Electromechanical properties of paper-derived potassium sodium niobate piezoelectric ceramics

Abstract

The small-signal dielectric and piezoelectric coefficients of paper-derived sodium potassium niobate, K_0.5Na_0.5NbO₃ (KNN), were compared with those of conventionally prepared samples. Results show similar functional properties of paper-derived KNN without significantly decreasing the small-signal piezoelectric coefficient. The structure and microstructure analysis of conventional KNN and paper-derived KNN did not reveal any significant difference in the crystal structure and grain size. However, the temperature-dependent inter-ferroelectric phase transition temperature estimated from the temperature-dependent dielectric permittivity data revealed a decrease of approximately 18°C for the paper-derived KNN and is possibly associated with the structural and microstructural defects. This work indicates that optimizing suspension chemistry and sintering conditions will be critical to enhance the functional response of paper-derived KNN further. Moreover, paper-derived ceramic processing, a novel and cost-effective additive manufacturing technology, can be potentially used to fabricate other electroceramics with a wide range of porosities and sizes as well as complex geometries and multilayer structures.

1 INTRODUCTION

Preceramic paper-derived fabrication techniques offer the possibility of large-scale production of numerous ceramic materials with different complex shapes as well as multilayer structures.¹ Importantly, depending on the type and the weight fraction of the inorganic filler/organic fiber, macro-and microscopic porosities can be tailored and enable the fabrication of lightweight ceramic products.² Preceramic paper is characterized by a high filler content of inorganic powder of more than 80 wt.%. After preparing a suspension consisting of filler particles, cellulose fibers, retention agents, and binders, a paper is produced by dewatering through a sieve. Achieving a high solid content, that is, retention of filler in a paper sheet is critical in achieving ceramic components with mechanical stability.¹ Retention can be increased by various mechanisms, distinguishing between mechanical retention (hooking of the powder particles to the fibers) and chemical retention (coagulation and flocculation processes in the suspension, which increase the effective particle size).^{3, 4}

Although the paper-derived ceramic processing technique has been mainly used for high-temperature refractory ceramics, such as Al₂O₃,⁵ SiC,² and Ti₃SiC₂,⁶ it is also feasible to fabricate electro-active ceramic materials.⁷ For example, Menge et al.⁷ demonstrated the fabrication of ferroelectric BaTiO₃ with acceptable functional properties using the paper-derived method. A small-signal piezoelectric coefficient, d₃₃, of 76 pC/N and relative permittivity, $${\varepsilon _r}$$, of 1241 were achieved in samples with 19% open porosity and 79% relative sintered density. This previous work on paper-derived BaTiO₃ showed promising electromechanical behavior without significantly decreasing functional response despite the relatively higher porosity. However, to fully realize the application potential for paper-derived ceramic processing for electroceramics, a further understanding of processing techniques and their relationship with functional electromechanical properties is required.

The designed architecture of piezoelectrics, that is, electromechanical metamaterials, is increasing because of their high specific piezoelectric constant at low material volume fraction and convenient tunability of functional response.⁸ However, methods for rapid and cost-effective processing of such architecture of piezoelectrics are not well established yet. One of the critical aspects of the paper-derived ceramic processing method, that is, their flexibility in preparing complex-shaped architecture, can be helpful in further developing additive manufacturing of piezoelectrics with complex architecture.

K_0.5Na_0.5NbO₃ (KNN) and related compositions are widely researched for their demonstrated functional properties and the potential to replace and/or overcome some limitations of commercially dominating lead-based compositions.⁹ The conventionally prepared KNN ceramic shows promising piezoelectric properties, such as a d₃₃ of 120 pC/N, $${\varepsilon _r}$$ of 500 with ferroelectric–paraelectric phase transition temperature, $${T_c}$$ at 420°C.¹⁰ For high-frequency (> 20 MHz) transducer applications, compositions with low density are advantageous because it is convenient to fabricate thicker ceramic elements for a specific frequency. In this regard, KNN with a density of 4.5 g/cm³ possesses a significant advantage over PZT with 8 g/cm³. Therefore, investigating the piezoelectric response of KNN-based composition fabricated using the paper-derived technique can be critical in evaluating the feasibility of possible applications. Moreover, considering that the paper-derived technique can be used to fabricate ceramic components with complex shapes, a feasibility study of the paper-derived KNN-based composition can open new opportunities for future applications. In this work, the temperature-dependent electromechanical response of paper-derived KNN ceramic was investigated and compared with conventionally prepared KNN. Our results demonstrated that the well-established paper-derived ceramic fabrication technique is effective for fabricating advanced functional ceramics for electromechanical applications. Importantly, this processing technique offers opportunities to produce functional ceramic components with different size and complex geometries.

2 EXPERIMENTAL METHODOLOGY

The suspension for the papermaking process was first optimized regarding the type and amount of raw materials, the amount and ratio of long and short fibers, and their influence on retention. Then, preceramic papers with this optimized composition of 30.1 wt.% K_0.5Na_0.5NbO₃ (KNN) ceramic powder (d₅₀ = 0.95 μm, CerPoTech, Norway), 57.4 wt.% pulp suspension (mixture of 40 wt.% long fibers (Orion EFC, Zellstoff Pöls, Austria) and 60 wt.% short fibers (Celbi PP, Cellulose Beira, Portugal), 9.6 wt.% cationic and anionic starch (Fibraffin K72, Fibraffin A5, Südstärke, Germany), and 2.9 wt.% retention agent (Percol 121 L, BASF, Germany) were produced using a Rapid-Köthen laboratory sheet former (Haage Laborblattbildner BBS-2, Estanit GmbH, Germany, DIN EN ISO 5269-2). In this method, the suspension was placed in a glass cylinder and homogenized by whirling. The water was then sucked out of the cylinder through a sieve with a mesh size of 50 μm and a diameter of 200 mm by applying a low vacuum (< 10⁴ Pa). Details of the preceramic paper formation can be found in our previous works.¹ The resulting preceramic paper with an approximate thickness of about 200 μm was dried at 95°C. The dried paper was cut to 30 mm × 30 mm size and pressed to five-layer laminates by hot pressing (Polystat 200T, Servitec Maschinenservice, Germany) at 80°C with 20 MPa load for 15 min. Before the sintering step, the preceramic paper was heat-treated to 600°C to remove organic additives. In the debinding step, the heating rate was 0.5 K/min, with a dwelling time of 60 min each at 250°C, 405°C, and 600°C. Sintering was conducted in a closed environment with samples embedded in constituent KNN powder. The sintering temperature of 1150°C with a dwelling time of 2 h and a heating rate of 3 K/min was used to fabricate both paper-derived and conventional KNN. The green conventional KNN samples were prepared by cold isostatic pressing at 180 MPa of the as-received KNN powder. After grinding, the thickness of the sintered paper-derived ceramics (5-layer laminate) was approximately 660 μm.

The microstructures of the preceramic paper in green and sintered state and of the conventional samples were investigated by scanning electron microscope analysis (SEM, Quanta 200, FEI, Czech Republic) after grinding and polishing with a 1 μm finish. The geometric density and porosity of the samples were determined by weight and volume measurements. Additionally, SEM micrographs with different magnifications were used to determine the micro-and macro-porosity by image analysis (ImageJ, Wayne Rasband). The micrographs were converted into binary images, and the black area representing the pores was measured. The crystal structure was examined by X-ray diffraction (XRD) measurements (Bruker D8 Advance eco, Bruker AXS GmbH, Germany), and the chemical analysis was performed using energy-dispersive X-ray spectroscopy measurement (EDX; INCA x-sight TVA3, Oxford Instruments, UK).

To evaluate the microstructure, μCT scans were performed on conventionally sintered and paper-derived KNN samples of cylindrical shape (diameter ≈ 4 mm) using a Skyscan 1172 (Skyscan, Belgium) with a tungsten tube (λ = 0.024 nm) and an 11 MP. The μCT scan was performed at 80 kV, 100 μA with a resolution of 1.69 μm/pixel for 180° and a rotation step size of 0.25°. The distance between image slices is equal to the resolution. To improve the scan quality, a combined Al/Cu filter was used to reduce the noise of the low-energy X-rays. The two-dimensional (2D) sinograms were reconstructed using NRecon with GPU support (GPUNrecon) (version 1.7.4.2, Bruker, Belgium), resulting in a 2D image stack for further analysis.

Two opposing sample surfaces were sputtered with Pt electrodes for dielectric and electromechanical characterization. The temperature-dependent relative dielectric permittivity of paper-derived and conventional KNN was investigated using a modified furnace and measured with an LCR meter (E4980AL, Keysight) at frequencies between 0.1 kHz to 1000 kHz during heating to 500°C with a rate of 2 K/min. Large-signal polarization and strain-electric field measurements were performed using a piezoelectric evaluation system (TF2000, aixACCT). A triangular bipolar electrical field of 3.5 kV/mm was applied with a frequency of 100 Hz. Electrical poling was conducted at 150°C with an applied electric field of 3 kV/mm for 20 min, followed by field cooling to room temperature. The temperature-dependent small-signal d₃₃ of the poled samples were characterized using a custom-built setup as previously described in detail.¹¹

3 RESULTS AND DISCUSSION

The SEM image of the paper surface (Figure 1) in the green state shows flocculated KNN filler powder partially attached to the cellulose fibers. This so-called hetero-flocculation can be explained by the different surface charges of the Percol 121L, which acts as flocculant, and the KNN powder coated with cationic starch.¹² An approximately 94% retention of the filler content was estimated by comparing the theoretical weight and measured weight of the paper. Such high filler content ensures the relatively higher sintered density of the paper-derived KNN samples. Importantly, by controlling the suspension compositions, the retention factor can be tailored to control the density/porosity of the paper-derived ceramic as described in previous work.^{1, 13}

FIGURE 1

Scanning electron microscope (SEM) micrograph of the preceramic paper surface in the green state (before sintering)

After lamination and sintering of the samples, a relative sintered density of approximately 86% was obtained. The achieved density of the paper-derived ceramic is slightly lower than that of the conventionally sintered KNN sample, that is, 89%. It is important to note that the sintered density of chemically unmodified KNN is generally relatively low and has been reported to be between 88%–92%.^14-16 Therefore, the observed low sintered density, that is, poor densification of conventionally prepared KNN, is not due to the processing or sintering condition used here, but rather a general phenomenon of this chemically unmodified KNN composition.¹⁷ The SEM images (Figure 2) show an overall homogenous microstructure for the paper-derived ceramic without observable individual layers, indicating excellent lamination and adhesion. In addition to the residual porosity, areas with different brightness are also visible. These areas indicate chemical inhomogeneities, that is, more niobium and less Na were detected in the brighter areas. The Nb-segregated areas were also visible for conventionally prepared KNN, indicating that the segregated area is not related to the paper-derived ceramic technique but a general feature of the sintered KNN ceramics fabricated for the as-received powder. However, the fraction of bright areas seems to be higher in the paper-derived ceramics. The porosity values determined by image analysis do not differ significantly, especially when the surface is considered (12.5% and 12.6% for the conventional and the paper-derived ceramic, respectively). For the cross-section, the paper-derived samples show higher values (13.9%) than that of the conventional (11.6%) sample, which can be attributed to the greater extent of macro-porosity, as shown in Figure 2D. The higher extent of macroporosity can be associated with the paper manufacturing process.

FIGURE 2

Scanning electron microscope (SEM) micrographs of the polished surface and cross-section of (A,B) conventionally prepared and (C,D) paper-derived KNN ceramics. The bright area in (D) is conductive silver and therefore does not describe the microstructure of the sample

The pore size distribution and pore shape were evaluated using CTAnalyzer (version 1.20.3, Bruker, Kontich, Belgium) on a volume of 1 000 × 1 000 × 320 px according to the porosity-dependent threshold. The size distribution was determined from the 3D volume, and the average pore shape was calculated for each layer in the 2D image stack. As can be seen in Figure 3A, the pore size of the conventionally prepared samples is a minimum of 4 μm and a maximum of 10 μm. The paper-derived KNN has a larger number of pores with a diameter > 6 μm and a maximum size of 13.8 μm. The pore sizes are in agreement with the SEM images. On the other hand, the pore form (Figure 3B) describes the shape of the pore with the marking points 1 = ideal sphere, 0.5 = elliptical pore, and 0 = crack. The paper-derived KNN samples have more elongated pores in the pore form factor range of 0.2–0.6 than those of the conventionally sintered KNN. Here, more spherical pores are found (range 0.8–1.0). This result again highlights the influence of processing route on the microstructure, that is, size and shape distribution of pores.

FIGURE 3

(A) Pore size distribution and (B) histogram of the pore form (shape) of conventional and paper-derived KNN

Figure 4 shows the X-ray diffraction patterns of as-received KNN powder with sintered ceramic samples of conventional and paper-derived KNN. The primary reflections from sintered ceramics match well with the as-received powder. However, both the sintered samples, that is, irrespective of the processing techniques used here, showed traces of secondary phases and are the possible origin of bright areas observed in SEM images (Figure 2). Such a secondary phase in KNN-based compositions generally indicates the existence of a tetragonal tungsten bronze (TTB) phase with the formula K₂Nb₄O₁₁, possibly due to the volatilization of alkali metal oxides, that is, Na₂O and/or K₂O, during the sintering.^{17, 18} It should be noted that the formation of the TTB secondary phase influences the electromechanical response and conductivity of KNN-based compositions.^{19, 20} The diffraction data indicate that sintering at 1150°C induces minor secondary phases in both cases, that is, 3% and 5% for conventional and paper-derived KNN, respectively, highlighting that the impurity phases are not due to the paper-derived ceramic processing technique. Nevertheless, further optimization of the sintering condition, such as atmospheric sintering with reduced oxygen partial pressure, can be used to prevent the formation of the secondary phases in KNN.^{18, 21}

FIGURE 4

X-ray diffraction patterns of the as-received KNN powder, sintered conventional (Conven.), and paper-derived (PD) KNN. Insets show variation in 110 pc and 200 pc reflections

In order to understand the functional dielectric response, for example, capacitance and loss, and their temperature-dependent variation in the crystal structure, temperature-dependent dielectric permittivity measurement up to 500°C was conducted. The temperature-dependent permittivity data for conventional and paper-derived KNN (Figure 5A,B) show a typical response with dielectric anomaly around 190°C and 400°C, indicating the interferroelectric phase transition from the orthorhombic-tetragonal phase and the ferroelectric–paraelectric phase, respectively.¹⁶ Interestingly, the phase transition temperatures seem to shift to a lower temperature, by approximately 18°C for the paper-derived KNN. Generally, a shift in the phase transition temperature can be related to stress and/or defects.^22-24 Previous works have shown that externally applied hydrostatic pressure generally shifts the phase transition temperature to a lower temperature, whereas uniaxial or biaxial stress shifts the transition temperature to a higher temperature.^{25, 26} It should be noted here that a shift in the phase transition temperature has also been associated with the variation in microstructure, that is, $${T_c}$$ shifts to higher temperature with decreasing grain size.^{27, 28} Considering that the average grain size (≈ 5 μm) is similar for both KNN samples, the influence of grain size can be excluded. Importantly, the room temperature dielectric loss is higher for the paper-derived KNN at lower frequencies (≤ 1 kHz), indicating increased conductivity in this sample, which can be associated with the space charge effect.²⁴ Interestingly, the change in the phase transition temperature can also be related to the increase in defect concentration, that is, the space charge effect associated with the migration of oxygen vacancies.²⁹ For example, in the case of BaTiO₃, the $${T_c}$$ can be lowered by 60°C–70°C by annealing in a reducing atmosphere and associated increased concentration of oxygen vacancies.³⁰ However, further temperature-dependent measurements with the different partial pressure of oxygen will be critical to confirm the observed lowering of phase transition temperatures in paper-derived KNN. Although the increased defect, that is, space charge, seems to be the primary origin of lower temperature shift of the phase transition temperature in the PD KNN, the influence of slightly higher extent of secondary phases observed in SEM and XRD data cannot be ruled out.

FIGURE 5

Temperature-dependent permittivity of (A) conventional and (B) paper-derived KNN ceramics

Considering the relatively higher porosity and observed secondary phases in the XRD data, it is essential to investigate the large-field polarization (P)—and strain (S)—electric field (E) measurements with increasing temperature to understand the feasibility of paper-derived KNN ceramics for electromechanical applications. The polarization and strain–electric field loops as a function of temperature are shown in Figure 6. Depending on the processing conditions and measurement frequencies, the E_c and Pr of polycrystalline KNN are reported to be in the range of 1.2–2 kV/mm and 15–19 μC/cm², respectively.^{31, 32} The room-temperature coercive field, E_c ≈1.4 kV/mm, and remanent polarization, P_r ≈ 13 μC/cm²– measured in this work, are within the range of those reported in the literature. The P–E loop of the conventional KNN does not show any significant variation with increasing temperature; however, the S–E loop measured at 200°C exhibited significantly higher maximum strain. This is likely because the temperature-dependent polymorphic phases coexist (see Figure 5A) in this temperature range for the conventional KNN, that is, phase instability regions exhibit enhanced extrinsic contributions such as non-180° domain wall motion and phase transitions.³³ The temperature-dependent P–E loops of paper-derived ceramics show a similar response to that of conventional KNN, although the remanent polarization is relatively low. Moreover, both samples showed an internal bias field (E_ib), that is, asymmetry in the positive and negative coercive field (E_c) of the P–E loop, $${E_{{\rm{ib}}}} = ( { - {E_c} + {E_c}} )/2\;$$, as shown in Figure 6F. Apparently, the paper-derived KNN exhibits a higher internal bias field, indicating a relatively higher extent of electrically-active defect species and related increased conductivity in the paper-derived ceramics and is possibly related to the higher macroscopic porosity in these samples. It should be mentioned that the fabrication process of paper-derived ceramics was water-based, as such increased leakage current can also be associated with the hygroscopic nature of the KNN.³³ Nevertheless, despite the higher leakage current, the paper-derived ceramic samples could be poled and achieve a stable piezoelectric constant comparable to that of conventional KNN.

FIGURE 6

Electric field-dependent polarization and strain response of (A, C) conventional and (B, D) paper-derived KNN measured at different temperatures. Variation in (E) remanent polarization, P_r and (F) internal bias field, E_ib as the function of temperature

In order to ascertain the impact of the processing method, that is, paper-derived and conventional processing, on the small-signal piezoelectric properties, the temperature-dependent direct small-signal direct piezoelectric coefficient, d₃₃, was characterized from 23°C up to 400°C as a function frequency (Figure 7). The conventional KNN (Figure 7A) shows a room-temperature d₃₃ value of approximately 100 pC/N without any significant frequency dependency, and the value increases with increasing temperature up to the interferroelectric phase transition, T_O-T (∼197°C), followed by a subsequent decrease with further increase in temperature. The observed high d₃₃ around the O-T phase boundary is due to the influence of permittivity and polarization.^{34, 35} The temperature-dependent response up to the T_O-T is consistent with those of previous reports of KNN-based compositions.^{36, 37} However, interpretation of the temperature-dependent d₃₃ response above 225°C is not conclusive due to the apparent contribution from the increased mobility of defects. The paper-derived KNN also shows a similar room-temperature d₃₃ value of approximately 100 pC/N, however, the dispersion of the d₃₃ value with measurement frequencies is higher even below the T_O-T. This increased frequency dependency is possibly related to the space charge effect associated with higher porosity and/or structural defects in the sample. A similar increase in frequency-dependency of d₃₃ was observed in porous PZT ceramics, where the increased porosity and associated defects showed increased frequency dispersion.³⁸ Similar to conventional KNN, the thermally activated defects contribute significantly to the piezoelectric response above 225°C. Although the origin of higher conductivity above 225°C in the KNN sample is not well understood, this could be associated with the observed secondary phase (see Figure 4). Interestingly, the T_O-T of the paper-derived KNN estimated from the temperature-dependent d₃₃ data was found to be 24°C lower than that of the conventional KNN. As explained for the temperature-dependent permittivity data, such lowering of the T_O-T is possibly related to the space charge effect. Nevertheless, the paper-derived KNN showed a stable piezoelectric response up to approximately 165°C and can be useful for sensing or for vibration energy harvesting applications below this temperature range.

FIGURE 7

Temperature-dependent small-signal d₃₃ of (A) conventional and (B) paper-derived KNN

KNN-based compositions possess unique advantages because of their inherent low theoretical density and relatively low-dielectric permittivity but with reasonable small-signal d₃₃ value. The effectiveness of piezoelectric materials for vibration energy harvesting can be evaluated by the product of the d₃₃ and g₃₃, known as the transduction coefficient, where the g₃₃ is the piezoelectric voltage coefficient ( = $${d_{33}}/{\varepsilon _r}{\varepsilon _o}$$). As such, the temperature-dependent voltage coefficient has been calculated for the KNN samples, as shown in Figure 8. The room-temperature g₃₃ was found to be 28 × 10⁻³ Vm/N, which is significantly higher than lead-free compositions such as BaTiO₃ and comparable to that of lead-based soft PZT compositions.^{38, 39} However, it should be mentioned here that only high g₃₃ value do not ensure optimized energy harvesting parameters, that is, the transduction coefficient. As such, despite the comparable g₃₃ of KNN with that of soft PZT, the transduction coefficient is lower for KNN due to the relatively low d₃₃ value. For example, additively manufactured BaTiO₃ showed a room-temperature g₃₃ value of 8.6 × 10⁻³ Vm/N with small-signal d₃₃ value of 150 pC/N, which equates a transduction coefficient of 1290 × 10⁻¹⁵ m²/N. In this study, the paper-derived KNN ceramic displayed a value of 2800 × 10⁻¹⁵ m²/N. It should be mentioned here that the g₃₃ value is highly influenced by the porosity and the dielectric permittivity of the sample. Moreover, the voltage coefficient does not deteriorate significantly with increased temperature below the T_O-T. Importantly, optimized functional properties of paper-derived KNN can be achieved by further optimizing the filler/paper content ratio as well as with improved sintering conditions. The paper-derived processing technique should be explored in the future for other high-performing KNN-based compositions.

FIGURE 8

Temperature-dependent variation in piezoelectric voltage coefficient for conventional and paper-derived KNN samples

4 CONCLUSION

Temperature-dependent electromechanical properties of paper-derived KNN ceramics have been investigated and compared with that of conventionally sintered KNN in this work. The small-signal dielectric and piezoelectric properties of paper-derived KNN are comparable to conventionally processed samples despite the minor difference in porosity, that is, ≈ 2% higher porosity for paper-derived KNN. The water-based paper-derived ceramic fabrication technique does not influence the crystal structure of sintered KNN, that is, no additional secondary phases due to the processing are evident. Notably, the microstructure-porosity can be optimized further by tailoring the suspension chemistry as well as the sintering condition. It should be noted here that the processing of KNN, even using conventional solid-state processing, is challenging due to alkali volatilization and the hygroscopic nature of the constituent elements. Despite the processing challenges associated with the fabrication of KNN-based materials, the data presented in this work indicate that the paper-derived ceramic fabrication method can be suitable and should be explored further to fabricate piezoelectrics with complex architecture.

ACKNOWLEDGMENTS

J.M., T.F., N.H.K., and K.G.W. gratefully acknowledge the Deutsche Forschungsgemeinschaft (DFG) for financial support of their contributions under grant numbers GRK2495D/F/H.

Open access funding enabled and organized by Projekt DEAL.