Stretchable piezoelectric elastic composites for sensors and energy generators

The search for a piezoelectric elastomer that gener ates an electrical signal when pressed and stretched has increased significantly in the last d eca e as they hold great promise in harvesting energy from human motion and monitoring human a ctivities. Here, the excellent elasticity of polydimethylsiloxane-based elastomers and the pi ezoelectric properties of lead zirconate titanate (PZT) were combined and, using a thermally a ctivated poling process, elastic piezoelectric composites were obtained. For this, two pol ydimethylsiloxane (PDMS) matrices with a molar mass of 139 kDa and 692 kDa and PZT fillers w ith particle sizes of 2 and 20 μm were used. For the same poling conditions, an increase i n the piezoelectric response with increasing amount of filler, filler size and molar mass of the polymer matrix was observed. Overall, d33 and d31 values of 2.7−40 pC·N -1 and 16−48 pC·N were achieved in this work with filler contents ranging from 37−72 vol.%. A composite mate ri l with a PZT filler content of 38 vol.% (20 μm particle size) in a commercially avail able PDMS with a Mw =139 kg·mol -1 exhibited high flexibility, good elasticity with long -term mechanical stretchability and high lon-


Introduction
Piezoelectric materials have the characteristic of generating an electrical signal by an induced mechanical stress and vice versa, described by the converse and direct piezoelectric effect.
They have been used in a wide range of applications such as ultrasound tranducers, sensors, actuators and energy harvesters. [1,2] Piezoelectricity can be found in both organic and inorganic materials, whereas ferroelectric ceramics exhibit by far the highest response. For instance, perovskite ceramics such as barium titanate (BTO) [3,4] or lead zirconate titanate (PZT) [5,6] have a high d * 33 piezoelectric coefficient, high temperature stability and a high dielectric constant, which render them attractive for energy conversion materials. Below their Curie temperature T c , BTO (T c = 130 °C) and PZT (T c ≥ 360 °C) feature a tetragonal or rhombohedral structure with a non-symmetric distribution of the central atom. To obtain a piezoelectric effect in a ferroelectric ceramic with randomly distributed grains, the polarization within different grains must be oriented as much as possible in the same direction. This can be achieved by subjecting the ferroelectric materials to a strong electric field via a so-called poling process, where it is assumed that the microscopic polar domains within the crystal structure are reoriented and a permanent polarization is introduced in the macroscopic sample, leading to a high piezoelectric response of the poled material. However, ceramics are brittle and therefore their application in flexible electronics is limited. Although more flexible alternatives such as poled polylactic acid [7], polyvinylidene difluoride [8,9] or disperse red 1-copolymethylmetacrylate composite in polydimethylsiloxane (PDMS) [10,11] exist, their piezoelectric response is relatively weak compared to ferroelectric ceramics. To increase the piezoelectric properties of polymeric materials, often a ferroelectric powder is mixed inside a polymeric matrix such as lead zirconium titanate (PZT), barium titanate or bismuth ferrite. [2,[12][13][14] Sebastian et al. demonstrated that even flexible polymer-based composite fibers with a 60 vol.% of ferroelectric fillers could be polarized by applying an electrical field up to 4 kV·mm -1 . [15] In order to achieve elasticity, ceramic particles can be dispersed in an elastic matrix such as PDMS, thus leading to mechanically stable materials. Such composites can undergo high deformation strain, which is an interesting feature for energy conversion from human motion, monitoring human activities and stretchable electronics. [16][17][18] In recent years, implantable piezoelectric biomedical devices which were powered by motions in the human body such as breath and heartbeat into electrical energy were developed which could pave the way for sustainable future health care technologies. [19][20][21] In this work, PZT particles of two different sizes are dispersed in two different high molecular weight PDMS matrices and the composites are poled in an electric field to enhance their respective piezoelectric properties, e.g., the d * 33 and d * 31 coefficients. These two coefficients represent the ability of a material to convert a pressure applied in poling direction (d* 33 ) or a tensile stress perpendicular to the poling direction (d* 31 ) into electric charge, which is exploited in sensor applications. We achieved highly stretchable materials with a filler content of 37−66 vol.%, which show a d * 33 value of 2.7−40 pC·N -1 and d * 31 value of 16−48 pC·N -1 and therefore exhibit attractive properties for sensors and energy harvesters. While the d * 33 values of PZT/PDMS composites have been extensively reported previously and used as a basis for sensor development, the d * 31 measurements of the corresponding composites was hardly studied, hence the potential of exploiting this feature of PZT/PDMS for energy harvesting was an integral part of this investigation. [2,12,[22][23][24][25][26] Additionally, stability measurements were conducted to check the performance of selected composites after subjecting them to different environmental treatmetns such as humidity and temperature. Furthermore, we demonstrated that the longitudinal piezoelectric effect of the prepared composite materials can be utilized to construct a working prototype sensor.

Synthesis of high molecular weight linear silanol end-functionalized PDMS
D 4 was introduced in a reaction vessel equipped with mechanical stirrer and cooling mantle.
Concentrated H 2 SO 4 was added as a catalyst and the reaction mixture was vigorously stirred for 60 min at 25 °C, after that distilled water (2 mL) was added and the stirring was continued for another 30 min. The mixture was stored as such for 18.5 h. The polymer was dissolved in a large amount of CHCl 3 ( ~ 2.5L), followed by repeatedly washed with an aqueous solution 5% Na 2 CO 3 and warm water In order to reach a neutral pH, followed by the removal of CHCl 3 at the rotary evaporator. The polymer was further dried in a vacuum oven at 150 °C for several hours in order to remove residual cycles which were present as side products of the reaction.

PDMS composites
The trimethylsilyl surface-modified PZT powder (7−24 g) dispersed in toluene (~10 mL) by tip sonication for 2 minutes was mixed with PDMS 139k or 692k (1.5 g) and (25-35% methylhydrosiloxane)-dimethylpolysiloxane (150 µL) in the three-roll-mill (Exakt, type 80 S) for 10 min. Thereafter, the composite was redispersed in toluene, stirred at room temperature for aproximately 2 hours, treated with a 50% solution of DBTDL in toluene (112 µL). The mixture was then bath sonicated for 5 minutes, and then casted onto a fluorinated polymer substrate using a doctor blade. The composites were let to stand at room temperature overnight and then heated in vacuo for 7 hours at 80 °C before further investigations. The composites had a film thickness of 170−250 μm (PDMS 139k) and 120−150 μm (PDMS 692k).

Characterization Methods
SEM images were obtained with a Nova NanoSEM 230 FEI and the investigated samples were sputtered with Au prior to the measurement. TGA was conducted with a Perkin Elmer TGA7 at a heating rate of 20 °C min −1 under a He gas flow. Impedance spectroscopy measurements were performed at room temperature with a Novocontrol high impedance Alpha Analyzer using a Hewlett Packard 16451B dielectric test fixture equipped with round electrodes on circular composite samples with round sputtered gold electrodes (20 mm diameter, 20 nm thickness) on both sides. The measurements were performed in a frequency range between 0.1-10 6 Hz. The amplitude of the probing ac electric signal applied to the samples was 1V, and the permittivity ɛ' was determined from the capacitance: C=ɛ'ɛ 0 A/d, where A is the electrode area, d is the thickness of the film and ɛ 0 is the vacuum permittivity. The permittivity values were averaged from the deviation in the film thickness of the corresponding film sample. Tensile tests were performed using a Zwick Z010 tensile test machine with a crosshead speed of 50 mm·min -1 . Tensile test samples, with a gauge width of 2 mm and a gauge length of 18 mm, were prepared by die-cutting. The strain was determined using a longitudinal strain extensometer. Stress versus strain curves were obtained by averaging measurements of 3 to 5 samples for each material. The Young's modulus was determined from the slope of the stress-strain curves at ≤ 10 % strain. Dynamic mechanical analysis (DMA) was carried out on an RSA 3 DMA from TA Instruments. Strips of 10 mm × 20 mm from selected samples were measured under a dynamic load of 2 g, at 2% strain in the frequency range of 0.1-10 Hz at 25 °C. The mechanical loss factor tan(δ) is given as the fraction of imaginary and real storage modulus at 2% strain.
Gel permeation chromatography (GPC) measurements were done with an Agilent 1100/PSS WinGPC 8.1 system in THF (flowrate: 1 mL/min) calibrated with PDMS standards (Polymer Standards Service). The dielectric breakdown was determined using flat rigid metal electrodes with an area of 0.25 mm 2 embedded in an epoxy resin, which hindered electromechanical actuation of the dielectric films. Ten measurements were conducted for each sample.

Poling
The composite was sandwiched between two 15 mm (diameter) flexible electrodes, consisting of GNPs/CB (300/150 30 wt.%) in PDMS, and two 15 mm (diameter) round aluminum electrodes. The metal electrodes were then connected to a voltage source and to the ground. Most composites were poled in an electric field of 3−18 V·µm −1 at 150 °C for 40 min and afterwards cooled to room temperature at a rate of 4 °C min -1 with the field still switched on. This precaution was used to avoid the relaxation of the domains back to their random orientation.
The field was maintained by a Stanford Research Systems PS350 high voltage source. Temperature control was ensured by a Lauda RC6 CP thermostatic oil bath connected to a closed glass container that housed the composite.
The d * 33 piezoelectric coefficient was measured by applying a constant force (0.6 N) on the poled part of the composite, which is covered by the electrode material, that led to a voltage output. The measuring device was calibrated with a reference ceramic sample with a d * 33 of 17.5 pC·N -1 in order to calculate the resulting d * 33 coefficient. The samples measured had two conductive electrodes on them, which were applied according to a previous procedure. [28] For the d * 31 measurements, a Zwick Z010 tensile testing machine was used to apply strain and measure the force acting on the rectangular films with dimensions of 3.5 cm × 2.5 cm (length × width). The metallic clamps of the tensile testing machine were covered with a layer of adhesive tape to insulate them from the sample. A strip of adhesive copper tape was attached to both the upper and lower clamps to make an electrical connection between the sample and the voltage readout via a current preamplifier. Both sides of the poled area of the electrets were covered with conductive electrode composites consisting of GNPs, CB and PDMS (M w = 139 kg·mol -1 ) which were screen printed on both sides of the sample (Supporting information SI 2). The electrodes were 11 mm in diameter. A connective strip consisting of carbon black powder, going from the poled area to the sample edge, was applied on each side of the film, facing opposite edges. After mounting the sample in the tensile testing machine, the copper tape contacted the carbon black and a Keithley 2000 recorded the current passed through a Stanford Research Systems SR570 low noise current preamplifier while the sample was subjected to cyclic strains. Selected samples were subjected to further treatment in an oven set at 90 % relative humidity (RH) at 25 °C and at 85 °C, respectively, in order to asses their performance after exposure to different environments.

Results and Discussion
In this work, we used PZT particles with a diameter of 2 and 20 µm as a piezoelectric filler and two elastic PDMS matrices with different molar masses of 139 and 692 kDa. A schematic illustration of the preparation steps is depicted in Figure 1 Size measurements conducted from the SEM micrographs concluded that the average size of those agglomerates was around 21 ± 6.5 μm similar to the size given by the supplier. Prior to the preparation of the composites, it was essential to subject the particles to surface treatment with HMDS to render them hydrophobic, thus enhancing their compatibility with the PDMS matrix. [29,30] Through the surface functionalization, the PZT particles could be dispersed more easily in toluene and did not sediment as fast as the untreated particles ( Figure 2c).  The materials prepared as described in the experimental section are named as Ex[y-z], where x represents the mass of PDMS used in kDa, y the size of the particles used in µm, and z the vol.% of PZT filler. SEM micrographs of the resulting composites with different particle size and filler content are featured in Figure 3. The effective amount of filler dispersed in the PDMS was determined from the TGA measurements where the total filler content consisted of the residual mass after burning off the PDMS polymer ( Figure 4). The vol.% of PZT particles within the composites was calculated using the densities of PZT (ρ=7.6 g/cm 3 ), of the cross-linker (ρ=0.98 g/cm 3 ) and that of PDMS (ρ=0.97 g/cm 3 ). Under inert atmosphere, the PDMS matrix should decompose without residual mass. [30] In general, the composites tendentially feature an enhancement in the thermal degradation stability with increasing filler content since they act as heat absorbants which is transferred to them from the polymer chains.
Furthermore, the compatibilized particles also act as physical cross-linking centers for the PDMS chains through interfacial interactions which reinforce the material. [31] As a result, the temperature of maximum weight loss rate T max,m (m for major) of the main degradation step increased with the filler content. However, an increase of the filler content also hinders the chemical cross-linking reaction of polymers, and it was also reported that ceramics can act as catalysts, which promote the depolymerisation of PDMS at elevated temperatures. [32] Hence, T max.m decreased again towards lower temperatures at high filler content as observed for composites using both 2 and 20 μm PZT particles. In addition, a weak degradation step between 300 to 350 °C noted as the temperature of maximum weight loss rate T max,w (w for weak) appeared for composites with filler contents above 60 vol% although it was much weaker than T max,s which suggests the degradation of free polymer chains due to the inhibited chemical cross-linking reaction (Table 1). [33] The composites prepared with PDMS 139k using 2 μm PZT exhibited higher thermal stability than the composites prepared with PDMS 692k with regards to T max,m , likely due to the higher chemical cross-linking density of the former due to its lower molecular weight, which limits the thermally-induced segmental movement which was also supported by the increasing filler content that contributed to more physical cross-linking centres ( Figure 4). [34] Higher molecular weight polymer composites exhibit an increased probability of random scission of polymer chains as a result of intramolecular and intermolecular redistribution reactions which leads to depolymerisation. [35] This trend is not observed with the composites utilizing 20 μm PZT particles since their total surface area is lower and therefore does not inhibit the cross-linking reaction between the polymer chains as heavily as in the case of 2 μm PZT particles. However, the lower surface area also leads to a decrease in their ability to effective-ly act as physical cross-linking centers, and hence lower T max,m was overall observed for 20 μm PZT composites. [36]     The composites exhibited an enhanced dielectric permittivity, which is attributed to the high ε' of PZT (ε' = 2'100). The increase in ε' was almost linear with increasing filler content ( Figure  5, Table 2). Here, ε' seemed independent from both the particle size and the M w of the PDMS matrix. For both particle sizes of PZT used, ε' values up to 34−35 were reached at a filler content of 66 vol. %. The obtained ε' values are in comparable range to PZT/PDMS composites reported earlier and are in good accordance to the model proposed by Yamada. [5] Although the filler content in the composites prepared in this work surpassed the percolation threshold, ε'did not register an abrupt increase since, despite the high ɛ' of the filler, PZT is not a conductive filler like metal particles which contribute to this phenomenon. [37,38] When the filler content exceeded 70 vol. %, ε' diminished due to the increased presence of air voids (ε' = 1) with higher filler content due to particle agglomeration. [37] The dielectric breakdown field,  The composites were also subjected to mechanical tensile testing (Supporting information SI 4 and 5). With increasing filler content, the strain at break values for composites PDMS 139k decreased due to the increasing brittleness of the materials ( Figure 6, Table 3). There seemed to be no substantial difference between the smaller and larger particles in terms of the Young's modulus Y and the strain at break values although the difference in Y and strain at breakvalues manifested itself at larger filler content. By using PDMS 692k instead of PDMS 139k, the strain at break value was enhanced due to the higher flexibility of the polymer chains. For composites using both 2 μm and 20 μm PZT particles dispersed in both PDMS 139k and 692k, respectively, Y increased initially with the filler content due to the reinforcing effect of the filler as additional physical cross-linking agents. As the surface-to-volume ratio was higher for the 2 μm particles, the resulting Y values were higher than for the 20 μm PZT composites. However, a decrease in Y at high filler contents (50 vol% or above) was observed due to the increasing agglomeration of particles with rising filler content and the subsequent dewetting of the polymer at the polymer/particle interphase (Table 3, Entries 3 and 4, 6 and 7, 9 and 10, 12 and 13). [41,42] As a result, the reinforcing effect of the particles diminished due to the reduction of the polymer/particle interactions, and hence the mechanical properties deteriorated which was manifested in both the reduction of Y and the strain at break value. Below a filler content of 50%, the 20 μm PZT/PDMS composites for PDMS 692k at filler contents showcased a higher strain at break value which indicated that the chemical cross-linking was less hindered and the polymer chains less entangled compared to the scenario where the smaller PZT particles are used as filler where the latter limits the diffusion of the polymer chains and act as physical cross-linkers.  DMA measurements performed on selected composites with roughly the similar filler content revealed a constant storage modulus, E', from 0 to 1 Hz. It has to be mentioned that composites with higher filler content were not suitable for this characterization due to their increasing brittle nature. In general the storage modulus increased with the amount of filler, which is indicative of the reinforcing character of the filler thus contributing to the stiffness of the material (Figure 7). The increase in the filler amount is associated with an increase in the particle surface area and in the resulting polymer/particle interface. [43] Hence, the loss modulus also increased since the reinforcement of PDMS with the filler required a larger force to induce the same stress. With increasing frequency, a decline in the modulus was observed, which could be due to the diminishing entanglements between the particles and the polymer chains. A steeper decline in the loss modulus, E'', is observed for the composites using PDMS 692k since the cross-linking density is lower compared to the composites using PDMS 139k, thus giving rise to softer and more flexible materials. [44] Both composites E139  and E692  are materials where the adjacent particles start to become more interconnected with each other as a result of the growing filler content. The increase in the frequency led to the separation of the interconnected particles, and hence the polymer/particle interface area increased. In the case of E139 , the increase in the polymer/particle interface led to a brief increase in both the storage modulus E' and the loss modulus E'' due to the destruction of the particle agglomerates and the ensuing activity of the resulting isolated particles as physical cross-linkers. This phenomenon was not observed for E692  where the increase in the polymer-particle interface area was smaller due to the lower surface-to-volume ratio of the 20 μm PZT particles and therefore had a lower positive impact on the mechanical properties of the corresponding composites in comparison to 2 μm PZT particles. Therefore, the initial increase in frequency led to decline in both E' and E''.  Entries 1−3). It is well known for composites that elevated temperature and higher electric field results in a higher piezoelectric constant (due to a higher mobility of dipole movement of the ferroelectric filler), on the other hand the maximum temperature is limited by the melting point and decomposition temperature of the polymer. Furthermore the maximum allowed electric field (E) is determined by the dielectric strength of the material (E b ), where E < E b , and further decreases with increasing temperature. Therefore, some of the different composites were polarized using different conditions. The breakdown field, E B , is reduced with increasing filler content due to particle agglomeration as explained earlier. Since the E B of the composite E139  was at 13 V·µm -1 , poling at 12 V·µm -1 was quite risky in terms of inducing dielectric breakdown. Therefore, E139  was poled at 6 V·µm -1 at a temperature of 155 °C instead. As a result, this composite exhibited a d * 33 of 22.7 pC·N -1 ( Figure 8, Table 4, Entry 5). Increasing the molecular weight of the PDMS matrix allowed the incorporation of more filler, while the dielectric breakdown field was enhanced. Therefore, a composite E692  could be prepared and poled at 12 V·µm -1 . However, it was difficult to remove it from the electrodes after poling at 150 °C since it easily broke apart during the removal because of its brittleness. To overcome this problem, the poling temperature was reduced to 100 °C, which enabled a facile removal of the sample from the electrodes after poling. The piezoelectric response of this poled composite was d * 33 = 30 pC·N -1 . With this composite, it has also been illustrated that the poling temperature increased the d * 33 of the resulting composite, which is in agreement with the literature. [47] For instance, when E692  was poled at an electric poling field of 12 V·µm -1 and the temperature was increased from 24 °C to 159 °C, the d * 33 coefficients increased from 20.5 to 30 pC·N -1 (Table 4, Figure 8, Entries 10 and 11).
Aside from the poling temperature, the electric field with which the composites are poled also has an impact on the d * 33 . The increase in the electric field leads to an enhancement in the d 33 as shown by E139  where an increase of the d * 33 from 15.8 to 22.7 pC·N -1 was observed when the electric field was increased from 3 to 6 V·µm -1 at 155 °C to 159 °C (Table 4, Entries 4 and 5). In direct comparison, E139  with larger particles exhibited a d * 33 = 40 pC·N -1 when poled at 159 °C and 12 V·µm -1 (Figure 8, Table 4). This is in line with previous reports which stated that the d * 33 increases with the particle size in PZT composites. [25,46] In comparison to the permittivity ɛ' which featured a linear increase with filler content, d *

33
values were quite small until 30 vol.% which corresponds to the percolation threshold for isotropically dispersed particles. [37] Therefore, d * 33 exhibited a significant increase above the percolation threshold as the composites experience a transition from a 0−3 to 1−3 composite where adjacent particles became interconnected (Figure 3). [5] The 20 μm PZT composites showcased higher d * 33 since there are less polymer/particle multilayers along the film thickness due to the larger particle size, therefore stress is more effectively transferred between the PZT domains.   SI 6). [11] Similar to the d * 33 , the d * 31 piezoelectric coefficient of composites with similar filler content, poling conditions and identical strains increased with the PZT particle size. The difference in the performance of the two composites E139  and E139 [20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38] can be seen in the generated charges and voltages ( Figure 9). The cyclic strain measurements conducted for the composites using the 2 μm at 25 % strain and at a frequency of 0.1 Hz led to an output of 1.58 V or 0.9 nC which corresponded to a d* 31 of 16 pC·N -1 (Figure 9a-c) with a tensile force of 2.9 N acting on the E139 . When E139 [20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38] was subjected to the same measurements, the voltage generated from the cyclic measurements at the same strain and frequency increased from 1.58 V to 2.3 V, with the latter voltage corresponding to generated charges of 1.25 nC and a d * 31 value of 30 pC·N -1 , as a result of increasing the PZT particle size from 2 to 20 µm. The corresponding tensile force to strain E139 [20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38] was 3.02 N. The larger output exhibited by the composites with 20 μm PZT particles composed of agglomerates of primary PZT particles is attributed to the higher connectivity between the particles due to the shorter interparticle distance. [48] The reduced interface area between the particle and the polymer associated with the increase in the size of the particle leads to less imperfections in the joining the particles. [49] In short, different stresses act on the PZT particles depending on their size, and in the case of 20 μm PZT particles the shorter interparticle distance between the primary particles within the agglomerates lead to a higher stress at the contact point between particles. [17,18] Hence, the positive correlation between stress and piezoelectric properties is reflected in the higher d * 33 and d * 31 values for the composites using the 20 μm PZT particles. The long-term reliability of the films were confirmed since the generated charges remained within the similar regime when assessing the performance at different time intervals (Figure   9a and d). Furthermore, composites containing 20 µm PZT particles dispersed in PDMS 692k were also subjected to d * 31 measurements at similar filler content (43 vol.%) in order to determine the effect of the polymer matrix ( Figure 10). Due to the increase in structural defects with increasing filler content caused by partice agglomeration, composites with filler content above 50 vol.% could not be investigated for the performed d * 31 measurements. For the synthesis of E692  more toluene had to be used, hence the formulated films were thinner than the composites with PDMS 139k using the same amounts of reagents. The presence of defects in the PDMS 692k composites could lead to a higher rupture probability, therefore the films were initially stretched to 15% strain, which yielded a d * 31 of ~ 40 pC·N -1 from an output voltage of 1.96 V (Figure 10a and c) and a charge of 0.91 nC at a force of 1.23 N. When the film was stretched to 25%, the generated d * 31 surpassed 40 pC·N -1 and the output voltage of 3.08 V, the charge of 2.11 nC at a force of 1.87 N was obtained. Unfortunately, the film experienced rupture after a few days of measurement (Figure 10b and d). As shown earlier the PDMS 692k matrix reduced the Young's modulus of the composite materials compared to PDMS 139k matrix, which should lead to an increase of output current and hence also a larger d * 31 coefficient since a higher stress acts between adjacent particles due to the more efficient stress transfer to the particles from the external source. [18] As can be seen from the measurement the d * 31 coefficient increased from 30 to to 48 pC·N -1 by changing the matrix from PDMS 139k to PDMS 692k. This renders it more beneficial for applications because higher output currents can be generated, or less input force is needed. The results obtained for the d * 31 are higher as compared to d * 33 , which is usually not the case. We assume that the capacitor effect related to the change in charge densities of the composites plays a major role. Furthermore, the force to generate the desired strains are higher than the strains to which the composites were subjected to during the d * 33 measurements. During cycling the thickness of the composite films will change, which affects the interparticle distance as mentioned earlier since they were incorporated within the composites alone through physical cross-linking.
Hence, a difference in the charge density will occur.
Although the generated voltages are comparable to the output of other PZT composites, the higher strains which can be generated with the composites featured in this work expand their potential as dielectric elastomers used to generate electricity. [50,51] Figure 9. The d * 31 and charges generated by E139  at a strain of 25 %, the generated charges and voltage from 5 cycles when the sample was strained by 25% at 0.1 Hz, 88 days after poling (top row) and d 31 and charges generated by E139 [20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38] at a strain of 25 % and the generated charges and voltage from 6 cycles when the sample was strained by 25% at 0.1 Hz, 10 days or 225 h after poling (bottom row). The effect of the environment on the material performance was also investigated for selected composites. After exposure to 90% relative humidity (RH) for 24 hours, the d * 31 of E139 , at a strain of 25 %, was altered from 30 to 36 pC·N -1 , which corresponded to a change in the charge from 1.92 nC to 1.8 nC. The material softened after this treatment, therefore the d * 31 increased despite the reduction in the generated charge. The exposure of E139  to 80 °C for the same period of time saw the d * 31 slightly change from 18.2 to 20.5 pC·N -1 and a change in the charge from 0.97 nC to 1.09 nC ( Figure 11). Thus, it has been found that such composites have not shown any major changes in their piezoelectric d * 31 coefficients especially at enhanced temperatures below T c thus indicating their ability to be used under those selected environmental conditions. Composite E139 [20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38] generated a voltage of 2.3 V when stretched by 25% and has been found to exhibit the best compromise between piezoelectric performance and mechanical flexibility and stability. To demonstrate the potential of the newly developed material for applications, a demonstrator consisting of a clamping/contact device and a measuring device with LCD display was constructed ( Figure 12). The piezoelectric film sensor is clamped in the contact device and contacted via two circuit boards on the front and back, which con-nect the sensor to the measuring device via cables. The sensor signal is amplified and filtered in the measuring device using a "voltage mode amplifier". Then the signal is digitized with the ADC. The first line from the LCD shows the signal curve and in the second line the current peak value is recorded for a few seconds. The average offset voltage of the measuring circuit has already been subtracted from the peak value. Only the positive signal deflection is taken into account in the measurement. With the voltage amplitude, the signal influence by the deformation of the piezo film can be compared and assessed. In the present device, the piezo film signal is loaded with 500 kΩ and then amplified with 20 V. The measured value full scale (0023) corresponds to an amplitude of about 165 mV of the piezo film. For example, a displayed value of 100 corresponds to a voltage of 16 mV on the piezo with a load of 500 kΩ. Figure 12 shows how the mechanical stress induced by pressing a plastic knob by a finger generates an electrical signal, which is detected by the black electronic device. Figure 12. The components used for the construction of press sensor (a), photograph of the piezoelectric film with black conductive films on top and bottom (b), schematic representation of the press sensor in the relaxed (c) and press state (d) as well as the photograph of the sensor and the electronic de-vice used to monitor the generated electrical signal. A change in the Sig Peak from 0 to 0.023 can be seen for the relaxed (e) and pressed sensor (f).

Conclusions
Flexible PZT/PDMS composite films with a thickness of 120−250 µm were prepared, starting from two PDMS matrices of different molar mass and surface functionalized PZT particles of 2 µm and 20 µm. Elastic composites with a large content of filler were successfully prepared by the doctor blading technique. They were poled in a strong electric field at elevated temperature in order to induce a remanent polarization of the ferroelectric particles responsible for the macroscopic piezoelectric response. The poled composites containing 20 µm particles produced larger piezoelectric response coefficients (both d * 31 and d * 33 ) than the composites using 2 µm PZT particles at similar filler content. The potential of the most promising composite in a press sensor was demonstrated. Because of the good elastic and piezoelectric properties these materials are highly attractive for applications in sensing, energy harvesting, and stretchable electronics.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.