Ultrahigh piezoelectric performances in soft lead zirconate titanate/polydimethylsiloxane composites by ethanol-assisted freeze casting

Abstract

Flexible materials capable of transducing mechanical signals into electrical responses underpin advances in wearable systems, soft robotics, and underwater sensing technologies. Ceramic–polymer piezoelectric composites can combine high electromechanical activity with mechanical compliance, but controlling phase distribution and architecture remains challenging. Here, we report an ethanol-assisted freeze-casting strategy that endows soft lead zirconate titanate/polydimethylsiloxane composites with an anisotropic architecture, enabling both high piezoelectric performance and excellent mechanical deformability. The composites achieve a high piezoelectric charge coefficient of 275 pC·N⁻¹, a piezoelectric voltage coefficient of 233 mV·m·N⁻¹, and an energy-harvesting figure of merit of 64.1 pm²·N⁻¹. The composites also exhibit robust mechanical flexibility, sustaining bending with radii below 5 mm, twisting, compressive deformation, and tensile strains up to 60%. The ethanol-assisted route suppresses freeze-induced cracking and improves three-dimensional compositional uniformity. This method offers a scalable and practical route for fabricating high-performance flexible piezoelectric composites for next-generation sensing technologies.

The authors report ethanol-assisted freeze casting of flexible PZT/PDMS composites with enhanced piezoelectric performance, improved structural uniformity, and excellent mechanical deformability, underscoring their potential for sensing technologies.

Introduction

Flexible piezoelectric materials have attracted significant interest for wearable electronics, soft robotics, underwater acoustics, mechanical energy harvesting and conversion, and soft sensing systems, where devices must conform to complex surfaces while maintaining high electromechanical responsiveness^1–4. However, an inherent trade-off exists between mechanical compliance and strong piezoelectric activity of piezoelectric materials. Traditional organic polymer-based piezoelectric films, such as poly(vinylidene fluoride) (PVDF) and its copolymers, exhibit good bendability. However, their piezoelectric charge coefficients are generally low, and the elastic stretchability remains limited, which restricts their practical applications^5,6. In contrast, dense ferroelectric ceramics, such as lead zirconate titanate (PZT) and potassium-sodium niobate (KNN), offer high piezoelectric charge coefficients but are intrinsically brittle and rigid⁷. Moreover, the inherently high dielectric permittivity of ferroelectric ceramics also suppresses piezoelectric voltage output under mechanical excitation, limiting their effectiveness in sensing applications^8,9. These intrinsic limitations have motivated the development of ceramic–polymer piezoelectric composites, in which ferroelectric ceramics are combined with linear polymer matrices to integrate strong piezoelectric activity and favorable mechanical properties^10,11.

Embedding high-performance ceramic fillers within polymer matrices is a widely adopted strategy to achieve enhanced piezoelectric performance while maintaining favorable mechanical properties^12–15. In particular, PZT fillers with diverse geometries, including particles^16,17, disks¹⁸, microfibers¹⁹, and nanofibers²⁰, have been widely incorporated into polymer matrices to tailor electromechanical properties of composites, with their orientation and spatial distribution critically governing performance²¹. Polymer matrices are selected based on the targeted mechanical requirements, with epoxy resins typically employed in applications demanding high stiffness and modulus²², whereas elastomers, such as PDMS, are preferred for imparting flexibility and mechanical compliance. However, the discontinuous nature of dispersed ceramic fillers limits interparticle coupling, reduces poling efficiency, impedes stress transfer, and suppresses the effective piezoelectric response^23–25. Aligned porous ferroelectric ceramics fabricated via freeze casting offer a compelling alternative^26,27. During unidirectional solidification, ice templating drives particle redistribution to form continuous lamellar channels, which promote more efficient poling and piezoelectric activity^28–31. However, conventional water-based freeze casting often suffers from macroscopic cracking due to significant volumetric expansion and uncontrolled dendritic ice growth^32–34. In particular, the low solid loadings required to achieve high porosity not only promote crack formation but also compromise the structural integrity and mechanical stability of the skeleton, making it difficult to maintain a continuous framework during subsequent processing. Freezing parameters, such as cooling rate and suspension composition, further influence the formation of these defects^35–37. These challenges have prompted the exploration of more effective processing strategies. For example, alternative solvents, such as camphene and tert-butyl alcohol (TBA), have been employed to fabricate porous ceramic materials due to their minimal volumetric change and reducing stress accumulation during freezing. However, camphene’s dendritic solidification produces branched pore networks that hinder the long-range alignment required for efficient poling^38,39. Its high melting point (~50 °C) imposes practical constraints, as slurry preparation must be performed at elevated temperature, limiting the choice of additives and complicating processing. TBA offers tunable pore architectures through adjustment of the TBA–water ratio and possesses a substantially higher saturated vapor pressure, allowing much faster sublimation than ice under identical freeze-drying conditions, which reduces energy consumption^40,41. Nevertheless, concerns over neurotoxicity and environmental impact limit its suitability for large-scale or industrial applications.

Previous studies indicate that water–ethanol mixtures can mitigate freeze-induced volume expansion and suppress dendritic ice growth, enhancing structural integrity^42,43. For example, water–ethanol systems have been employed in the preparation of LAGP/PEO composite skeletons for solid-state lithium electrolytes, enabling the formation of mechanically robust ion-conducting frameworks⁴⁴. They have also been utilized to produce high-performance bacterial cellulose aerogels that serve as drug carriers and hemostatic materials⁴⁵. However, ethanol-assisted freeze casting has not yet been explored for the fabrication of piezoelectric materials, and the effects of water–ethanol casting solvents on the structure and performance of piezoelectric composites remain unclear.

In this work, we introduce an ethanol-assisted freeze-casting approach to fabricate structurally robust PZT skeletons. After PDMS infiltration, the resulting PZT/PDMS composites combine excellent flexibility with markedly enhanced piezoelectric, ferroelectric, and hydrostatic performance. The preparation procedure is schematically illustrated in Fig. S1. The assistance of ethanol moderates dendritic ice growth and mitigates solvent expansion during freezing, while reducing supercooling and promoting the formation of continuous, well-aligned lamellar pore channels. This results in significantly improved alignment and continuity of the PZT phase along the freezing-gradient direction, effectively suppressing macroscopic cracking. A systematic comparison of composites prepared via original water-based and ethanol-assisted routes demonstrates that the ethanol-assisted process affords markedly enhanced piezoelectric and ferroelectric properties, together with improved structural uniformity. For clarity, samples utilizing water and water–ethanol as casting solvents are denoted as “original” and “EtOH-assisted,” respectively. Notably, the EtOH-assisted composite achieves exceptional piezoelectric performance at a PZT volume fraction of only 17.0 vol%, delivering a piezoelectric coefficient d₃₃ of 275 pC N⁻¹, voltage coefficient g₃₃ of 233 mV m N⁻¹, and piezoelectric energy-harvesting figure of merit FOM₃₃ of 64.1 pm² N⁻¹. The corresponding hydrostatic parameters reach a high d_h of 155 pC N⁻¹, g_h of 132 mV m N⁻¹, and hydrostatic figure of merit FOM_h of 20.4 pm² N⁻¹, outperforming those of the original water-based counterparts. Furthermore, flexible piezoelectric sensors were fabricated from the optimized composites to evaluate their device-level response, demonstrating promising capability for self-powered pressure sensing, real-time motion monitoring, and underwater acoustic sensing. These results demonstrate that ethanol-assisted freeze casting is an effective strategy for high-performance flexible piezoelectric systems and a scalable pathway toward next-generation deformable electromechanical devices.

Results

Phase structure and microstructure

The particle-size distribution and crystal structure of the raw PZT powder were characterized and shown in Fig. 1. As seen in Fig. 1a, with additional powder characterization provided in Figs. S2 and S3, the particle-size distribution exhibits a nearly unimodal profile, with d₁₀ ≈ 0.50 μm, d₅₀ ≈ 0.76 μm, and d₉₀ ≈ 1.19 μm, indicating a relatively narrow size distribution and good powder uniformity, which facilitates the formation of stable and uniformly dispersed slurry. XRD patterns of the raw powder and sintered porous skeleton (Fig. 1b) correspond to a single perovskite PZT phase, with no detectable secondary phases in the 20°–80° range, demonstrating that the intrinsic crystal structure of PZT remains unchanged after freeze-casting and sintering processes. The sharper peaks observed in the sintered skeleton indicate enhanced crystallinity due to grain growth during sintering.

Fig. 1. Structural characteristics of PZT powders, freeze-cast skeletons, and PZT/PDMS composites.

a Particle size distribution of the raw PZT powder. Error bars represent the standard deviation from three independent measurements. b X-ray diffraction (XRD) patterns of the raw PZT powder and sintered porous PZT skeleton. SEM images of the EtOH-assisted freeze-cast green body before sintering, showing the top view (c) and side view (d). e SEM image of the EtOH-assisted sintered PZT skeleton (side view), together with that of the original sintered skeleton for comparison. f SEM images of PZT/PDMS composites prepared by the original and ethanol-assisted routes (side view). Corresponding EDS elemental mappings (Pb, Ti, Zr, and Si) of the EtOH-assisted composite, confirming the homogeneous infiltration of PDMS into the porous PZT skeleton. Original and EtOH-assisted denote samples prepared using water and water–ethanol as casting solvents, respectively.

As revealed in Fig. 1c, d, freeze casting induces a distinct anisotropic lamellar architecture in PZT skeletons. During the freezing process, the growth direction of ice crystals is primarily governed by the imposed temperature gradient and the wetting behavior at the cold interface^46,47. In this work, a single cold source was applied at the bottom of the mold, causing ice crystals to preferentially propagate along the vertical freezing direction while remaining largely unconstrained in the horizontal plane. Figure 1c shows the morphology of the EtOH-assisted green body on the surface perpendicular to the freezing direction, and multiple lamellar regions with distinct in-plane orientations can be observed. Within each region, the lamellae exhibit nearly parallel alignment, whereas neighboring lamellar regions with distinct in-plane orientations are cross-linked. Figure 1d further presents cross-sectional SEM images obtained at different in-plane cutting angles along the freezing direction, which clearly reveal the pronounced anisotropy and high degree of lamellar alignment. Continuous plate-like pore channels extend along the freezing direction, with occasional interlamellar bridges connecting adjacent lamellae. This three-dimensional lamellar framework provides two key structural advantages. On one hand, the continuous lamellae aligned with the freezing direction ensure excellent longitudinal connectivity, enabling efficient transmission of the electric field and mechanical stress during poling and deformation. On the other hand, compared with the strictly parallel lamellae in an idealized 2–2 composite, the presence of cross-linked lamellar regions together with interlamellar bridging may help suppress cracking or delamination of the green body during drying and demolding, thereby improving its mechanical integrity. Such cross-linked lamellar architecture is expected to homogenize the stress distribution and promote long-range stress transfer²⁶, which is beneficial for fully exploiting the intrinsic piezoelectric properties of the ceramic phase. The microstructures of sintered skeletons are shown in Fig. 1e, indicating that the lamellar architecture is well preserved after sintering. Notably, compared with the skeleton produced from the original water-based slurry, the EtOH-assisted skeleton exhibits well-organized pore walls, higher alignment, and improved structural regularity.

To characterize the microstructure and polymer infiltration of the PZT/PDMS composites, cross-sections of original and EtOH-assisted PZT/PDMS composites were examined by SEM and EDS (Fig. 1f). Compared with the original composite, the EtOH-assisted composite exhibits more ordered continuous interconnection along the freezing direction, resulting in a more regular lamellar structure. This contrast is consistent with the orientation and connectivity of the corresponding PZT skeletons shown in Fig. 1e. This structural difference is further supported by X-ray computed tomography (CT) (Fig. S5), which reveals more continuous and better-aligned lamellar channels in the EtOH-assisted composites. Quantitative analysis based on the micro-CT data further shows higher values of the structural regularity parameter for the EtOH-assisted composites (Fig. S6 and Supplementary Note 1), together with increased lamellar thickness and interlamellar spacing relative to the original samples (Fig. S7). Notably, the 5 wt% EtOH-assisted composite exhibits the highest structural regularity parameter, whereas a slight decrease is observed at 10 wt% ethanol, suggesting that excessive ethanol addition may partially reduce structural regularity. EDS maps of both composites reveal a complementary spatial distribution of ceramic-related elements (Pb, Ti, and Zr) and the polymer-related Si signal, with Si mainly located in the interlamellar regions (Figs. 1f and S4). This consistent distribution, together with the CT observations (Figs. S5 and S8), confirms that PDMS effectively penetrates and fully fills the PZT skeletons in both systems, yielding dense composite structures without noticeable voids.

The PZT volume fraction in PZT/PDMS composites was determined from density measurements using the Archimedes method. Assuming a two-phase composite, the PZT volume fraction φ_PZT was calculated as

$${\phi }_{\mathrm{PZT}}=\frac{\left({\rho }_c-{\rho }_{\mathrm{PDMS}}\right)}{\left({\rho }_{\mathrm{PZT}}-{\rho }_{\mathrm{PDMS}}\right)}$$ 1

where ρ_c is the measured density of the composite, and ρ_PDMS and ρ_PZT are the densities of PDMS and dense PZT (1.0 and 7.8 g cm⁻³, respectively). With increasing PZT solid loading in the freeze-casting slurry, ${\phi }_{\mathrm{PZT}}$ increased accordingly. The PZT volume fraction ranges from 17.1 to 38.8 vol% in the original composites and from 13.6 to 39.7 vol% in the EtOH-assisted composites.

Dielectric, ferroelectric, and piezoelectric properties

Figure 2a–c summarizes the dielectric properties of poled PZT/PDMS composites with different PZT volume fractions. For both the original (Fig. 2a) and EtOH-assisted composites (Fig. 2b), the dielectric constant increases monotonically with increasing PZT content. Within the 1–10⁴ Hz range, the dielectric constant gradually decreases as the frequency increases. Owing to the strong insulating capability of the PDMS matrix, the dielectric loss remains low (tanδ < 0.1), significantly lower than that of dense PZT ceramics (Fig. S9). Above 10⁴ Hz, slight fluctuations appear in both dielectric constant and dielectric loss, presumably arising from interference caused by thickness-mode resonances activated after poling. As shown in Fig. 2c, at comparable PZT volume fractions, the EtOH-assisted composites exhibit slightly higher dielectric constant at 1 kHz than the original composites, consistent with the enhanced ceramic continuity produced by the EtOH-assisted freeze-casting process. Notably, the dense PZT ceramic exhibits a much larger dielectric constant than either composite series.

Fig. 2. Dielectric and ferroelectric properties of PZT/PDMS composites with different PZT volume fractions.

a Dielectric constant and dielectric loss of the original composites. b Dielectric constant and dielectric loss of the EtOH-assisted composites. c Dielectric constant at 1 kHz for the original and EtOH-assisted composites, with dense PZT ceramic included for comparison. d Polarization–electric field hysteresis loops of the original composites. e Polarization–electric field hysteresis loops of the EtOH-assisted composites. f Maximum polarization (P_max) and remanent polarization (P_r) of the original and EtOH-assisted composites. Original and EtOH-assisted denote composites prepared using water and water–ethanol as casting solvents, respectively.

According to effective medium theory, ideal 2–2 or 1–3 architectures subjected to electric fields along the ceramic alignment direction follow a parallel mixing behavior. Under this assumption, the effective dielectric constant ${\epsilon }_{\mathrm{parallel}}$ of the composite can be described as:

$${\epsilon }_{\mathrm{parallel}}={\phi }_{\mathrm{PZT}}{\epsilon }_{\mathrm{PZT}}+\left(1-{\phi }_{\mathrm{PZT}}\right){\epsilon }_{\mathrm{PDMS}}$$ 2

where φ_PZT is the PZT volume fraction, and ε_PZT and ε_PDMS refer to the dielectric constants of dense PZT ceramic and PDMS, respectively. However, the experimentally measured dielectric constants of both freeze-cast composites are substantially lower than the values predicted by this parallel model. This deviation is primarily ascribed to microstructural imperfections within the porous PZT skeleton, which reduce the effective ceramic connectivity, along with local electric-field inhomogeneities associated with the complex PZT/PDMS architecture. Importantly, a reduced dielectric constant can be beneficial for enhancing the piezoelectric voltage response of the composites.

Beyond dielectric behavior, the ferroelectric response of the composites is equally critical, as it directly governs their polarization dynamics and ultimately dictates the attainable piezoelectric performance. Figure 2d, e show the polarization–electric field (P–E) hysteresis loops of the original and EtOH-assisted composites under the electric field of ±6 kV mm⁻¹ at 10 Hz. All samples exhibit obvious ferroelectric hysteresis behavior, demonstrating that ferroelectric domain switching occurs throughout the investigated composition range. Normally, in ferroelectric materials, the squareness of a P–E hysteresis loop, representing the fraction of polarization retained after removal of the external field, is commonly quantified by the ratio $\frac{P_{\mathrm{r}}}{P_{\mathrm{s}}}$. Here, P_s is the saturated polarization, and P_r denotes the remanent polarization after the external electric field is removed. In this work, the maximum polarization P_max, calculated as $\frac{∣P_{\max +}∣+∣P_{\max -}∣}{2}$ at ±6 kV mm⁻¹, is taken as the index for the saturated polarization. Accordingly, the ratio $\frac{P_{\mathrm{r}}}{P_{\max }}$ is used as the indicator of loop squareness, as shown in Fig. S10. Notably, even for the EtOH-assisted composite with only 13.6 vol% PZT, $\frac{P_{\mathrm{r}}}{P_{\max }}$ reaches 0.76, indicative of a relatively square and well-developed hysteresis loop despite the low ceramic content, implying that ferroelectric domain switching has already been activated. Moreover, the EtOH-assisted composites exhibit consistently higher $\frac{P_{\mathrm{r}}}{P_{\max }}$ values than the original samples across the entire investigated range of PZT volume fraction. Correspondingly, their P–E loops are more saturated and square-shaped, with higher polarization levels, demonstrating that ethanol assistance facilitates more complete and reversible ferroelectric switching. As shown in Fig. 2f, both the maximum polarization P_max and remanent polarization P_r for the EtOH-assisted composites increase nearly linearly with ceramic volume fraction and remain substantially higher than those of the original composites. In particular, at high ceramic contents of approximately 30–40 vol%, the P_max and P_r values of the EtOH-assisted composites are nearly twice as high as those of the original composites, highlighting the significant enhancement in effective polarizable volume and domain-switching efficiency achieved through EtOH-assisted freeze casting.

To further understand the origin of the enhanced ferroelectric performance, finite element simulations based on the reconstructed microstructures from X-ray CT were performed (Supplementary Note 2 and Fig. S11). The results show that the EtOH-assisted composites exhibit a higher average electric field within the PZT phase than the original composite, indicating more effective electric-field concentration in the aligned lamellar ceramic framework. The constructed continuous and well-aligned lamellar ceramic channels could promote more uniform and stable local electric-field distribution during polarization, thereby enhancing cooperative domain switching and improving polarization efficiency. Consequently, the microstructure engineered via EtOH-assisted freeze casting substantially optimizes polarization pathways and internal electric-field distribution, enabling the composites to achieve significantly higher polarization strengths at equal or lower ceramic loadings.

Piezoelectric performance is a key parameter for evaluating piezoelectric materials. Various parameters are employed to evaluate the sensing and energy-harvesting performance of piezoelectric materials^48,49. The longitudinal piezoelectric charge coefficient d₃₃ quantifies charge generation, while the piezoelectric voltage coefficient ($g_{33}=\frac{d_{33}}{{\epsilon }_0{\epsilon }_{\mathrm{r}}}$) determines the voltage output, which is essential for pressure sensing. For piezoelectric pressure sensors, the open-circuit voltage is expressed as $V_{\mathrm{OC}}=g_{33}\cdot \sigma \cdot t$, where $\sigma$ is the applied stress, and t refers to the thickness of the piezoelectric layer. This implies that materials with higher d₃₃ and lower ε_r generate larger voltage signals, thereby enhancing sensor sensitivity. The energy-harvesting figure of merit FOM₃₃, defined as ${FOM}_{33}=d_{33}\cdot g_{33}=\frac{d_{33}^2}{{\epsilon }_0{\epsilon }_{\mathrm{r}}}$, reflects the capability to convert mechanical stimuli into electrical output in the longitudinal 33 mode. Figure 3 summarizes the evolution of the piezoelectric charge coefficient d₃₃, voltage coefficient g₃₃, energy-harvesting figure of merit FOM₃₃ and the thickness-mode electromechanical coupling coefficient k_t versus PZT volume fraction. As shown in Fig. 3a, d₃₃ increases steadily with increasing PZT content in both composite systems, with the EtOH-assisted samples consistently outperforming their original counterparts. With PZT loading of only 17.0 vol%, the EtOH-assisted composite achieves a d₃₃ of 275 pC N⁻¹, which is approximately 68% of the value of dense PZT ceramics (403 pC N⁻¹). When the PZT content reaches 39.7 vol%, the d₃₃ further increases to 413 pC N⁻¹, slightly exceeding that of the dense ceramic. This behavior can be attributed to the EtOH-assisted freeze-cast lamellar skeleton, which provides an efficient load-transfer network and facilitates effective activation of the ceramic phase. In addition, the compliant PDMS matrix in the composite may weaken lateral clamping and thereby enhance the longitudinal piezoelectric response⁵⁰. The effect of ethanol content on the piezoelectric response was further examined. As shown in Fig. S12, the 5 wt% EtOH-assisted composites deliver the highest d₃₃ among the investigated ethanol contents, further supporting the selection of 5 wt% ethanol as an optimal composition. This result is consistent with the structural analysis above, which shows that 5 wt% ethanol yields the highest structural regularity.

Fig. 3. Piezoelectric properties of PZT/PDMS composites with varying PZT volume fractions.

a Piezoelectric charge coefficient d₃₃, b piezoelectric voltage coefficient g₃₃ and c piezoelectric energy-harvesting figure of merit FOM₃₃ of the original and EtOH-assisted composites. d Comparison of FOM₃₃ and d₃₃ values obtained in this work with those of previously reported piezoelectric materials, including PVDF-based polymers, polymer-matrix ceramic composites, porous ceramics and dense ceramics. e Effective thickness-mode electromechanical coupling coefficient k_t,eff of the original and EtOH-assisted composites. Dense PZT ceramic is included for comparison in subgraphs (a–c, e). Original and EtOH-assisted denote composites prepared using water and water–ethanol as casting solvents, respectively.

The piezoelectric voltage coefficient g₃₃ characterizes the electric field generated under applied mechanical stress. As illustrated in Fig. 3b, g₃₃ of the original composites decreases monotonically with increasing PZT volume fraction, correlating with the sharp increase of dielectric permittivity at higher ceramic loadings seen in Fig. 2c. This trend reflects the typical dilemma between dielectric permittivity and voltage sensitivity in ceramic–polymer composite systems. However, for the EtOH-assisted composites, g₃₃ displays the non-monotonic trend versus PZT loading, achieving the highest value of 242.5 mV m N⁻¹ at 15.1 vol% PZT. This is ascribed to the fact that, at lower PZT contents, the limited d₃₃ restricts voltage output, whereas the rapidly increased permittivity suppresses g₃₃ with high PZT loading. Over the entire range of PZT volume fraction, the EtOH-assisted composites maintain higher g₃₃ than the original composites. For comparison, the dense PZT ceramic exhibits a much lower g₃₃ of 17.7 mV m N⁻¹ due to their extremely high permittivity. These results highlight that the EtOH-assisted freeze-cast composites are capable of achieving a favorable combination of high d₃₃ and low permittivity even at relatively low ceramic loadings, giving rise to an optimal g₃₃.

The energy-harvesting figure of merit (${FOM}_{33}=d_{33}\cdot g_{33}$) is obtained to evaluate the energy harvesting performance and shown in Fig. 3c. For the original composites, FOM₃₃ remains near 25 pm² N⁻¹, with a minor peak of 28 pm² N⁻¹ at 29.6 vol% PZT. In contrast, the EtOH-assisted composites exhibit significantly enhanced FOM₃₃ within the range of 15.1–18.3 vol% PZT, reaching a pronounced maximum of approximately 64 pm² N⁻¹ at 17.0 vol%, which is 226 and 896% of that of the original composites and dense PZT ceramics, respectively. This enhancement arises from the favorable balance between high d₃₃ and moderate permittivity at appropriate ceramic contents, demonstrating the effectiveness of the EtOH-assisted freeze-casting strategy in improving longitudinal energy-conversion efficiency.

Figure 3d and Table S1 compare the d₃₃–FOM₃₃ characteristics of the present composites with representative piezoelectric materials, including PVDF-based polymers, polymer–ceramic composites, porous ceramics, and dense ceramics. The original composites reside in the region of moderate d₃₃ and moderate FOM₃₃, whereas the EtOH-assisted composites shift toward the region of high d₃₃ and high FOM₃₃. Particularly, the optimal FOM₃₃ of EtOH-assisted composites outperforms that of many previously reported piezoelectric materials.

In addition, the thickness-mode electromechanical coupling coefficient (k_t) serves as a key indicator of how effectively a material converts mechanical energy into electrical energy. For the composite systems, the calculated k_t values were regarded as effective thickness-mode electromechanical coupling coefficients k_t,eff. k_t,eff was determined from the resonance (f_r) and anti-resonance (f_a) frequencies extracted from the impedance spectra (Figs. S13–S15), and was calculated using the following equation:

$$k_{\mathrm{t},\mathrm{eff}}=\sqrt{\frac{\pi }{2}\cdot \frac{f_{\mathrm{r}}}{f_{\mathrm{a}}}\cdot \tan \left(\frac{\pi }{2}\cdot \frac{f_{\mathrm{a}}-f_{\mathrm{r}}}{f_{\mathrm{a}}}\right)}$$ 3

As shown in Fig. 3e, the EtOH-assisted composites display higher k_t values than the original composites, which confirms that EtOH-assisted freeze casting enhances the effective electromechanical coupling in the thickness-extensional mode, supporting its potential use in vibration-energy harvesters and thickness-mode transducers.

Generally, for applications, such as underwater acoustic detection and static-pressure sensing, the hydrostatic piezoelectric response is a critical performance metric. The hydrostatic piezoelectric response is influenced by several parameters, including the longitudinal and transverse piezoelectric coefficients as well as the dielectric constant. The hydrostatic charge coefficient is defined as $d_{\mathrm{h}}=d_{33}+2d_{31}$, where d₃₁ is the transverse piezoelectric coefficient. The hydrostatic voltage coefficient $g_{\mathrm{h}}=\frac{d_{\mathrm{h}}}{{\epsilon }_0{\epsilon }_{\mathrm{r}}}$ directly determines the material-level sonar sensitivity. For a piezoelectric element of thickness t (in m), g_h (in Vm N⁻¹), and the intrinsic sensitivity M (in dB) can be estimated as $M=20\log \left({10}^{-6}{g}_{\mathrm{h}}t\right)$⁵¹. The hydrostatic sensitivity figure of merit (${FOM}_{\mathrm{h}}=d_{\mathrm{h}}{g}_{\mathrm{h}}=\frac{d_{\mathrm{h}}^2}{{\epsilon }_0{\epsilon }_{\mathrm{r}}}$) comprehensively evaluates performance under hydrostatic loading by considering both pressure sensitivity and transducer signal-to-noise characteristics, serving as a key metric for evaluating piezoelectric materials in underwater acoustics and hydrostatic pressure sensing applications⁵¹. The behavior of the transverse piezoelectric coefficient d₃₁ was first studied. The radial-mode impedance spectrum of the dense PZT ceramic (Fig. S16) was used to extract its d₃₁ value from the resonance and anti-resonance frequencies. As shown in Fig. 4a, the transverse piezoelectric coefficient d₃₁ is negative, with its magnitude increasing monotonically with PZT volume fraction. Based on the relation $d_{\mathrm{h}}=d_{33}+2d_{31}$, the corresponding hydrostatic piezoelectric charge coefficient d_h was calculated and shown in Fig. 4b. With increasing PZT fraction, d_h increases for both composite systems. Specifically, the EtOH-assisted composites consistently exhibit higher d_h values than the original ones, indicating more efficient charge generation under hydrostatic loading.

Fig. 4. Hydrostatic piezoelectric properties of PZT/PDMS composites with varying PZT volume fractions.

a Transverse piezoelectric coefficient d₃₁, b hydrostatic piezoelectric charge coefficient d_h, c hydrostatic piezoelectric voltage coefficient g_h, d hydrostatic energy harvesting figure of merit FOM_h of the original and EtOH-assisted composites. Dense PZT ceramic is included for comparison in subgraphs (a–d). e Comparison of FOM_h and d_h values obtained in this work with those of previously reported piezoelectric materials, including PVDF-based polymers, polymer-matrix ceramic composites, porous ceramics and dense ceramics. Original and EtOH-assisted denote composites prepared using water and water–ethanol as casting solvents, respectively.

The hydrostatic piezoelectric voltage coefficient g_h shows an overall decreasing trend with increasing PZT content (Fig. 4c), and the EtOH-assisted samples consistently outperform their original counterparts at comparable PZT volume fractions. Subsequently, the hydrostatic sensitivity figure of merit FOM_h is obtained and displayed in Fig. 4d. For both composite systems, FOM_h shows a non-monotonic dependence on PZT loading, peaking at an intermediate composition. The EtOH-assisted composites achieve a maximum FOM_h of 20.4 pm² N⁻¹ at 17.0 vol% PZT, which is more than three times of 6.1 pm² N⁻¹ at 29.6 vol% PZT in the original composites, confirming the advantages of the EtOH-assisted strategy. Notably, as seen in Fig. 4c, d, the g_h and FOM_h values of both series of composites remain significantly higher than those of dense PZT ceramics. This enhancement is ascribed to the lamellar PZT skeleton, which reduces lateral electromechanical coupling, while the low-modulus PDMS matrix provides mechanical damping that suppresses lateral load transfer, thereby enabling efficient hydrostatic piezoelectric conversion.

Figure 4e and Table S2 compare the d_h and FOM_h values obtained in this work with those of previously reported piezoelectric materials. It is shown that the original composites locate in the region of moderate d_h and moderate FOM_h. In contrast, the EtOH-assisted samples simultaneously achieve high d_h and high FOM_h, outperforming previous reported materials. Their exceptional hydrostatic piezoelectric performance places them among state-of-the-art piezoelectric materials, implying the advantage and potential for hydrostatic sensing and underwater acoustic detection.

Mechanical flexibility

Apart from electrical properties, the mechanical flexibility is equally important for piezoelectric materials in conformal applications. Therefore, the mechanical properties of the EtOH-assisted composite containing approximately 16 vol% PZT, which exhibits superior electrical performances, were systematically evaluated. Figure 5a shows the fabricated strip-shaped sample with uniform geometry. The sample maintains its structural integrity at a bending radius of approximately 5 mm and withstands large-angle twisting without fracture under manual manipulation, evidenced in Fig. 5b, c. These observations indicate that the lamellar PZT skeleton, encapsulated and supported by the PDMS matrix, enables cooperative deformation without cracking. Supplementary Movie 1 further visualizes repeated bending and shows that no macroscopic cracks or damage occur even after multiple cycles of large-curvature bending and recovery, highlighting the excellent flexibility and deformation tolerance of the composite.

Fig. 5. Mechanical flexibility and compositional homogeneity of PZT/PDMS composites.

a Photograph of the strip-shaped EtOH-assisted composite sample. b Demonstration of the bending flexibility of the EtOH-assisted composite. c Demonstration of twisting flexibility of the EtOH-assisted composite. d Lateral compression cyclic tests of the EtOH-assisted composite sample, conducted at 10, 20, 30, 40, 50, 60, and 70% strain, with 11 cycles applied at each strain level at a compression speed of 10 mm min⁻¹. e Longitudinal compression cyclic tests of the EtOH-assisted composite sample, conducted at 10, 20, 30, 40, 50, 60, and 70% strain, with 11 cycles applied at each strain level at a compression speed of 10 mm min⁻¹. f Lateral tensile stress–strain curve of the EtOH-assisted composite measured at a stretching speed of 5 mm min⁻¹. g XY position-dependent PZT volume fraction in the original composite sample (36 mm × 60 mm). h XY position-dependent PZT volume fraction in the EtOH-assisted composite sample (36 mm × 60 mm). i Z position-dependent comparison of PZT volume fraction between the original and EtOH-assisted composite sample (height = 36 mm). Original and EtOH-assisted denote composites prepared using water and water–ethanol as casting solvents, respectively.

To further assess the structural stability of the composite under different principal loading directions, the material was machined into regular cuboid specimens for cyclic compression tests along the transverse direction (perpendicular to the freezing temperature gradient) and the longitudinal direction (parallel to the gradient). As shown in Figs. 5d and S17, under transverse compression, the stress–strain curves of cycles 2–11 nearly coincide when the target strain is below 60%, following the initial preloading cycle 1. This high degree of overlap indicates that the composite maintains excellent structural integrity and cyclic stability during repeated compression, demonstrating good cyclic stability and fatigue resistance. When the strain exceeds 60%, pronounced buckling features and structural failure are observed in the sample. Longitudinal compression produces a different mechanical response, as shown in Figs. 5e and S18. At 10% strain, the longitudinal compressive stress is higher than the transverse stress. This result indicates that the continuous lamellar PZT framework carries and transfers external loads more effectively when compressed along the lamellar direction, resulting in a higher apparent modulus. When the strain exceeds 10%, the stress grows more slowly with further strain, suggesting the onset of local structural damage. Beyond 50% strain, a significant divergence in the cyclic curves is observed alongside the emergence of macroscopic cracks. Therefore, longitudinal loading is more prone to induce mechanical failure within the composite architecture.

The tensile deformability of the composite was also characterized. As shown in Fig. 5f, the samples sustain an engineering strain exceeding 60% in the lateral direction, and the stress–strain curve increases steadily until fracture. This behavior indicates that the lamellar PZT skeleton produced via EtOH-assisted freeze casting is fully infiltrated by the PDMS matrix and forms strong interfacial adhesion. In contrast, the tensile extensibility in the longitudinal direction is lower, with a reduced fracture strain (Fig. S19). Consequently, this unique lamellar composite architecture endows the material with tensile ductility that surpasses that of most piezoelectric composites. In addition, the electrode integrity after repeated deformation is further verified in Fig. S20.

In general, the EtOH-assisted PZT/PDMS composite not only achieves excellent piezoelectric performance but also exhibits outstanding deformation capability under bending, twisting, compression and tension. The transverse direction shows significantly better deformability and cyclic stability than the longitudinal direction, making it advantageous for flexible devices that require large bending or conformal deformation, whereas the longitudinal direction can be exploited for highly sensitive sensing.

Three-dimensional compositional uniformity

Compositional uniformity is critical for practical applications. To quantify the effect of ethanol addition on the macroscopic compositional uniformity of the composites, samples (36 mm × 60 mm × 7 mm) were sectioned in the XY plane, and the local density of each subregion was measured to determine the corresponding PZT volume fraction (φ_PZT). As shown in Fig. 5g, the original samples exhibit pronounced in-plane fluctuations, with φ_PZT ranging from 8.9 to 28.0 vol%. In contrast, the EtOH-assisted samples display a much more uniform distribution in Fig. 5h, with φ_PZT confined to a narrow range of 15.0–16.6 vol%. The macroscopic morphologies of samples are shown in Fig. S21. Penetrating cracks can be observed in the green bodies of the original sample in Fig. S21a, b, and the corresponding sintered PZT skeleton in Fig. S21c exhibits more severe cracking, some of which extend through the entire thickness. The regions containing these apparent defects correspond to locally reduced PZT volume fractions. By comparison, in Fig. S21d–f, the EtOH-assisted green bodies and sintered skeletons are nearly crack-free, indicating that EtOH-assisted freeze casting improves the in-plane compositional uniformity of the PZT phase and enhances the structural integrity of the composites. Notably, as shown in Fig. S22, similar improvements in structural uniformity and crack suppression are observed in other ceramic systems, including BaTiO₃, TiO₂, and Al₂O₃, suggesting the general applicability of the ethanol-assisted freeze-casting strategy. To further assess compositional uniformity along the freezing direction, 36-mm-thick composites were sectioned along the Z direction, and φ_PZT was analyzed at different heights (Fig. 5i). The original samples exhibit a clear vertical gradient, with φ_PZT decreasing from 28.2 vol% at the bottom to 16.0 vol% at the top. In contrast, the EtOH-assisted samples show much smaller variations, with φ_PZT confined to 16.1–18.4 vol%. These results demonstrate that ethanol enhances both in-plane and through-thickness compositional uniformity of PZT within the polymer matrix, achieving compositional uniformity in three dimensions. Moreover, the EtOH-assisted route also exhibits good reproducibility, and the d₃₃ values of samples prepared in four independent batches remain highly consistent (Fig. S23).

The underlying reasons for the improved compositional uniformity are as follows. Firstly, the addition of ethanol lowers the freezing point and reduces supercooling (Fig. S24), resulting in a more stable freezing front and facilitating the formation of aligned lamellar pore channels⁵². Secondly, pure water generally undergoes about 9% volumetric expansion during freezing, and this expansion is constrained by the mold and the surrounding slurry. During bottom-up solidification, the confined expansion generates residual stress, which in turn promotes penetrating cracks and upper–lower compositional differences⁵³. In the EtOH-assisted system, ethanol contracts upon freezing and can partially offset the expansion of water, thereby reducing internal stress accumulation⁵⁴. Thirdly, a water-5 wt% ethanol mixture behaves as a typical binary solvent. Its freezing process does not occur at a single temperature but proceeds through a solid–liquid coexistence region. At the early stage of freezing, the water phase solidifies first and forms lamellar ice crystals, while the remaining liquid gradually becomes ethanol-rich^55–57. These lamellar ice crystals act as geometric templates for the packing of PZT particles, and the ethanol-rich liquid serves as a soft lubricating layer. This liquid phase redistributes local stresses during ice formation and volume changes, reducing macroscopic internal stress and suppressing crack formation. It also retains limited fluidity, allowing PZT particles to migrate or rearrange along liquid channels rather than being instantly immobilized. Such a moderated rearrangement would promote the formation of a more uniform, continuous and ordered ceramic skeleton, and the increased structural order may be tightly associated with the enhanced piezoelectric performance. In short, the EtOH-assisted freeze-casting strategy enables superior three-dimensional compositional uniformity and structural integrity at the centimeter scale, demonstrating high reproducibility and potential for scalability. For flexible piezoelectric devices, large-area sensing arrays demand stringent material uniformity and batch-to-batch stability. The EtOH-assisted process thus offers a practical and scalable route for producing large-area, high-throughput, and performance-stable PZT/PDMS piezoelectric composites.

Sensing performance

To evaluate the feasibility of the EtOH-assisted PZT/PDMS composite for self-powered sensing applications, piezoelectric composite sensors were fabricated, as schematically illustrated in Fig. 6a, with the fabrication process shown in Fig. S25. The electromechanical response of the sensors was systematically evaluated under cyclic compressive loading. Figure 6b–d present the open-circuit voltage, short-circuit current, and transferred charge signals under different applied forces at a loading frequency of 3 Hz. The amplitudes of all output signals increase monotonically with increasing force, indicating a typical piezoelectric response. Compared with the original composite (Fig. S26), the EtOH-assisted composite exhibits significantly enhanced output signals, further highlighting the advantage of the EtOH-assisted structure. Forward/reverse connection and unpolarized-state tests (Figs. 6e and S27) show that the signals reverse polarity upon switching connections, while the unpolarized sample exhibits negligible output, confirming that the response originates from the intrinsic piezoelectric effect rather than extrinsic contributions. The sensor also exhibits stable output over a frequency range from 0.5 to 10 Hz (Fig. S28) and excellent durability under repeated loading. As shown in Fig. 6f, the open-circuit voltage remains highly stable over 15,000 loading cycles at 3 Hz, with the average peak voltage showing only a slight variation from 2.16 V in the first 10 cycles to 2.13 V in the final 10 cycles.

Fig. 6. Electromechanical sensing performance of the EtOH-assisted PZT/PDMS composite-based piezoelectric sensor.

a Schematic illustration of the sensor structure and electrical measurement configurations under forward and reverse connections. b Open-circuit voltage, c short-circuit current, and d transferred charge of the composite under cyclic compressive loading with different applied forces at 3 Hz. e Output voltage signals under forward connection, reverse connection, and in the unpolarized state. f The durability of the sensor evaluated from the open-circuit voltage over 15,000 loading cycles at 3 Hz. g Linear fitting of the output voltage as a function of applied pressure. h Real-time voltage response during holding and releasing of a ball. i Real-time voltage response during finger tapping and releasing.

The relationship between output voltage and applied pressure was analyzed in detail. As shown in Fig. 6g, the sensor exhibits a piecewise linear response. In the low-pressure region (<40 kPa), a high voltage sensitivity of 140 V MPa⁻¹ is obtained (R² = 0.99). At higher pressures (40–200 kPa), the output voltage continues to increase approximately linearly, although the sensitivity decreases to 108 V MPa⁻¹. This reduction in sensitivity at higher loading levels may be associated with partial micro-scale damage or local structural degradation within the brittle PZT lamellae, which lowers the effective electromechanical coupling. As shown in Fig. S29, the EtOH-assisted composite exhibits consistently higher pressure sensitivity than the original composite across both pressure ranges. In addition, the generated charge shows a linear dependence on applied force for both composites (Fig. S30), allowing the effective piezoelectric charge coefficients to be estimated. In the low-pressure region, the extracted values are in reasonable agreement with the quasi-static d₃₃ results, further supporting the reliability of the measured electromechanical response.

Benefiting from the high sensitivity and mechanical flexibility, the composite demonstrates strong potential for real-time motion monitoring. As shown in Fig. 6h, distinct voltage signals are generated during the holding and releasing of a ball. Under finger tapping (Fig. 6i), the sensor produces clear and repeatable responses, further demonstrating its applicability in wearable sensing systems. In addition, the EtOH-assisted PZT/PDMS composite was evaluated for underwater acoustic sensing (Fig. S31 and Supplementary Note 3), exhibiting a clear acoustic response and potential for flexible hydrophone applications. In short, the EtOH-assisted PZT/PDMS composite-based sensor integrates high piezoelectric voltage sensitivity, good linearity, frequency stability, and long-term durability, demonstrating strong potential for flexible sensing applications.

Discussion

This work established an EtOH-assisted unidirectional freeze-casting strategy to fabricate flexible PZT/PDMS piezoelectric composites. The results demonstrate that introducing a small fraction of ethanol effectively modulates ice-templating behavior, enabling the formation of highly aligned and crack-suppressed lamellar PZT skeletons. After PDMS infiltration, the resulting composites exhibit markedly improved centimeter-scale structural uniformity and homogeneous ceramic–polymer integration. Moreover, the EtOH-assisted composite achieves a longitudinal piezoelectric coefficient d₃₃ of 275 pC N⁻¹, a voltage coefficient g₃₃ of 233 mV m N⁻¹, and an energy-harvesting figure of merit FOM₃₃ of 64 pm² N⁻¹. It further delivers hydrostatic piezoelectric coefficients of d_h = 155 pC N⁻¹ and g_h = 132 mV m N⁻¹, together with a high hydrostatic figure of merit (FOM_h = 20.4 pm² N⁻¹). In addition, the composite exhibits excellent mechanical compliance, sustaining bending with radii below 5 mm, large-angle twisting, repeated compression, and tensile strains above 60%, depending on the loading direction. The EtOH-assisted composite also demonstrates promising self-powered sensing performance, together with potential for motion monitoring and underwater acoustic applications. Ethanol-assisted freeze casting offers a practical, environmentally benign, and scalable pathway for producing flexible ceramic–polymer piezoelectrics with outstanding electromechanical performance, mechanical compliance, and sensing potential. This strategy also provides broadly applicable insights for designing next-generation ceramic–polymer architectures for wearable electronics, soft robotics, and underwater electromechanical technologies.

Methods

Materials

Lead zirconate titanate (PZT) powders were purchased from Bangci Electronic Technology (Yancheng, China). Polyvinyl alcohol (PVA; alcoholysis degree 87–89 mol%, viscosity 80–110 mPa s) was obtained from Aladdin. Polydimethylsiloxane (PDMS; Sylgard 184) was supplied by Dow Corning and used with a base-to-curing-agent mass ratio of 10:1. Ethanol was purchased from Shanghai Titan Scientific Co., Ltd. All materials were used as received without further purification.

Preparation of PZT/PDMS composites

PZT/PDMS composites were fabricated through a unidirectional freeze-casting process, as illustrated in Fig. S1. Composites with water or water-5 wt% ethanol mixture as casting solvents were respectively prepared. Freeze-casting slurries were formulated by dissolving 3 wt% polyvinyl alcohol at 90 °C, followed by dispersing PZT powders and ball milling at 260 rpm for 1 h to ensure homogenization. The slurries were then poured into molds with a copper base and polytetrafluoroethylene (PTFE) sidewalls. Unidirectional freezing was induced by immersing the copper base into liquid nitrogen, enabling upward solidification. During the freezing process, the advancing ice front expelled PZT particles and generated aligned lamellar ice templates. The frozen samples were then demolded and freeze-dried under a pressure below 10 Pa for 72 h to sublimate ice without structural collapse, yielding porous PZT green bodies with oriented lamellar channels. The green bodies were sintered in a muffle furnace at 850 °C for 2 h, producing mechanically robust porous PZT skeletons, which were subsequently infiltrated with PDMS prepolymer (base/curing agent = 10:1, w/w) under vacuum conditions. After curing at 50 °C for 4 h, the final PZT/PDMS composites were obtained. The composites were machined into cuboids using a computer numerical control mill, coated with silver paste on both sides, and then cured at 120 °C for 20 min to form electrodes. Finally, the samples were polarized in silicone oil under an electric field of 4 kV mm⁻¹ for 20 min at room temperature. The poling conditions were selected based on the results in Fig. S32 and Supplementary Note 4.

For comparison, the preparation and characterization of dense PZT ceramic references are described in Supplementary Note 5 and Figs. S33, S34.

Characterization

The crystal structure was analyzed by X-ray diffraction (XRD, Rigaku D/max-2550, Japan) using Cu Kα radiation (λ = 1.5406 Å). The microstructure and elemental mapping were investigated using a field-emission scanning electron microscope (FE-SEM, ZEISS Merlin Compact, Germany) equipped with an energy dispersive X-ray spectroscopy (EDS) system. The three-dimensional microstructure of the composites was characterized by X-ray CT with an X-ray microscope (Xradia 620 Versa, Carl Zeiss, Germany). The particle size distribution of the PZT powder was measured using a laser particle size analyzer (Mastersizer 2000, Malvern Instruments Ltd., UK). The dielectric properties were measured using a broadband dielectric spectrometer (Novocontrol Technologies GmbH & Co. KG, Germany) over the frequency range from 1 to 10⁵ Hz. The polarization–electric field (P–E) hysteresis loops were measured with a TF Analyzer 1000 (AixACCT Systems, Germany) at 10 Hz. The resonance frequency, anti-resonance frequency, and electromechanical coupling coefficient were extracted from the resonance spectra measured with an impedance analyzer (Keysight E4990A, USA). The piezoelectric coefficients d₃₃ and d₃₁ were measured using a quasi-static piezoelectric meter (ZJ-4AN, Institute of Acoustics, Chinese Academy of Sciences, China). Mechanical tensile and compressive tests were performed on the composites using a universal testing machine (Shimadzu AGS-X, Japan). The electrical outputs were recorded using an electrometer (Keithley 6514, USA). Differential scanning calorimetry (DSC) curves were acquired using a DSC Q2000 (TA Instruments) at a cooling rate of 5 °C min⁻¹ under a nitrogen atmosphere to determine the solidification temperature of the slurry.

Author contributions

Y.X., L.Y. and Y.S. conceived the idea and designed the research. Y.X., L.Y. performed material fabrication, freeze-casting experiments, and electrical and mechanical characterizations, and drafted the manuscript. M.Z. performed the finite-element simulations. E.X. assisted with ferroelectric measurements. M.Z., X.L. and M.Y. contributed to data visualization and figure preparation. Y.X., L.Y. and P.H. contributed to data analysis and experimental guidance. Y.S. supervised the project, provided critical insights, and revised the manuscript. All authors contributed to data interpretation, analysis, and manuscript preparation.

Peer review

Peer review information

Nature Communications thanks Jinwook Yeo and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

Y.S. discloses support for the research of this work from the National Natural Science Foundation of China (grant nos. 52421001, 92566301, 52572133, and 52302149).

Data availability

All data supporting this study and its findings are available within this paper and its Supplementary Information. The data corresponding to this study are available from the corresponding authors upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains Supplementary material available at 10.1038/s41467-026-73758-x.