550 rel. cm${^{-1}}$ is an anti-symmetric Raman mode along the polar direction (perpendicular to the oxygen octahedral basal plane). The E$_{\mathrm{g}}$ ($\nu _2$) Raman mode is less intense than the A$_{\mathrm{1g}}$ ($\nu _1$), but is magnified when the Raman polarization is parallel to in plane polarization. Then, it can be said that the polar direction is parallel to the sample plane surface in those regions (in red colour) of the sample presenting the red spectrum shown in Fig. 22.4f, while the blue spectrum represent a ferroelectric domain region with out-of-plane polarization.

In Fig. 22.4g, a colour-coded image indicates sample regions where the Raman spectra correspond to the ones presented in Fig. 22.4f (red and blue colour regions respectively). Raman spectra having same Raman shift are classified by colours and the colour intensity corresponds to the Raman intensity. As can be seen in this image, adjacent stripes present different Raman spectra. The studied Raman modes are certainly coupled and the determination of crystal orientation in adjacent striped regions cannot be determined unequivocally with the above measurement only. However, this measurement shows that there are relevant differences related to the polarization direction in the stripe regions corresponding to alternations of in-plane and out-of-plane orientations. This allows the confirmation of a 90$^\circ$ domain wall between adjacent stripe regions, as previously expected based on the topography. This example demonstrates the sensitivity and applicability of the Confocal Raman technique for detecting the presence of domains at the micro-scale.

3.2 Insights into the Details of the Ferroelectric Domain Structure

In addition to the domain identification, CRM allows a determination of the nature of domain walls and correlation between the structure and piezoelectric properties. Raman spectra have been measured following a line crossing perpendicularly the 180$^\circ$ domains in a stripe region where in-plane polarization is predominant (region marked in red in Fig. 22.4f and corresponding to red spectrum, see also the marked blue line in Fig. 22.5). In Fig. 22.5b, c we show the BO$_{6}$ octahedron Raman modes in two adjacent 180$^\circ$ domains, marked by A and B points in Fig. 22.5a. The differences in E$_{\mathrm{g}}$ and A$_{\mathrm{1g}}$ Raman modes intensity presumably account for different polar orientations, while the variations in the Raman shift of modes associated to the BO$_{6}$ octahedron allow determining changes in polarization, which are associated with modification of the constant force of the octahedron due to deformation or stress. It is worth noticing that there are two characteristic aspects in the E$_{\mathrm{g}}$ and A$_{\mathrm{1g}}$ Raman shift evolution along the line crossing the domains. First, the peaks shift in different directions (a polarization change), probably attributed to differences in polarization orientation, which is compensating between both modes. Second, the Raman shift amplitude in the E$_{\mathrm{g}}$ Raman mode is nearly 4 times higher than the one in the A$_{\mathrm{1g}}$ Raman mode. These evidences indicate that, in addition to the polar orientation change, the value of cell distortion is also modified and therefore the polarization changes (in value as well as in orientation). The alternating polarization differences in 180$^\circ$ domains tetragonal templates are resolved by 60$^\circ$ and 120$^\circ$ orthorhombic domains, which resulted in T-O alternating domains, coexistent in the in-plane-polarization region.

Fig. 22.5

Reprinted with permission from [52]. Copyright 2012 Royal Society of Chemistry

Characterization of the 180$^\circ$ domain walls by confocal Raman spectroscopy: a Optical microscopy image of KNN ceramic samples, with two domains separated by a 90$^\circ$ domain wall. The 180$^\circ$ domain walls under analysis are present at a 45$^\circ$ angle to the 90$^\circ$ domain wall; note that the 180$^\circ$ domains located on the left of the 90$^\circ$ domain wall fail to preserve the symmetry and the angle is slightly lower than 45$^\circ$. The width of the 180$^\circ$ domains is in the range of 750–1500 nm. b, c Magnified Raman spectra and Lorentzian fits of domain structures in the frequency range between 500 and 700 rel. cm${^{-1}}$ corresponding to the points labelled (A) and (B) in the image shown in (a). These spectra are fitted to the sum of two Lorentzian peaks, ascribed to the E$_{\mathrm{g}}$ ($\nu _2$) and A$_{\mathrm{1g}}$ ($\nu _1$) Raman modes, respectively. d Evolution of the A$_{\mathrm{1g}}$ and E$_{\mathrm{g}}$ modes which were measured following the blue arrow marked in the panel (a).

3.3 Some Clues About the Origin of the Ferroelectric Domain Structure: The Stress Sensitivity of Raman Spectroscopy

As mentioned above, at room temperature KNN presents a crystallographic structure composed of coexisting orthorhombic (O) and tetragonal (T) phases, a phenomenon known as Polymorphic Phase Boundary (PPB). The high temperature tetragonal structure formed at the T$_{\mathrm{c}}$ (cubic-tetragonal transition) acts as domain template of the room temperature orthorhombic structure. The change of the polarization direction of the orthorhombic structure implies a crystal lattice strain that is governed by the existing tetragonal domain configuration. The internal stress results in the appearance of an unusual polarization relaxation at the 90$^\circ$ domain wall in (K,Na)NbO$_{3}$ ceramics (Fig. 22.6).

Fig. 22.6

Reprinted with permission from [52]. Copyright 2012 Royal Society of Chemistry

Stress analysis around complex domain structures: a Color coded Raman image showing the Raman shift of the A$_{\mathrm{1g}}$ mode. b Evolution of the A$_{\mathrm{1g}}$ mode Raman shift following the arrow marked in (a). c Statistical analysis of the number of spectra corresponding to different Raman shift values of the A$_{\mathrm{1g}}$ mode. d Schematic representation of the three-dimensional domain structure.

Taking advantage of the stress sensitivity of Raman spectroscopy, we used Confocal Raman maps to monitor the distortion of the oxygen octahedra originated by the appearance of the ferroelectric domains. In Fig. 22.6a we show the Raman shift of the A$_{\mathrm{1g}}$ mode over two adjacent stripe regions. The evolution of the A$_{\mathrm{1g}}$ mode, in Fig. 22.6b, shows clearly an abrupt decreasing in the Raman shift of this mode that occurs at the 90$^\circ$ domain wall. The occurrence of alternating regions with T and T-O domains, as a result of the new crystalline structural distortions, must be accommodated at such 90$^\circ$ domain boundary. The decrease of polarization at the 90$^\circ$ domain walls is consequence of such stress, which can be evaluated by statistical analysis of the A$_{\mathrm{1g}}$ mode Raman shift of all spectra shown in Fig. 22.6c.

To sum up, we feel that the methodology proposed here is a powerful tool to evaluate complex domain structures, which allows imaging of the ferroelectric domains with high spatial resolution and therefore it offers the opportunity to build a simple model that explain the nature of domain walls and correlation between the structure and piezoelectric properties (Fig. 22.6d).

4 A Potential Technological Application: Ferroelectric Domain Wall Motion Induced by Polarized Light

In the last section we will elucidate the surprising ability to move ferroelectric domain walls of a BaTiO$_{3}$ (BTO ) single crystal by varying the polarization angle of a coherent light source [53]. Along the different sections of this chapter, we have emphasized that ferroelectric materials are characterized by exhibiting spontaneous and stable polarization, which can usually be reoriented by an applied external electric field. Various ferroelectric devices rely on the electrically switchable nature of this polarization, such as nonvolatile ferroelectric random access memory (FeRAM). In such memory devices, the storage of data bits is achieved by motion of domain walls that separate regions with different polarization directions. Therefore, an external voltage pulse can switch the polarization between two stable directions, representing “0” and “1”. This behavior is responsible for a read/write process that can be completed within nanoseconds. The major drawback of FeRAM is that it requires a circuitry access, limiting their practical implementations for commercial use due to their difficult integration into devices as compared to their conventional magnetic random access memory counterparts. Consequently, there is a need for a method for switching the polarization of ferroelectric materials without the need of a circuitry access.

The engineering of ferroelectric domains is now advancing at a rapid pace. There has been progress in the understanding of the behavior of ferroelectric domain walls, especially on the control of their movement. New research studies are ongoing in order to find effective methodologies capable of modulating ferroelectric domain motion. Recently, in a pioneering study, T. Sluka and co-workers [54] demonstrated the correlation between the existence of “strongly” charged domain walls (sCDW) in ferroelectric BaTiO$_{3}$, and their enhanced electromechanical properties (electron-gas-like conductivity while individual domains remained excellent insulators). Therefore, the discovery of this stimulant behavior from sCDW described for BTO-based materials and the potential technological applications from these enhanced functionalities raises a need for an efficient and non-invasive method to switch ferroelectric domains without the need of electrical connections or physical contact, for example, by means of an electrostrictive film.

4.1 Resolving the Origin of the BaTiO$_{3}$ Complex Domain Structure

As we have shown in the previous section, a large variety of methods are used to characterize the ferroelectric domain structure. For that, it is crucial to know both structural and topographic properties that are at the core of engineering of ferroelectric domains. AFM and Raman imaging contribute to the analysis of the ferroelectric domain structures by visualizing the distribution of the individual domains based on the differences in the characteristic Raman spectra of each one them.

In an effort to shed light on the relationship between the switching of ferroelectric domains and their functional properties (e.g. the storage of data bits), we have chosen a BaTiO$_{3}$ (BTO) single crystal as a model case. The BTO single crystal used in this study was produced by top-seeded solution growth (TSSG) and provided by PIKEM Ltd (UK). The $5\times 5\times 1$ mm$^\text {3}$ BTO crystal was grown with (100) orientation, or a-plane. The single-crystal was polished using 1 $\upmu$m diamond paste, and then cleaned with acetone and ethanol before characterization. No further thermal and/or chemical etching was used to reveal the domain structure, thus avoiding topographic artifacts induced by these processes. The sample was maintained at T > 25 $^\circ$C during the AFM and CRM measurements.

A basic identification of the structure and of the crystalline orientation of the single crystal reveals that the sample presents a tetragonal symmetry and two different crystallographic orientations, (001) or c-plane and (h00) or a-plane [53]. At first the BTO single crystal was characterized from a topographical point of view by optical microscopy and AFM (Fig. 22.7a–d). The morphology map indicates that the domain structure is mainly composed of domains of 40–50 $\upmu$m width and adjacent domains appear alternately as protrusions and troughs (height difference of ${\sim }120$ nm) (Fig. 22.7c). The AFM scan also reveals the domain boundary topography, associated with soft transitions (Fig. 22.7d). Moreover, such transitions are asymmetric: left boundary of the protrusions presents a characteristic step (1.5–3 $\upmu$m), while this step is not observed for right boundary.

Fig. 22.7

Modified and reprinted with permission from [53]. Copyright 2015 Nature Publishing Group

Resolving the BaTiO$_{3}$ complex domain structure: a Optical micrograph of the domain structure. b AFM image of BTO crystal, showing the domain topography inside the marked white box of (a). Scale bar, 20 $\upmu$m. c AFM topography scan along the red arrow of (a). d Detail of the domain boundary topography, corresponding to the left boundaries LB-1 and LB-2 of panel (c). The Raman image showing the domain distribution at the surface (recorded in the white rectangle marked in (a)) by colour code (e) as well as in the depth scan (f). Raman spectra having the same spectral shift are shown using the same colour and the colour intensity correlates with the Raman intensity. Scale bar, 20 $\upmu$m. g Main Raman spectra of BTO Raman image associated with the three different colours: red = a-domain, blue = c-domain, green = b-domain, collected at the points marked as A, B, and C in panel e, respectively. The numbers next to the vibrational peaks represent the main atomic motions (see Table 22.1 for assignments). The inserts show magnified Raman spectra, ascribed to the 4 and 5 Raman modes, respectively. h The Figure displays a schematic illustration of the domain structure, which has been built by combining the AFM and Raman imaging information, shown in (a–d) and (e–g) respectively.

Although the topography is an essential pillar for the determination of the domain structure, CRM allows studies at a local scale of the structural deformations of perovskites, induced both by the tilting of BO$_{6}$ octahedra and by the cationic displacements [52]. CRM is here combined with AFM in the same apparatus, thus giving direct correlations between topography and local structure (Fig. 22.7e, f). The Raman spectra are collected at a plane located just below the surface of the sample where the Raman intensity is maximized. The selected area ($150 \times 30$ $\upmu$m) is the same previously studied by AFM (Fig. 22.7b). The acquisition time for a single Raman spectrum was 1 s (1 pixel). Thus the Raman image consisting of $150 \times 30$ pixels (4500 spectra) required 75 min for the planar-section. Features such as Raman peak intensity, peak width or Raman shifts are fitted with algorithms in order to compare information and to represent the derived Raman image. The assignments of the observed Raman modes, both symmetry and nature (first and second order), are summarized in the Table 22.1.

Table 22.1 Raman modes and their mode symmetry assignments in tetragonal BaTiO$_{3}$ single crystal. The table summarizes both symmetry and nature (first and second order), of the Raman modes of the BaTiO$_{3}$ phase. According to the nuclear site group analysis, Raman active phonons of the tetragonal P4mm (C$_{4v^1}$) crystal symmetry are represented by 3A$_{1}$ + B$_{1}$ + 4E. Long-range electrostatic forces induce the splitting of transverse and longitudinal phonons, which results in split Raman active phonons represented by 3[A$_{1}$ (TO) + A$_{1}$ (LO)] + B$_{1}$ + 4[E (TO) + E (LO)]

Raman spectra having the same Raman shift are classified by colours and the colour intensity corresponds to the Raman intensity. The combination of colours results in a Raman image of the surface (Fig. 22.7e) and a Raman depth scan image of the cross-section (Fig. 22.7f). Figure 22.7g shows the average Raman spectra obtained in adjacent domains: A (red) and C (blue) points. The a-domains (so called because its (100) crystallographic orientation) are easily detectable since they are characterized by the annihilation of two Raman modes (4 and 6 in Fig. 22.7g), while the presence of the both Raman modes is attributed to the c-domains (so called because its (001) crystallographic orientation). With the aid of optical microscopy and confocal Raman microscopy, mappings of the domain structure are obtained at the surface and deeper within the sample (Fig. 22.7e, f, respectively). Two well-known types of domains are identified by their characteristic Raman peak (a- and c-domains) and we also found a new type of domain (green colour, point B in Fig. 22.7e), which appears at the surface like a domain boundary. For the sake of simplicity, we will use hereafter the term b-domain to refer to this boundary. From their Raman signature, the b-domains (green colour, point B in Fig. 22.7e) appear clearly as a combination of both the a-domain and c-domain that could be explained as a bundle of subdomains. Indeed, recent studies have revealed the existence of a-c-subdomain structure in thin single-crystalline slices of BaTiO$_{3}$, referred  as “superdomains” [62]. The formation of complex domains accounts for the existence of stress in thin layers, that usually relax for thicknesses >100 nm. In addition, this fact can produce a small onset of electrostatic potential across the two sides of a 90$^\circ$ domain wall, which should have a dominant effect on the migration of charges that accumulates electron and oxygen vacancies at domain walls. As a relevant result, we believe that the b-domains clearly appear in the a-c-domain wall due to crystalline stress higher in the a-c-domain wall than in the c-a one. Due to the boundary conditions of the crystal surface, the c-domain protrudes accordingly with out-of-plane distortion, while the a-domain is depleted. AFM (Fig. 22.7a–d) and Raman imaging information (Fig. 22.7e–g) help in the building of a simple model of the complex domain structure (Fig. 22.7h) by which it is possible to explain the origin of the appearance of a new type of domain on the BTO. The scheme shows a domain structure composed of a-domains and c-domains with a head-to-head configuration of the polarization vectors, which are represented in red and blue colours respectively. The head-to-head configuration maximizes the internal stress close to the domain wall. As a consequence of this internal stress the a-c-domains are hindered by b-domains, which are represented in green colour. The insert of the b-domains structure shows how internal stress at the domain wall is minimized by a bundle of alternate a-domain and c-domain.

4.2 “In Situ” Ferroelectric Domain Switching Using a Polarized Light Source: A Potential Technological Application

Following the above analysis and interpretation, the b-domains are associated with a significant stress degree and according to the literature by an uncompensated charge [63]. Strongly charged domain walls (sCDW) generate photovoltages that could imply an interaction of photons with the accumulated charge. Consequently, a response of the sCDW related either to charge or stress redistribution could be expected when exposed to light. In addition, as we noted in the introduction of this section, an emerging application of ferroelectric materials are FeRAM devices, where the polarization switching of the ferroelectric domain is the key to the storage of data bits.

To explore such possibilities, in situ observations of polarization switching are carried out by applying polarized light. In all cases the incident light is normal to the sample surface and the polarization direction of light parallel to the surface. The angle between the polarization of light and the depth scan direction varies with increments of $\varDelta \varTheta =15^\circ$ between $\varTheta =0^\circ$ and $\varTheta =90^\circ$. The scanning direction is always perpendicular to the a-c-domain wall. Thus, in the a-domain case the polarization of light and consequently its associated electric field (E) is parallel to in plane polarization for $\varTheta =0^\circ$ (E parallel to P) and perpendicular for $\varTheta =90^\circ$. Therefore, the electric field has no effect on the polarization vector of BTO-cell for $\varTheta =0^\circ$ (E and P are parallel) and has a maximal impact for $\varTheta =90^\circ$ (E and P are perpendicular). Thus, for the a-domain, the effect of the electric field could imply an in-plane rotation of the polarization vector that remains parallel to the crystal surface even with re-orientation of its polarization direction under polarized light.

For the c-domain, the polarization of light is always perpendicular to the out-of-plane polarization. Whatever the $\varTheta$ angle, the electric field associated to polarized light has the same effect. For the b-domains, a combination of both behaviors is expected as a result of the bundle of a-c-subdomains. Thus b-domains are probably the most sensitives to electric field and thus to the polarized light.

Fig. 22.8

Reprinted with permission from [53]. Copyright 2015 Nature Publishing Group

Motion of ferroelectric domains under polarized light: a–g Sequence of Raman depth scan images showing the switching of the c-domains and b-domains in the BTO cross section for different angles of light polarization between 0$^\circ$ $\rightarrow$ 90$^\circ$. The scheme localized at the top of (a) represents the angle $\varTheta$ of the polarized light in the XY plane. Additionally, on the left of each image the light polarization angle value is indicated (a–g). The scale bar is 1 $\upmu$m. h–n Sequence of Raman shift depth scans images showing the Raman shift at each pixel corresponding to the A$_{1}$(TO$_{2}$) Raman mode 2 in Table 22.1. The colour code of the bar corresponds to maximum (220 rel. cm${^{-1}}$) and minimum (205 rel. cm${^{-1}}$) values of Raman shift of (h–n). The relative motion of domain is illustrated along the line marked as F $\rightarrow$ G in (a–g), and is plotted in black as a function of the polarization angle ($\varTheta$) in (o). The error bars show the standard deviation ($\sigma$) of the relative motion from the measurement for each angle of light polarization on a given sample. Moreover, panel o also shows the Raman shift evolution for three main points in the complex domain structure representing a-domain, a-c-domain wall and b-domain which are inscribed as 1$\rightarrow$1’, 2$\rightarrow$2’ and 3$\rightarrow$3’ in (h–n), respectively. The error bars show the standard deviation of the Raman shift evolution for three main points in the domain structure for each angle of light polarization on a given sample.

Figure 22.8a–g shows a sequential depth scan Raman images obtained for various $\varTheta$ angles. The D-E-line serves as a guide to the eye to follow the domain motion. Clearly, ferroelectric domains move along the X axis for both the b-domains (green) and the c-domains (blue). When the polarized light reaches $\varTheta = 90^\circ$, the Raman spectrum of the c-domain evolves toward b-domain (Fig. 22.8a–g), accounting for a structural change. The a-domain (red), located on top surface of the sample, appears as unchanged. The local change of domain configuration seems to imply the displacement of the b-domain (green) to the detriment of the c-domain (blue). This observation supports our previous hypothesis of charged b-domains: these charges are highly sensitives to electric field and associated to the motion of domain walls.

However, it is noteworthy that the domain wall motion observed here implies a modification of tetragonality and polarization, at least at local scale. As we have demonstrated along the sections presented here, the Raman shift is an indicator of the crystal stress and correlates with both tetragonality and polarization [7, 64]. Indeed, a modification of the chemical environment (atomic displacement, local stress ...) changes the force constants of chemical bonds and thus modify the phonon frequency of some Raman modes (Raman shift). By analogy with the visible spectrum and compared to references, a Raman shift increase of the Raman peak frequency is called “blueshift” (higher energy of phonon), while a Raman shift decrease is called “redshift”. Raman shifts reveal chemical bond variations in the BO$_{6}$ octahedron, associated with the crystalline stress: we used confocal Raman imaging to understand this complex domain structure. The A$_{1}$(TO$_{2}$) Raman active phonon is a symmetrical mode, which is detected as relatively strong scattering signal in BTO because of a near-perfect equilateral octahedral symmetry. Figure 22.8h–n display depth scans Raman shift images related to the A$_{1}$(TO$_{2}$) Raman mode, for the different light polarization angles as indicated. Raman shifts as large as ca. 15 cm$^{-1}$ (from 205 to 220 rel. cm${^{-1}}$) are determined in the complex domain structure. The Raman shift also evolves with the light polarization angle, with a translation of the domain structure, as stated above. The b-c-domain walls move and the region with local charge accumulation moves accordingly, undergoing redshift with increasing $\varTheta$. The redshift seems to be more relevant for $\varTheta >45^\circ$ and the most relevant redshift occurs in the a-domain, indicating stress relief, or to some extent a structural modification of the crystal lattice. As a whole, the domain structure promotes a polarization reduction by the effect of the polarized light that stimulated the domain motion.

Figure 22.8o summarizes the extent to which the domain position is affected by the light polarization angle. This relative motion is illustrated by using as a reference the line F $\rightarrow$ G in Fig. 22.8a–g. This analysis shows that the relative motion of the domains is approximately $2.16\pm 0.09$ $\upmu$m when the light polarization direction changes from 0 to 90$^\circ$. Accordingly, the relative domain displacement presents two different regimes as a function of the light polarization angle. From 45 to 90$^\circ$ the relative motion approximately triplicates the one observed between 0 and 45$^\circ$, Fig. 22.8d–g. In addition, there is a progressive change of the c-domain nature that reaches the top of the single crystal surface, which for the observed region becomes a b-domain in nature. The previous experiments are attempted in two ways: the first one, the light polarization angle increasing in steps of 15$^\circ$ from 0$^\circ \rightarrow$ 90$^{\circ }$ and thereafter decreases from 90$^\circ \rightarrow$ 0$^\circ$. For the second one, the light polarization angle randomly switches between the selected steps. In all cases the domain motion is reversible and remains within the error bars of Fig. 22.8o. In addition, the experiments are performed on different days and different regions along the complex domain structure, reaching similar results that confirm the reproducibility.

To summarize this final section, we demonstrate how polarized light with a wavelength of 532 nm can induce the motion of ferroelectric domain walls of a BaTiO$_{3}$ single crystal without any physical contact. It is worth pointing out that this stimulant behavior observed here could have potential technological applications leading to a non-contact read-out method in FeRAM devices or remote control of piezoelectric actuators.

5 Outlook

Raman spectroscopy can be used to establish relationships correlating the functional properties and crystalline phase changes (i.e. intrinsic contribution) on ferroelectric materials, respectively. Being a fast measurement technique and requiring only small quantities of sample material it is a powerful technique for material testing, also for industrial applications and processes. In combination with confocal microscopy, the identification and distribution of crystalline phases can be used to design new ferroelectric materials with improved properties.

Confocal Raman Microscopy (CRM) allows us to acquire at high resolution the microstructural features such as the ultrastructure of the ferroelectric domain configuration (extrinsic contribution) and their highly ordered architecture at a scale ranging from few nanometers to several microns. For a correct assessment, it is necessary to determine both the structural and topographic features that are at the core of potential technological applications. To facilitate that, we took advantage of the combination of CRM (chemical information) with AFM, (topographic information), to obtain molecular-level information on the relationship of the crystalline structure with the domain formation as well with the effect of polarization.

Finally, one of the most important points of this chapter is the demonstration that advanced measuring techniques such as CRM (especially coupled with AFM) can be successfully used to resolve and manipulate ferroelectric domains for piezoelectric engineering. As an example, the methodology proposed here has allowed us to demonstrate how polarized light can induce the motion of ferroelectric domain walls of a ferroelectric material without any physical contact. This breakthrough observation could lead to memory devices without electrical connections, by converting light energy directly into ferroelectric domain wall motion on a BaTiO$_{3}$ single crystal. In more general terms, the coupling between coherent light and the crystal orientation opens up exciting new opportunities for materials science.