Development of Porous Piezoceramics for Medical and Sensor Applications

The use of porosity to modify the functional properties of piezoelectric ceramics is well known in the scientific literature as well as by the industry, and porous ceramic can be seen as a 2-phase composite. In the present work, examples are given of applications where controlled porosity is exploited in order to optimise the dielectric, piezoelectric and acoustic properties of the piezoceramics. For the optimisation efforts it is important to note that the thickness coupling coefficient kt will be maximised for some non-zero value of the porosity that could be above 20%. On the other hand, with a good approximation, the acoustic velocity decreases linearly with increasing porosity, which is obviously also the case for the density. Consequently, the acoustic impedance shows a rather strong decrease with porosity, and in practice a reduction of more than 50% may be obtained for an engineered porous ceramic. The significance of the acoustic impedance is associated with the transmission of acoustic signals through the interface between the piezoceramic and some medium of propagation, but when the porous ceramic is used as a substrate for a piezoceramic thick film, the attenuation may be equally important. In the case of open porosity it is possible to introduce a liquid into the pores, and examples of modifying the properties in this way are given.


Introduction
In many cases, functional materials are selected as a trade-off between a set of requirements. When these requirements are in contradiction, a composite could be a more optimal material, combining at least two materials with very different properties. The preferred structure of a composite depends on the physics governing the application as well as on manufacturability.
Porous materials can be considered composites where the secondary phase is air. Many examples are known from other fields, e.g., fur, wool etc. for thermal insulation, foam for mechanical damping, and biological bone structure for optimizing strength/mass. As will be apparent from this study, piezoelectric ceramics is a field where a number of significant advantages are obtained by introducing porosity. Before proceeding with a brief literature overview for porous piezoceramics, the terminology of composite electroceramics will be reviewed.
In a classical paper, Newnham et al. introduced a very practical notation for composites based on the concept of connectivity [1]. The connectivity of a phase is the number of dimensions in which it is fully connected and thus can take the values of 0, 1, 2 and 3. Furthermore, by convention, the active (e.g., piezoelectric) phase is mentioned first in the case of a diphasic composite. Some common examples of the use of Newnham's notation are 2-2 composites for multilayer structures, 0-3 for

Characterisation of Bulk Discs
Porous piezoceramic bulk samples of the material Ferroperm™ Pz37 [17] were prepared as described in the experimental section and a number of relevant properties were measured. To begin with, the density ρ is determined from the mass and geometrical dimensions. The relative permittivity in the 33 direction (ε33,r) is determined from the capacitance measured at 1 kHz and the geometric dimensions. The piezoelectric charge coefficient d33 has been measured directly, and the voltage coefficient has then been calculated from the following expression: The frequency constants for the same modes (Nt and Np, respectively) have been determined from the relevant resonance frequency as shown in the following example for the thickness mode (where h is the thickness): The coupling coefficients in the thickness and the planar mode (kt and kp, respectively) have been determined according to common equations given in reference [18] on the basis of impedance spectra. As described by Alemany et al. [19], lossy piezoelectric materials (such as porous piezoceramics) offer special challenges for characterisation using impedance spectra and a detailed study would involve determining dielectric, piezoelectric and mechanical properties as complex quantities, cf. [10]. However, this study is focused on properties with direct importance for pulse-echo applications so only mechanical losses are considered.
The mechanical quality factor Qm has a special significance for pulse-echo applications, since a high Qm (i.e., low mechanical loss) leads to prolonged "ringing" in the transducer after an excitation has been applied, which would be seen as noise in the measurement. It is commonly determined by two different methods, which often tend to give different results. One involves the minimum impedance |Z|min at resonance, the free capacitance C (conventionally measured at 1 kHz) and the resonance and antiresonance frequencies [18]

Characterisation of Bulk Discs
Porous piezoceramic bulk samples of the material Ferroperm™ Pz37 [17] were prepared as described in the experimental section and a number of relevant properties were measured. To begin with, the density ρ is determined from the mass and geometrical dimensions. The relative permittivity in the 33 direction (ε 33,r ) is determined from the capacitance measured at 1 kHz and the geometric dimensions. The piezoelectric charge coefficient d 33 has been measured directly, and the voltage coefficient has then been calculated from the following expression: The frequency constants for the same modes (N t and N p , respectively) have been determined from the relevant resonance frequency as shown in the following example for the thickness mode (where h is the thickness): The coupling coefficients in the thickness and the planar mode (k t and k p , respectively) have been determined according to common equations given in reference [18] on the basis of impedance spectra. As described by Alemany et al. [19], lossy piezoelectric materials (such as porous piezoceramics) offer special challenges for characterisation using impedance spectra and a detailed study would involve determining dielectric, piezoelectric and mechanical properties as complex quantities, cf. [10]. However, this study is focused on properties with direct importance for pulse-echo applications so only mechanical losses are considered.
The mechanical quality factor Q m has a special significance for pulse-echo applications, since a high Q m (i.e., low mechanical loss) leads to prolonged "ringing" in the transducer after an excitation has been applied, which would be seen as noise in the measurement. It is commonly determined by two different methods, which often tend to give different results. One involves the minimum impedance |Z| min at resonance, the free capacitance C (conventionally measured at 1 kHz) and the resonance and antiresonance frequencies [18]: This method is rather convenient, since all parameters except C can be determined from the impedance spectrum commonly used to determine the coupling coefficient. However, the value of |Z| min is rather sensitive to spurious modes.
The second one is directly based on the 6 dB peak width as seen from the conductance spectrum, where f 1 and f 2 can be found as the maximum and minimum, respectively, of the susceptance close to the resonance [19]: Though not being insensitive to spurious modes, this method has the advantage that the operator can clearly see if a spurious mode is disrupting the measurement. Finally, the acoustic impedance Z a is found as the product of density and sound velocity from the thickness mode: The common unit for Z a is the megarayleigh, with 1 MRayl = 10 6 kg/(m 2¨s ). Table 1 compares these properties of Pz37 before and after oil saturation to the two material types most commonly used for pulse-echo transducers: PZT 5A and K81 lead metaniobate. There are a number of points to note from these measurements. Starting with the density of Pz37, the density before oil infiltration indicates a porosity close to 25% (taking the theoretical density of PZT to be approximately 8 Mg/m 3 ). Upon oil infiltration the density increases by 3.3%, which indicates a considerable degree of open porosity. This agrees well with the observation that in this type of system, the connectivity changes from 3-0 to 3-3 in the porosity range from 20% to 30% [10].
The measured value of the relative permittivity of Pz37, 1058, is intermediate between that of the standard soft PZT and K81. In order to understand the rather large decrease in permittivity, a comparison is useful with the work of Wersing et al. [3], who considered porous piezoceramics with a wide range of porosity and 3-3 or 3-0 connectivity. It was found that the set of formulae of Bruggeman for permittivity as a function of porosity describe these cases quite well and that the two types of connectivity differ by less than 6% up to a porosity of 30%. Specifically for piezoceramics with 3-3 connectivity and porosity p up to 60%, the following approximation is valid for the relative permittivity: where ε(0) is the relative permittivity of the dense piezoceramic. Taking the latter value to be 1700 and the porosity to be 25%, yields an expected relative permittivity of 1063, which is in good agreement with the measured value. Turning now to the piezoelectric properties, it is worth mentioning that thanks to a high d 33 of the porous ceramics and the aforementioned reduced permittivity, Equation (1) yields a very high g 33 coefficient compared to the other two types listed in Table 1. This is important for sensor applications using a voltage amplifier.
The sensitivity at resonance relevant for the pulse-echo application is more adequately described by k t , and this is higher for Pz37 than for the dense soft PZT. It may appear counterintuitive that introducing an inactive phase could cause the thickness coupling coefficient to increase, but this has been observed earlier in comparisons with equivalent low-porosity discs [10,11] and confirmed by KLM-type modelling of composites of soft PZT and air, with connectivity ranging between 3-3 and 3-0. k t generally showed a non-monotonous dependence on porosity and for both types of connectivity, a significant increase was seen in a wide range of porosity [20]. The impedance spectrum close to the thickness resonance showed an interesting variation with oil infiltration. To begin with, before oil infiltration these particular discs (with an aspect ratio of 0.1) showed spurious modes at the thickness resonance. Two minima in the absolute impedance, of almost equal depth, were seen, the one at lower frequency yielding a surprisingly high k t « 0.61 and the other one yielding an average k t of 0.546. On the other hand, after oil infiltration, the first minimum (equivalent to k t « 0.61) was significantly less pronounced and a much cleaner resonance appeared yielding an average k t of 0.543. Our interpretation is that the true k t of the porous, dry material is close to 0.55 and that the introduction of oil might have damped the spurious mode. The observations reported here agree well with the findings in the interesting work on filling pores of piezoceramics with oil or araldite [21].
The planar coupling factor k p shows a rather different dependence on porosity, namely a significant reduction. This is good agreement with previous observations and correlates with tendencies seen for transverse modes (k 31 , d 31 ) [10]. For the pulse-echo application, planar coupling is undesired because harmonics of this may affect the thickness resonance, and indeed the very low k p of K81 is considered an important virtue in this case. The measured value of 0.357 for Pz37 is between that of K81 and the standard soft PZT, and even after oil infiltration the k p /k t ratio remains acceptable.
The most important point to note about the mechanical quality factors is that they are lower for Pz37 than for the standard soft PZT, at least for the thickness mode, and apparently exhibit a minor increase upon oil infiltration (although the Q m measurement showed some scatter the tendency was consistent for all samples). The frequency constant shows a larger increase with porosity for the thickness mode than for the planar one [3], and as expected, the porosity dependence is even stronger for the acoustic impedance (cf. Equation (5)). Z a of Pz37 is only slightly affected by oil infiltration and the values are very close to that of K81. The low acoustic impedance makes it easier to transfer acoustic energy to media such as water (Z a 1.5 MRayl) and oil.
In summary the porous piezoceramic Pz37 offers significant improvements over standard soft PZT in a number of respects and it is clearly relevant to test the performance in a pulse-echo transducer.

Transducer Characterisation
With the porous bulk discs described in the previous section, 5 transducers of each type were manufactured as described in the experimental section. There was one clear, strong thickness mode at the nominal frequency, indicating good acoustic matching (cf. discussion of spurious modes above). The pulse-echo response of two such transducers built with dry piezo-elements is shown in Figure 2a.
To begin with, some features of the curves should be pointed out. The strong negative signal (off-scale) at t < 5 µs is the excitation pulse, and between 5 and 10 µs the so-called main-bang ringdown is seen. This is the period where high acoustic energy reverberates in the piezo-element. The next feature, from 23 µs and slowly decaying for some 10 µs, is an echo coming from the backing material. A secondary backing echo starts at around 72 µs. The distinct peaks starting at around 51 µs are the target echo, which is of course the desired signal. The fact that the red and the grey curves show good agreement is a sign of reproducibility, and it should also be noted that the target echo is much stronger than the main-bang ringdown and the backing echoes. Figure 2b shows the pulse-echo response of two transducers with oil-saturated piezo-elements and in this case the two curves are nearly indistinguishable. Apart from this, a slightly higher amplitude in the main-bang ringdown can be seen. This could be associated with the apparent minor increase in the mechanical quality factors upon oil infiltration.
Materials 2015, 8, page-page material. A secondary backing echo starts at around 72 μs. The distinct peaks starting at around 51 μs are the target echo, which is of course the desired signal. The fact that the red and the grey curves show good agreement is a sign of reproducibility, and it should also be noted that the target echo is much stronger than the main-bang ringdown and the backing echoes. Figure 2b shows the pulse-echo response of two transducers with oil-saturated piezo-elements and in this case the two curves are nearly indistinguishable. Apart from this, a slightly higher amplitude in the main-bang ringdown can be seen. This could be associated with the apparent minor increase in the mechanical quality factors upon oil infiltration. In order to examine the bandwidth of the pulses in Figure 2, a Fourier analysis is performed, and the result is shown in Figure 3. The fractional bandwidth (i.e., the bandwidth found from the −6 dB points with respect to maximum amplitude, divided by the centre frequency) is above 100% for all four measurements, which is a very respectable performance for this application. A closer look reveals that the average of two similar transducers decreases from 114% to 105% as a result of the oil infiltration, and this is in good agreement with the slight decrease in kt noted in the previous section. In order to examine the bandwidth of the pulses in Figure 2, a Fourier analysis is performed, and the result is shown in Figure 3. The fractional bandwidth (i.e., the bandwidth found from the´6 dB points with respect to maximum amplitude, divided by the centre frequency) is above 100% for all four measurements, which is a very respectable performance for this application. A closer look reveals that the average of two similar transducers decreases from 114% to 105% as a result of the oil infiltration, and this is in good agreement with the slight decrease in k t noted in the previous section.  In summary, the results obtained with the oil-saturated transducers based on the porous Pz37 are very promising for pulse-echo applications at elevated pressure, such as downhole drilling.

Thick-Film Transducers for Ultrasonic Imaging
Since thick films are closely integrated with a substrate, the characterisation of functional properties of the material itself is significantly more complicated than in the case of separate bulk elements. Therefore, this section will deal with functional characterisation of thick-film transducers only, and as will be seen, the main difference in terms of pulse-echo measurements is the higher frequencies involved. For some typical properties of the InSensor™ thick film used, the reader is referred to [22]. Figure 4a is a close-up view of the target echo from a pulse-echo measurement (with reflector placed in focal point) performed on a single-element thick-film transducer for ultrasonic imaging such as the one described in Section 3.2 below. The echo is seen to be very well-defined, and the Fourier analysis shown in Figure 4b reveals a −6 dB fractional bandwidth of 130% and a centre frequency of 22.1 MHz. This is a very good performance for high-resolution applications such as ultrasonic skin imaging. For this transducer the porosity plays a similar role as described in the section on porous bulk discs: increasing kt for improved sensitivity; reducing the lateral coupling factor k31; reducing Qm and thus improving the bandwidth; and reducing the acoustic impedance. It should be mentioned that for ultrasonic imaging, high bandwidth combined with high frequency translates into high resolution [15]. Because of the thick film being integrated with the substrate as mentioned above, there are important constraints on the substrate. To begin with, it should be chemically and physically compatible with the thick film at the processing temperatures, and furthermore it should play the role of a backing material, being well-matched acoustically to the film and able to dissipate acoustic energy. For the thick films described in this work, a porous ceramic substrate has been used. In summary, the results obtained with the oil-saturated transducers based on the porous Pz37 are very promising for pulse-echo applications at elevated pressure, such as downhole drilling.

Thick-Film Transducers for Ultrasonic Imaging
Since thick films are closely integrated with a substrate, the characterisation of functional properties of the material itself is significantly more complicated than in the case of separate bulk elements. Therefore, this section will deal with functional characterisation of thick-film transducers only, and as will be seen, the main difference in terms of pulse-echo measurements is the higher frequencies involved. For some typical properties of the InSensor™ thick film used, the reader is referred to [22]. Figure 4a is a close-up view of the target echo from a pulse-echo measurement (with reflector placed in focal point) performed on a single-element thick-film transducer for ultrasonic imaging such as the one described in Section 3.2 below. The echo is seen to be very well-defined, and the Fourier analysis shown in Figure 4b reveals a´6 dB fractional bandwidth of 130% and a centre frequency of 22.1 MHz. This is a very good performance for high-resolution applications such as ultrasonic skin imaging. For this transducer the porosity plays a similar role as described in the section on porous bulk discs: increasing k t for improved sensitivity; reducing the lateral coupling factor k 31 ; reducing Q m and thus improving the bandwidth; and reducing the acoustic impedance. It should be mentioned that for ultrasonic imaging, high bandwidth combined with high frequency translates into high resolution [15]. Because of the thick film being integrated with the substrate as mentioned above, there are important constraints on the substrate. To begin with, it should be chemically and physically compatible with the thick film at the processing temperatures, and furthermore it should play the role of a backing material, being well-matched acoustically to the film and able to dissipate acoustic energy. For the thick films described in this work, a porous ceramic substrate has been used.  In order to benefit from the many advantages offered by electronic beamforming, the logical next step is to move from single-element transducers to multi-element array transducers. Since thick films are generally deposited by high-resolution additive processes such as screen printing and pad printing, they offer great possibilities of high-level integration and cost-effective manufacturing. The thick-film array prepared in this work is an example of such a transducer with 32 elements defined by parallel lines of the top electrode. It is important to note that this is a so-called kerfless design, i.e., the individual elements are not physically separated at the level of the thick film [23], which makes the patterning simpler and reduces the obtainable pitch (i.e., centre distance between elements).
Pulse-echo measurements of the individual elements of 6 such arrays yield a centre frequency of 11.4 MHz and a −6 dB fractional bandwidth of 55% in average (see example in Figure 4c,d, with reflector 5 mm away), which is somewhat reduced in comparison with that shown in Figure 4a,b for a single-element transducer. The printing method is not quite the same for the flat 32-element array transducer as for the curved single-element transducer-the former is deposited by screen printing and the latter by pad printing-but experience shows that the two methods yield similar performance. Instead the reason should be sought in the less favourable geometry of the long, narrow line elements, and the lower target frequency that dictates a higher film thickness (performance is seen to decrease slightly for films with centre frequencies approaching 10 MHz). It is worth mentioning, however, that the fractional bandwidth of 55% compares well with the more conventional 2-2 composite transducers made by dicing bulk ceramic and backfilling with polymer (see for example [24]). In order to benefit from the many advantages offered by electronic beamforming, the logical next step is to move from single-element transducers to multi-element array transducers. Since thick films are generally deposited by high-resolution additive processes such as screen printing and pad printing, they offer great possibilities of high-level integration and cost-effective manufacturing. The thick-film array prepared in this work is an example of such a transducer with 32 elements defined by parallel lines of the top electrode. It is important to note that this is a so-called kerfless design, i.e., the individual elements are not physically separated at the level of the thick film [23], which makes the patterning simpler and reduces the obtainable pitch (i.e., centre distance between elements).
Pulse-echo measurements of the individual elements of 6 such arrays yield a centre frequency of 11.4 MHz and a´6 dB fractional bandwidth of 55% in average (see example in Figure 4c,d, with reflector 5 mm away), which is somewhat reduced in comparison with that shown in Figure 4a,b for a single-element transducer. The printing method is not quite the same for the flat 32-element array transducer as for the curved single-element transducer-the former is deposited by screen printing and the latter by pad printing-but experience shows that the two methods yield similar performance. Instead the reason should be sought in the less favourable geometry of the long, narrow line elements, and the lower target frequency that dictates a higher film thickness (performance is seen to decrease slightly for films with centre frequencies approaching 10 MHz). It is worth mentioning, however, that the fractional bandwidth of 55% compares well with the more conventional 2-2 composite transducers made by dicing bulk ceramic and backfilling with polymer (see for example [24]).
For such a multi-element transducer, cross-talk is an important parameter to consider, especially in view of the kerfless design. This has been measured by comparing the voltage amplitude for the four elements closest to the exited element and an average value of´38 dB has been obtained. Figure 5 shows a comparison for two representative arrays. This is a surprisingly low cross-talk for a kerfless design, especially when compared with the value of´24 dB obtained with the 2-2 composite transducer mentioned before [24]. Clearly, the strongly reduced lateral coupling induced by porosity (as discussed in the case of k p of the porous bulk discs) is crucial for obtaining such a low cross-talk with this kerfless design.
The successful use of these thick-film multi-element transducers for ultrasonic imaging is described in a recent paper [16].

Materials 2015, 8, page-page
For such a multi-element transducer, cross-talk is an important parameter to consider, especially in view of the kerfless design. This has been measured by comparing the voltage amplitude for the four elements closest to the exited element and an average value of −38 dB has been obtained. Figure 5 shows a comparison for two representative arrays. This is a surprisingly low cross-talk for a kerfless design, especially when compared with the value of −24 dB obtained with the 2-2 composite transducer mentioned before [24]. Clearly, the strongly reduced lateral coupling induced by porosity (as discussed in the case of kp of the porous bulk discs) is crucial for obtaining such a low cross-talk with this kerfless design.
The successful use of these thick-film multi-element transducers for ultrasonic imaging is described in a recent paper [16].

Bulk Transducers
A number of discs (diameter 24.6 mm, thickness 2.5 mm) were prepared from a commercial soft PZT, Ferroperm™ Pz37 (Meggitt Sensing Systems, Kvistgaard, Denmark) [17], which in this article will be referred to as Pz37. This ceramic type has been designed to have a high level of porosity (by means of an organic pore former), approximately 25%, in order to obtain specific acoustic properties. The discs were poled using the same poling field as used for an equivalent low-porosity soft PZT, which was well below the breakdown field of the porous material (thus no dielectric breakdown was observed). Some of the discs were infiltrated with silicone oil by using a pressure cycle: vacuum-overpressure-vacuum. The number of samples for each category (without or with oil impregnation) was 5.
For the transducer fabrication, wires were soldered to the disc before bonding it to a cylinder of backing material with similar diameter and embedding in a plastic housing as indicated in Figure 6a. In order to reduce back wall echoes, the impedance of the backing was matched to that of the ceramic. Finally, the front face was coated with an epoxy layer with a thickness corresponding to one quarter of the wavelength (for impedance matching), and the back side and electrical leads were protected by potting (Figure 6b). Again, for each category the number of transducers was 5.
For the characterisation of the discs, the following equipment was used: Hewlett-Packard 4278A capacitance bridge (Keysight Technologies, Santa Rosa, CA, USA), Agilent E4990A impedance analyser (Keysight Technologies, Santa Rosa, CA, USA), PiezoMeter System PM200 & PM300 d33 meter (Piezotest Ltd., London, UK).  16 and 19) of two linear arrays. The average crosstalk is found to be approximately´38 dB.

Bulk Transducers
A number of discs (diameter 24.6 mm, thickness 2.5 mm) were prepared from a commercial soft PZT, Ferroperm™ Pz37 (Meggitt Sensing Systems, Kvistgaard, Denmark) [17], which in this article will be referred to as Pz37. This ceramic type has been designed to have a high level of porosity (by means of an organic pore former), approximately 25%, in order to obtain specific acoustic properties. The discs were poled using the same poling field as used for an equivalent low-porosity soft PZT, which was well below the breakdown field of the porous material (thus no dielectric breakdown was observed). Some of the discs were infiltrated with silicone oil by using a pressure cycle: vacuum-overpressure-vacuum. The number of samples for each category (without or with oil impregnation) was 5.
For the transducer fabrication, wires were soldered to the disc before bonding it to a cylinder of backing material with similar diameter and embedding in a plastic housing as indicated in Figure 6a. In order to reduce back wall echoes, the impedance of the backing was matched to that of the ceramic. Finally, the front face was coated with an epoxy layer with a thickness corresponding to one quarter of the wavelength (for impedance matching), and the back side and electrical leads were protected by potting (Figure 6b). Again, for each category the number of transducers was 5.
For the characterisation of the discs, the following equipment was used: Hewlett-Packard 4278A capacitance bridge (Keysight Technologies, Santa Rosa, CA, USA), Agilent E4990A impedance analyser (Keysight Technologies, Santa Rosa, CA, USA), PiezoMeter System PM200 & PM300 d 33 meter (Piezotest Ltd., London, UK). The set-up for pulse-echo measurements is shown in Figure 7a. The platform supporting the transducer had adjustable angle with respect to the target surface. The signal was generated with a Panametrics 5077 square-wave pulser/receiver (GE Measurement, Boston, MA, USA). For most of the testing described in this paper, the signal was set at 100 V and the period of the pulse was varied to maximise the return signal (see Figure 7b).

Thick Films
The single-element thick-film transducer (Figure 8a) was manufactured by pad-printing a gold bottom electrode, InSensor™ TF2100 PZT thick film (Meggitt Sensing Systems, Kvistgaard, Denmark) [22] (approx. 20 μm thick) and silver top electrode onto a porous ceramic substrate. The multi-element array transducers shown in Figure 8b,c were prepared in a similar manner, except that screen printing was used and the thickness was approximately 90 μm. 6 arrays were manufactured, all showing similar performance.
The pulse-echo measurements were performed using a JSR Ultrasonics DPR500 dual pulser/receiver (Imaginant Inc., Pittsford, NY, USA) with remote pulsers in combination with an Agilent Infiniium DSO8064A oscilloscope (Keysight Technologies, Santa Rosa, CA, USA). The set-up for pulse-echo measurements is shown in Figure 7a. The platform supporting the transducer had adjustable angle with respect to the target surface. The signal was generated with a Panametrics 5077 square-wave pulser/receiver (GE Measurement, Boston, MA, USA). For most of the testing described in this paper, the signal was set at 100 V and the period of the pulse was varied to maximise the return signal (see Figure 7b). The set-up for pulse-echo measurements is shown in Figure 7a. The platform supporting the transducer had adjustable angle with respect to the target surface. The signal was generated with a Panametrics 5077 square-wave pulser/receiver (GE Measurement, Boston, MA, USA). For most of the testing described in this paper, the signal was set at 100 V and the period of the pulse was varied to maximise the return signal (see Figure 7b).

Thick Films
The single-element thick-film transducer (Figure 8a) was manufactured by pad-printing a gold bottom electrode, InSensor™ TF2100 PZT thick film (Meggitt Sensing Systems, Kvistgaard, Denmark) [22] (approx. 20 μm thick) and silver top electrode onto a porous ceramic substrate. The multi-element array transducers shown in Figure 8b,c were prepared in a similar manner, except that screen printing was used and the thickness was approximately 90 μm. 6 arrays were manufactured, all showing similar performance.

Thick Films
The single-element thick-film transducer (Figure 8a) was manufactured by pad-printing a gold bottom electrode, InSensor™ TF2100 PZT thick film (Meggitt Sensing Systems, Kvistgaard, Denmark) [22] (approx. 20 µm thick) and silver top electrode onto a porous ceramic substrate. The multi-element array transducers shown in Figure 8b,c were prepared in a similar manner, except that screen printing was used and the thickness was approximately 90 µm. 6 arrays were manufactured, all showing similar performance.

Conclusions
This work has shown two very promising and commercially relevant examples of the use of porous piezoceramics. It is clear that introducing and controlling porosity creates new functional materials that are able to compete with existing materials and also enable new devices.
For the thick films, it should be mentioned that pad printing and screen printing are additive and cost-effective manufacturing methods with very useful characteristics in terms of integration with a substrate also acting as a backing material.

Conclusions
This work has shown two very promising and commercially relevant examples of the use of porous piezoceramics. It is clear that introducing and controlling porosity creates new functional materials that are able to compete with existing materials and also enable new devices.
For the thick films, it should be mentioned that pad printing and screen printing are additive and cost-effective manufacturing methods with very useful characteristics in terms of integration with a substrate also acting as a backing material.