Introduction

Since the inception and development of Pb(ZrxTi1-x)O3 (PZT) perovskite solid solutions [1,2,3,4,5], PZT has been extensively explored as a model piezoelectric material with renewed focus into the origins of piezoelectricity [6,7,8]. The maximum piezoelectric response is found near the morphotropic phase boundary (MPB) at x ~ 0.52 [9]. Compositional modification by aliovalent substitution of donor and/or acceptor dopant ions on both A- and B-sites in the perovskite structure is extensively documented for tailoring of piezoelectric properties [9, 10]. Donor-doped PZT or ‘soft’ PZT ceramics, commonly with La3+ on the A-site (PLZT) and Nb5+ on the B-site (PNZT), are generally characterised by greater polarisation, piezoelectric coefficients (d33 > 600 pC/N), and dielectric losses (tan δ) through increased extrinsic domain wall contributions primarily underpinned by suppression of the Curie temperature (TC) [11]. ‘Hard’ PZT ceramics, where acceptor dopants introduce domain wall pinning defects, exhibit decreased piezoelectric activity (d33 < 300 pC/N), reduced tan δ from suppressed extrinsic effects, along with large coercive fields (Ec) and relatively high TC suitable for high drive applications [12].

In the last few decades, there have been numerous efforts to push the limits of electromechanical properties, namely through development of binary Pb(Zn1/3Nb2/3)O3-PbTiO3 (PZN-PT)[13] and Pb(Mg1/3Nb2/3)O3-PbTiO3 (PMN-PT)[14] as well as ternary Pb(Mg1/3Nb2/3)O3-PbZrO3-PbTiO3 (PMN-PZT)[15] and Pb(In1/2Nb1/2)O3-Pb(Mg1/3Nb2/3)O3-PbTiO3 (PIN-PMN-PT)[16] single crystals with very high d33 > 1500 pC/N. Textured ceramics produced by templated grain growth (TGG) with the same composition as reported single crystals have also been developed with high d33 coefficients approaching 1000 pC/N [17, 18]. Textured ceramics provide a pathway to harness the effects of piezoelectric anisotropy while engineering the microstructure to tailor the piezoelectric properties of these materials [19]. Despite the strong enhancements in electromechanical behaviour through controlling crystallographic anisotropy, many of the existing materials to date are limited by low-temperature rhombohedral–tetragonal phase transitions (Tr-t < 150 °C), low Curie temperatures (TC < 230 °C), and small coercive fields (Ec < 9 kV/cm) which risk depoling as a result of temperature rises from heat generation or high drive fields [20].

PZT provides an advantage with greater coercive fields (Ec > 10 kV/cm) and high Curie temperatures (TC > 230 °C) to mitigate the risk of depoling in transducers. Incongruent melting across the whole PZ-PT solid solution range [21] has presented a substantial challenge in the development of PZT single crystals [22], commonly resulting in significant compositional segregation, crystal size limitations, and twinning defects. The literature on developments in textured PZT ceramics is limited to tape casting and hot pressing of PZT powders with cubic morphology [23] or tape casting of PZT powder with a large volume fraction (~ 40 vol.% of < 111 > oriented Ba6Ti17O40) of seeds [24]. In all cases, achieved texture fractions were low (f < 60%), but all authors noted that excess PbO played a critical role in promoting textured grain growth. The use of heterogenous seed crystals in TGG requires strict selection criteria: similar crystal structure with minimised lattice mismatch, sufficient chemical stability to survive the epitaxy process, and an understanding of the effect of the seed as a composite system within the host material [25]. Whilst there is significant published work on Ba- [26] and Sr-doped PZT [27] with varying dopant levels across the whole solid solution range, there is minimal literature on the effect of fixed low-level inclusions varying the PZT Zr:Ti ratio, representative of TGG. The addition of dopants necessitates the modification of the Zr:Ti to compensate for seed chemistry to retain a near MPB composition with optimised piezoelectric properties.

This work lays out critical factors in the development of grain-oriented PZT ceramics utilising platelet-shaped BaTiO3 seeds and demonstrates a proof of concept for texturing PZT. The effects of BaTiO3 seeds in the microstructure are explored and the relationship between the PZT matrix composition and resulting piezoelectric properties are established.

Methodology

Fabrication of textured ceramics

The matrix compositions selected for this investigation were a PZT powder synthesised in-house (referred to simply as PZT) and a commercial powder designated PZT-5A1 (Morgan Technical Ceramics, UK). PZT powder was synthesised from mixed metal oxides of PbO (99.9% purity, Alfa Aesar), ZrO2 (99% purity, Sigma-Aldrich), and TiO2 (rutile, ≥ 99.9% purity, Sigma-Aldrich). Oxide powders were blended by vibratory milling together in isopropanol in a 1 L Nalgene bottle with 350 g of a 1:1:1 mass ratio of 3, 5, and 10 mm diameter YSZ milling media spheres for 72 h, dried overnight, sieved to remove any agglomerates, and calcined at 800 °C for 2 h. The PZT powder was re-milled for 72 h to break up aggregates that formed during calcination. PZT-5A1 was similarly prepared by vibratory milling for 72 h. Inspection of the as-milled powders, shown in Fig. 1, revealed particles sizes below 1 µm. Both in-house synthesised PZT and PZT-5A1 material were studied to give an insight into whether TGG of PZT is possible based on the controlled preparation of PZT material and in compositions with dopant elements that may affect textured microstructure evolution.

Figure 1
figure 1

SE images of (a) in-house PZT and (b) commercial PZT-5A1 particles after vibratory milling for 72 h

Fibre-textured PZT ceramics with [001]c preferred orientation were seeded using 5 vol.% BaTiO3 templates (Entekno Materials, Turkey) (Fig. 2) that have a platelet morphology as shown in Fig. 3 [28]. [001]c oriented PZT-5A1 ceramics were seeded using 10 vol.% BaTiO3 platelets. Commercial PZT powders commonly include liquid phase formers to aid densification during sintering. The exact quantity of liquid phase former present in the PZT-5A1 powder is not disclosed by the manufacturer; therefore, the inclusion of an increased vol.% of BaTiO3 platelets is theorised to promote initial textured grain epitaxy at the known detriment of late stage textured grain impingement. PZT slurries (56 vol.% solids loading) with 5 vol.% BaTiO3 seeds with platelet morphology had excess PbO (5 wt.%) added to facilitate grain growth during TGG. Slurries made with PZT-5A1 and 10 vol.% BaTiO3 platelets contained no added PbO excess. All slurries were formed using an aqueous binder system (WB4101, Polymer Innovations, USA), mixed using a speed mixer at 1200 rpm for 20 min, and tape cast using a doctor blade setup with a gap height of 200 µm. Dried tapes were sectioned, stacked, and uniaxially laminated under an applied pressure of 15 MPa at 75 °C, then hydrostatically laminated at 200 MPa at 75 °C for 20 min. Organic binders were burnt out by heating at 0.1 °C/min to 350 °C and holding for 9 h, followed by heating to 450 °C and holding for 5 h. Green bodies were enclosed in covered alumina crucibles with PbZrO3 (PZ) + 5 wt.% ZrO2 sacrificial powder to ensure controlled PbO loss, and heated at 5 °C/min to 1250 °C for 2, 4, and 8 h holds to investigate textured microstructure evolution.

Figure 2
figure 2

XRD pattern for platelet-shaped BaTiO3 seeds

Figure 3
figure 3

SE images of a platelet-shaped BaTiO3 seed (a) flat facing and (b) edge facing

Fabrication of untextured ceramics

Randomly oriented PbyBa1-y(ZrzTi1-z)O3 specimens, with calcined powder compositions ranging from an undoped MPB composition x = 0.52 to x = 0.61 with fixed 5 vol.% BaTiO3 additions, were synthesised by a conventional solid-state route using PbO (99.9% purity, Alfa Aesar), ZrO2 (99% purity, Sigma-Aldrich), TiO2 (rutile, ≥ 99.9% purity, Sigma-Aldrich), and BaTiO3 (Entekno Materials, Turkey). Powder regents were ball milled together in isopropanol for 24 h, dried overnight, sieved to remove any agglomerates, and calcined at 800 °C for 2 h. The calcined PZT powder was re-milled with 5 vol.% BaTiO3 platelets and 10 wt.% PVA binder solution for 24 h. Green bodies were uniaxially pressed at 125 MPa, with the BaTiO3 platelets, destroyed during milling, randomly aligned within the pressed samples. The pressed bodies were subsequently enclosed with PZ + 5 wt.% ZrO2 sacrificial powder during sintering. Initial binder burnout, heating at 0.1 °C/min to 550 °C and holding for 3 h, was followed by heating at 5 °C/min to 1250 °C for 2 h. The final calculated Zr concentration after the addition of BaTiO3 is quoted to distinguish individual specimens.

Structural and microstructural characterisation

The structure and orientation of synthesised multilayers were examined using a D2 PHASER x-ray diffractometer (Bruker) operating in Bragg–Brentano geometry with Cu Kα radiation source (λ = 1.5408 Å), fitted with 2.5° Soller slits, 1 mm divergence slit, and 0.5 and 2.5 mm Ni filters to remove Cu Kβ radiation. Samples were analysed between the 2θ range of 20–80°, using a step size of 0.02° and rate of 0.4°/sec. Lotgering factors were obtained by comparing the relative peak intensities with those of a randomly oriented powder XRD spectrum to quantify the degree of orientation in the multilayer samples [29].

Powder x-ray diffraction data to elucidate the crystal structure and lattice parameters of crushed pellets mixed with Si standard reference material (SRM 640d, NIST) was obtained using a STOE STADI P (CuPSD) x-ray diffractometer (STOE & Cie. GmbH) operating in transmission geometry with a monochromated Cu Kα radiation source, fitted with 6 mm divergence and detector slits. Samples were analysed between the 2θ range of 20–80°, using a step size of 0.02°. Standard x-ray tube working conditions were 40 kV and 40 mA.

Ceramic sample densities were measured via the Archimedes method to qualify the sintering parameters in fabricating samples suitable for electric characterisation. Multilayer fracture surfaces and polished untextured ceramics were examined by scanning electron microscopy (SEM, Inspect F, FEI) to investigate grain structure and morphology. The average grain size for untextured specimens was calculated using the linear intercept method [30].

Electrical characterisation

The dielectric behaviour of ceramic samples was analysed from capacitance and tan δ values measured at 1 kHz up to 500 °C using an E4980A Precision LCR meter (Agilent Technologies Inc.). Rhombohedral–tetragonal phase transition temperatures (Tr-t) were determined from the minimum of the second derivative plots of capacitance with respect to temperature in the low-temperature regime. Curie temperatures (TC) were determined from first derivative plots when the derivative equals zero. Au-electroded specimens were poled at 180 °C under an applied field of 4 kV/mm for 30 min. Quasi-static d33 was measured at a frequency of 100 Hz using a Berlincourt PM300 d33 Piezometer (Piezotest Pte. Ltd), and large signal bipolar polarisation and strain hysteresis data was collected from an aixACCT TF Analyzer (aixACCT Systems GmbH).

Results and discussion

X-ray diffraction of [001]c oriented PZT multilayers

XRD patterns for fired Pb(Zr0.52Ti0.48)O3 ceramics are shown in Fig. 4(a). Randomly oriented undoped PZT ceramics have a tetragonal perovskite structure as indicated by the significant splitting in (002)/(200) peaks at 43° and 45° 2θ. All patterns are absent of any detectable PbO or pyrochlore secondary phases. When the green bodies are seeded with BaTiO3 templates, the relative intensities of (00 l) reflections increase substantially by 2 h. The suppression of all (hkl) peak intensities apart from (00 l) reflections indicates an achievement of perovskite c-axis fibre texture along the [001]c direction. The Lotgering factor of f(001) = 0.91 for PZT after 2 h increases marginally to f(001) = 0.94 after 8 h. The XRD pattern of untextured PZT-5A1 in Fig. 4(b) indicates that the structure is less tetragonal than undoped PZT in Fig. 4(a) as seen by the decreased (001)/(100) and (002)/(200) peak splitting at 22° and 45° 2θ, respectively. After sintering for 2 h, a slight shoulder peak is visible on the (001) and (002) reflections at 22° and 45° 2θ, which is indicative of the tetragonal structure of the BaTiO3 perovskite seeds. With continued heating, the shoulder peak disappears suggesting BaTiO3 enters into solid solution with the PZT matrix. In this case, BaTiO3 seeds can be considered to act as an isovalent dopant, with Ba2+ substituting Pb2+ on the A-site and excess Ti4+ shifting the Zr:Ti ratio of the bulk composition towards the tetragonal side of the MPB. It is therefore necessary to explore a suitable starting Zr:Ti ratio to compensate for BaTiO3 seeding and so that ultimately a composition closer to the MPB may be achieved in textured samples.

Figure 4
figure 4

XRD patterns for (a) untextured PZT reference and [001]c oriented PZT multilayers (5 vol.% BaTiO3) with varying sintering times; and (b) untextured PZT-5A1 reference and [001]c oriented PZT-5A1 multilayers (10 vol.% BaTiO3). * Denotes secondary phase peaks

Microstructural characterisation of [001]c oriented PZT multilayers

SEM images of fractured multilayer surfaces are shown in Fig. 5. Clear inhomogeneity in the grain microstructures is apparent in all textured specimens, the degree of which differs between undoped PZT and PZT-5A1. When sintered for 2 h at 1250 °C, the PZT microstructure is composed of large anisotropically shaped grains with 10–20 µm pores dispersed amongst a matrix of finer sized 5–10 µm equiaxed grains. The presence of elongated porosity and the absences of significant Z-contrast in the back scattered electron images suggests that BaTiO3 is not thermally stable within the PZT matrix at temperatures > 1200 °C. Coarse grains > 20 µm may originate from heteroepitaxial nucleation on the (00 l) BaTiO3 crystal faces prior to their dissolution in the PZT sample, whereas the PZT-5A1 microstructure shows an initial textured layer surrounding remnant porosity. The size difference between textured grains and fine equiaxed matrix grains earlier in the firing cycle, highlighted in Figs. 5(a, c), provides sufficient driving force for continued textured grain growth via Ostwald ripening [31]. After 8 h, for both undoped PZT and PZT-5A1, a coarse grained microstructure is achieved as seen in Figs. 5(b, d).

Figure 5
figure 5

SEM images of undoped PZT multilayer fracture surfaces after sintering for (a) 2 h and (b) 8 h. Fracture surfaces of PZT-5A1 multilayers after sintering for (c) 2 h and (d) 8 h

The final densities of ~ 90% theoretical resulting from the large remnant porosity indicate that more work is needed to improve the processing of BaTiO3-seeded PZT ceramics. Greater control of PbO liquid phase concentration and distribution, matrix particle size/template size ratio, and employment of a pre-densification step may be required to obtain specimens suitable for piezoelectric measurements. Moreover, the large structural deviation from the MPB region as a result of the dissolution of BaTiO3 seeds should be taken into consideration when tailoring compositions for maximum strain behaviour. The following sections therefore focus on providing a stronger understanding of how the Zr:Ti matrix ratio of seeded compositions influences structure and properties.

X-ray diffraction of ceramic PZT with 5 vol.% BaTiO3 seeds

XRD patterns for PbyBa1-y(ZrzTi1-z)O3 compositions within the range 0.49 ≤ z ≤ 0.58, are shown in Fig. 6, along with calculated lattice parameters in Fig. 7. Undoped PZT ceramics, with a nominal MPB composition of x = 0.52, have been reported to show significant phase coexistence between tetragonal P4mm and rhombohedral R3m ferroelectric phases when synthesised via solid-state methods [32] due to each structure being near-energetically degenerate [33]. XRD patterns for samples with compositions 0.49 ≤ z ≤ 0.52 were indexed according to a single tetragonal perovskite structure, with distinct differences in c/a lattice parameters and an absence of secondary phase tetragonal peaks suggesting that the BaTiO3 seeds have dissolved within the PZT matrix. Increasing the initial Zr:Ti ratio results in a decrease in 2θ spacing between tetragonal (00 l)t/(h00)t peak doublets until the appearance of a rhombohedral (00 l)r singlet peak indicating phase coexistence at z = 0.53. XRD patterns from compositions 0.54 ≤ z ≤ 0.58 show (00 l)r singlet peaks consistent with a rhombohedral structure. The effect of Ba2+ substitution (RBa 1.61 Å) with a larger 12-coordinated ionic radius than Pb2+ (RPb 1.49 Å)[34] on the A-site as well as additional Ti4+ incorporation explains why the rhombohedral boundary of the MPB region is shifted from the expected Zr:Ti ratio from 0.52:0.48 to 0.54:0.46.

Figure 6
figure 6

XRD patterns for PbyBa1-y(ZrzTi1-z)O3 with Zr concentration (a) z = 0.49 ≤ z ≤ 0.53 and (b) z = 0.54 ≤ z ≤ 0.58. * Denotes Si standard peaks

Figure 7
figure 7

Calculated lattice parameters and cell volume with respect to increasing Zr concentration

Microstructural characterisation of ceramic PZT with 5 vol.% BaTiO3 seeds

All untextured ceramic samples synthesised via the conventional solid-state route were found to have densities ≥ 95% theoretical. Figure 8 shows the grain microstructure and morphology of bulk specimens. No continuous pore channels can be observed, with only isolated pores along grain boundaries and triple points present. The grain morphologies are distinctly equiaxed. The appearance of certain large voids > 5 µm may be due to the dissolution of broken BaTiO3 template particles during sintering, similar to that seen in the textured ceramics in Fig. 5(c).

Figure 8
figure 8

(a) SE and (b) BSE image of bulk untextured PbyBa1-y(ZrzTi1-z)O3 microstructure

The lack of contrast in the BSE image in Fig. 8(b), excluding pores, suggests incorporation of Ba2+ throughout the PZT matrix and no remnant BaTiO3 grains. The average grain size calculated for untextured ceramics specimens was 9 µm.

Dielectric behaviour of ceramic PZT with 5 vol.% BaTiO3 seeds

The room temperature relative permittivity (εr), tan δ, TC, and Tr-t values are shown in Fig. 9. Figure 10 shows the temperature dependencies of εr and tan δ for both tetragonal and rhombohedral compositions.

Figure 9
figure 9

(a) Room temperature εr and tan δ and (b) TC and Tr-t with respect to Zr concentration

Comparison with undoped MPB PZT, where x = 0.52, indicates that ~ 5 vol.% seeds depress the TC by ~ 40 °C. This agrees with trends seen in other studies on Ba-doped PZT across the whole solid solution range [35]. TC also decreases with increasing Zr:Ti ratio, from 356 to 324 °C for 0.49 ≤ z ≤ 0.58. The increase in εr on the tetragonal side can be attributed to increased contributions of ferroelectric domain wall vibrations as the bulk composition approaches the MPB. The maximum εr just on the tetragonal side of the MPB agrees with the previous literature on PZT [9]. There is a marked drop in εr for 0.52 ≤ z ≤ 0.54 which continues across the MPB into the rhombohedral phase field.

All rhombohedral compositions show a secondary dielectric anomaly identified as a rhombohedral–tetragonal phase transition (Tr-t) which increases in temperature with increasing Zr:Ti ratio in Fig. 10(c). The Tr-t first appears in the composition z = 0.54 which agrees with the lattice structure characterisation that indicates this is the rhombohedral side of the MPB region. Since the z = 0.54 composition is so closely situated near the MPB region, the energy required to induce a rhombohedral–tetragonal structural distortion is at a minimum and results in a Tr-t of 160 °C. As the structure becomes increasingly rhombohedral with increasing Zr concentration from z = 0.54 to z = 0.58, the Tr-t increases up to 313 °C following the trend representative of many relaxor-based PbTiO3 materials on the rhombohedral side of the MPB [36]. The trend in Tr-t is also reflected in the tan δ data in Fig. 10(d) as structural distortions in the perovskite unit cell near the Tr-t phase transition temperatures contribute to changes in unit cell polarisation and dielectric losses.

Figure 10
figure 10

(a, c) Temperature dependenence of εr and (b, d) tan δ for tetragonal 0.49 ≤ z ≤ 0.53 and rhombohedral 0.54 ≤ z ≤ 0.58 compositions with general formula PbyBa1-y(ZrzTi1-z)O3

Ferroelectric hysteresis of ceramic PZT with 5 vol.% BaTiO3 seeds

Figure 11 shows polarisation hysteresis and strain hysteresis loops for PZT ceramic samples with the optimum composition z = 0.53 and PZT-5A1. Both the optimised composition and PZT-5A1 exhibit squared ferroelectric polarisation loops and ‘butterfly’ strain loops. Saturation, Ps, and remnant, Pr polarisation values for z = 0.53 (Ps = 41.8 µC/cm2, Pr = 37.1 µC/cm2) are comparable to the PZT-5A1 specimen (Ps = 43.1 µC/cm2, Pr = 34.1 µC/cm2). The Ec for the z = 0.53 composition is slightly decreased at 9.3 kV/cm compared to PZT-5A1 variant at 9.7 kV/cm because the energy barrier to ferroelectric switching is observed to be lower on the rhombohedral side of the MPB than on the tetragonal side. Strain-field data give a maximum strain, Smax, of 0.25% for x = 0.53.

Figure 11
figure 11

(a) Polarisation vs. electric field (P-E) hysteresis loops for Pb0.947Ba0.053(Zr0.531Ti0.469)O3 and PZT-5A1 variant, and (b) strain vs. electric field (S-E) hysteresis loops

Piezoelectric properties

Quasi-static Berlincourt d33 measurements are shown in Fig. 12. Piezoelectric charge coefficients show a general improvement compared to conventional undoped PZT. There is an increase in d33 moving from tetragonal compositions towards the MPB, to a maximum d33 = 453 pC/N for z = 0.53, comparable to commercial ‘soft’ PZT compositions [37]. The enhanced d33 in randomly oriented ceramics with solid solution incorporation of 5 vol.% BaTiO3 seeds with MPB-like compositions is ascribed to a lowering of the TC, allowing for easier polarisation rotation and facilitating domain wall motion [38].

Figure 12
figure 12

Berlincourt d33 coefficients for randomly oriented ceramics with Zr concentration 0.49 ≤ z ≤ 0.58

Conclusions

Proof of concept for the fabrication of 94% [001]c fibre-textured PZT ceramics was established using 5 vol.% BaTiO3 seeds. Dissolution of BaTiO3 into the PZT matrix resulted in a tetragonal structural shift away from the MPB region. Crystal structure characterisation and mapping of the temperature-sensitive phase transition behaviour of BaTiO3-modified PZT ceramics indicated that the MPB region shifts slightly towards the rhombohedral side of the PZT phase space. A series of BaTiO3-modified compositions of PZT (PbyBa1-y(ZrzTi1-z)O3) was explored where the final Zr concentration ranged from z = 0.49 to z = 0.58. The dielectric, ferroelectric, and electromechanical behaviour of the modified compositional space was established to lay the groundwork for future developments focused on texturing PZT ceramics. An initial PZT matrix composition of Pb(Zr0.56Ti0.44)O3 resulted in an identified MPB composition of Pb0.947Ba0.053(Zr0.531Ti0.469)O3 exhibiting ‘soft’ PZT behaviour with a d33 = 453 pC/N similar to that of commercial PZT-5A1 with d33 ~ 410 pC/N and comparable polarisation/strain hysteresis data. This suggests that initial PZT matrix compositions with increased Zr:Ti ratios are desirable when compensating for the incorporation of BaTiO3 seeds to ultimately achieve near MPB-like properties.