Abstract
Acoustic emission (AE) sensors are capable of detecting elastic waves emitted when cracks advance in engineering structures. To provide real time continuous monitoring, such sensors would have to remain permanently in place, but this is uneconomic and impractical to realise in many situations. The availability of low cost and small AE sensors is hampered by the poor piezoelectric properties of materials that can easily be processed and the processing challenges associated with materials that exhibit high piezoelectric properties. Here, a ZnO surface treatment for lead zirconate titanate (PZT) ceramic particles is presented that enhances the interfacial bonding between PZT and polymer to allow mechanically robust composite films to be created that exhibit d33 piezoelectric coefficient of 15 pC/N after corona poling at 100°C. Films of modified PZT–polymer material have been used to detect simulated acoustic emission events. A comparison between such devices and commercially available PZT ceramic based devices is drawn.
Introduction
Piezoelectric materials can be used in a variety of sensors to gather information about the environment in which they sit. Examples of piezoelectric sensors include accelerometers, pressure sensors, microphones, ultrasound transducers and hydrophones.1 There is an increased drive to miniaturise such devices in order to increase robustness and reduce cost by moving away from high cost assembly based manufacturing routes. In the manufacture of such miniature devices, the challenges of integration play a significant role as materials with widely different properties and processing requirements are brought together in a processing environment.2, 3 Techniques such as physical vapour and sol–gel deposition have been used for the creation of thin films, while ink based processed such screen, spray and jet printing using ceramic inks and ceramic sol–gel inks2, 4 have been used to create thick film system. In each of these cases, high temperatures (<450°C) are required for the creation of integrated piezoelectric films. This requirement for high temperatures to crystallise or densify the ceramic phase means that low thermal stability materials, such as polymers and many metals, cannot be successfully integrated with these piezoelectric materials. In addition, it means that such coatings cannot easily be applied to large structures or in situ. Piezoelectric polymer materials, such as polyvinylidene difluoride, offer an alternative to ceramic materials yet typically exhibit low piezoelectric properties (d33, 10–30 pC/N) and lower Curie temperatures limiting their operation in higher temperature environments. An alternative is to consider ceramic–polymer composite systems, which can readily be applied at low temperatures using low cost film forming techniques potentially allowing economic integration of piezoelectric films and the creation of sensors on a variety of substrates. Such lead zirconate titanate (PZT)–polymer matrix composites have historically exhibited low piezoelectric properties. Bulk PZT materials exhibit d33 piezoelectric coefficients in the range of 300–450 pC/N, with thin and thick film materials of comparable composition exhibiting values between 80 and 120 pC/N.4 In comparison, PZT–polymer composites typically exhibit d33 values of between 10 and 30 pC/N5, 6 due to the low levels of interconnection between the piezoelectric ceramic particles, the lower proportion of ceramic phase as well as the difficulty in transferring the mechanical stress across the polymer/ceramic interface. In this work, we report on a novel treatment to enhance the bonding between the polymeric and active ceramic phases and thereby maximising the stress transferral while also maintaining a high volume fraction of ceramic phase without compromising the mechanical integrity of the film.
To demonstrate the effectiveness of this approach, composite acoustic emission sensors have been fabricated and their performance was compared to commercially available devices. Acoustic emission monitoring is an area of non-destructive structural health monitoring, which has generated increased interest recently.7 Acoustic emission monitoring can be used to predict component failure, allowing the component to be replaced before failure results in costly unplanned downtime or a risk to health and safety. Acoustic emission testing has found application in monitoring the health of both static and dynamic structural components.8–10
Acoustic emissions are caused by the growth of structural defects, such as cracks, within materials. The mechanical energy released as a consequence of this crack advancement travels through the structure as elastic waves (between 100 kHz and 1 MHz) and is detected by the AE sensor. In steel structures, the wavelength of these emissions is of the order of 10–30 mm, larger than the typical dimensions of the AE device. In this context, the piezoelectric element is deformed in its thickness direction, generating a piezoelectric response due to the d33 piezoelectric coefficient. In order to maximise the response of the device, the d33 coefficient of the material should be maximised.
The choice of AE sensors as a case study is of interest as there are a number of pertinent issues with contemporary AE monitoring technology including the fact that current AE sensors are too expensive to enable continuous monitoring11 and that the size of current AE sensors is also prohibitive to the monitoring of inaccessible locations or locations where a protruding body would cause negative effects, such as on aerodynamic or hydrodynamic structures. Lead zirconate titanate–polymer composite film devices can provide solutions to these issues. With the deposition of disposable devices directly onto a structural element, which requires monitoring, it would be possible to continuously monitor the structure as the devices can be left in situ on the component. The low temperature deposition also enables the monitoring of a broad array of structural elements as the temperatures employed during device manufacture will not cause damage to the structure.
Methodology
Surface modified PZT powder was mixed with low molecular weight (10 000–26 000) polyvinyl alcohol (PVOH; Alfa Aesar) and water to create an ink that could be applied to the surface of an aluminium foil test piece. The mass ratio of PZT/PVOH/H2O was 7∶1∶10 and, once dried, resulted in films with a 55 vol.-% powder loading. The surface of the PZT particles were modified with a ZnO nanocoating by adding 15 g of PZT powder (Ferroperm PZ26) to a solution consisting of 0·42 g of hexamethylenetetramine, 0·89 g of zinc nitrate and 124 mL of deionised water. The suspension was stirred and held at 90°C for 2·5 h to allow the growth of ZnO on the surface of the PZT particles. The final proportion of ZnO present with respect to PZT was 2 vol.-%, equating to a 3 nm thick coating. On cooling, the solution was washed with deionised water three times and dried at 70°C overnight.
Inks were deposited onto the Al foil by mounting the foil onto a rigid substrate angled at 45° and allowing the ink to flow down the foil leaving behind a film of uniform thickness. Once dry, circular Cr/Au electrodes were applied to the top of the film by evaporation using an Edwards E480 coater. The Al foil was used as the counter electrode. The device was corona poled under a field of 10 kV with a device to pin electrode separation of 10 mm. Poling was carried out at 100°C for 60 min, and then the device was cooled to room temperature while maintaining the poling field. Electrical characterisation was carried out using a Wayne Kerr precision component analyser 6425 to measure the capacitance and loss of the film. A Berlincourt type PM25 piezometer system was used to measure the d33 piezoelectric coefficient of the device.
The PZT–polymer composite AE device was benchmarked against a PICO AE sensor, which is a single element, resonant sensor, commercially available from Physical Acoustics Corporation (PAC).12 Impedance measurement was carried out between 50 kHz and 1 MHz using a Hewlett-Packard 4192A LF impedance analyser. It was shown that the PICO device exhibited a resonant response between 450 and 600 kHz. There was no such behaviour observed with the PZT–polymer composite AE device, which is expected due to the smaller size and compliant nature of the structure.
For testing, the sensors were mounted on a steel plate 250mm×250mm and 5 mm thick. The sensors were mounted 50 mm from opposite edges of the plate with the source of simulated acoustic emission equidistant from both sensors and in the middle of the plate. Grease was used as a coupling agent, and the sensors were held onto the surface of the plate using a 113 g weight. The sensors were connected to 2/4/6 variable preamplifiers from PAC. The signals from the PZT–polymer device and the PICO sensor were preamplified by 40 and 60 dB respectively. AEWin data logging software from PAC was employed to monitor the signals emitted from the AE devices. A threshold value of 30 dB was applied to the PICO sensor, and a value of 40 dB was applied to the PZT–polymer device. A sampling rate of 5 MHz was employed. In addition, an upper filter of 3 MHz and a lower filter of 1 kHz were employed. Hsu–Nelson testing13 was carried out where a 2H pencil lead was broken at the centre of the steel plate equidistance from the two sensors. In total, 60 Hsu–Nelson tests were carried out.
Results and discussion
Figure 1 shows a scanning electron micrograph of the modified powder and (ZnO)PZT–PVOH films. It can be seen that the films are able to deform plastically without breaking, indicating that, despite the low levels of polymer present (45 vol.-%), the PZT/polymer interface is strong. PVOH has previously been shown to bond strongly with ZnO,14, 15 which is thought to occur via hydrogen bonding from the –OH polymer groups to the zinc vacancy sites16 and OH2+, OH and O− groups of the ZnO.17 While PVOH could also bond to PZT across the OH type surface groups, the bonding appears more effective in the case of ZnO modified PZT possibly due to the greater number of bond sites.
Images (SEM) of ZnO modified PZT–PVOH thick films:
After poling, the PZT–polymer device was found to exhibit a d33 of 15·2 pC/N, a relative permittivity of 86·3 and loss of 0·058. This compares to a d33 of 330 pC/N exhibited by high density PZ26 (PZT4 type material) and between 50 and 80 pC/N for ceramic thick film PZ264 on a variety of substrates. Similar values of relative permittivity have been reported for PZT epoxy films containing 60–70 vol.-% ceramic phase.5, 6 A d33 of 17 pC/N has been reported for 60 vol.-% PZT epoxy films5 made using a PZT5H type material, which exhibits a significantly higher d33 compared to the material used in this study (∼580 pC/N versus ∼330 pC/N). A PZT–polymer film made with untreated PZ26 exhibited a d33 of 8 pC/N, which represents a similar film/bulk property ratio to that observed in the PZT5H system reported in the literature. This indicated that further increases in piezoelectric properties (potentially up to 27 pC/N) could be obtained using a PZT5H type of material in place of PZ26. The X-ray diffraction analysis results for the film are shown in Fig. 2. The peaks associated with the PZT are clearly visible and correspond to a randomly oriented powder. Also evident is the broad peak associated with the (101) reflection of PVOH and the characteristic pattern of wurtzite ZnO. The a and c lattice parameters are slightly smaller than expected from the Joint Committee on Powder Diffraction Standards (JCPDS) database but still within the values reported for nanosized ZnO. No further secondary phases were identified, indicating that no detectable reaction had occurred between the PZT and ZnO.
X-ray diffraction pattern for ZnO modified PZT–PVOH composite film; diamond symbols indicate peaks associated with ZnO
Representative time domain plots of the signals detected by the (ZnO)PZT–PVOH device and the PICO commercial device can be seen in Fig. 3a and b respectively. The maximum amplitudes of the time domain signal detected by the composite device and the PICO commercial device were 140 and 345 mV respectively, following preamplification. The signal/ noise ratio (maximum amplitude of the AE divided by the amplitude of the noise signal before AE detection) of the PZT/polymer device was found to be 2·5, slightly below the industry recommended standard of 4.18 The AE signal was detected 20 μs earlier by the PZT–polymer composite device than the PICO commercial device.
Representative time domain plots showing emitted signal from a (ZnO)PZT–polymer sensor and b PICO commercial sensor in response to AE generated by Hsu–Nelson testing
Before preamplification, the signal detected by the composite device was 20 times smaller in amplitude than that detected by the PICO device. This correlates with the approximate difference in d33 of the PZT–polymer device (15 pC/N) and typical bulk PZT (∼300pC/N) from which the PICO sensor active element is manufactured.19 AGU Vallen Wavelet Software (release A2009·1027) was employed to transform the time domain signal using Gabor wavelets using a window of 10 μs and a 100 Hz frequency resolution. The wavelet transforms of the time domain signal detected by the PZT–polymer and the PICO sensors are presented in Fig. 4a and b respectively.
Gabor transform spectrograms of emitted signal from a PZT/polymer sensor and b PICO commercial sensor in response to AE generated by Hsu–Nelson testing
The maximum amplitude of the PICO sensor was found to result from the interaction of the zeroth order symmetric S0 wave with the resonant sensitivity peak of the PICO sensor, which was observed in an impedance measurement of the device, at a frequency between 450 and 500 kHz. The intersection of the S0, the zeroth order asymmetric wave (A0) and the first order asymmetric wave (A1) exhibited lower amplitude in the signal generated by the PICO device due to the intersection taking place at a frequency of 340 kHz. The response of the PICO device was shown by an impedance sweep to be at a minimum at a frequency of 340 kHz. The interaction of the AE wave with the edge of the plate resulted in reflections, which were detected by the PICO device.
The Gabor transform of the signal generated by the PZT–polymer sensor shows that there were periodic signals between 20 and 400 kHz and between 600 kHz and 1·2 MHz, which occurred at intervals of 13 and 5 μs respectively. These signals were detected before the AE arrived at the PZT–polymer sensor and were not apparent in the signal emitted by the PICO device, indicating that the signals were a result of noise between the PZT–polymer sensor and the AEWin data logger and not a characteristic of the signal or the steel test plate.
The S0 and A1 Lamb wave components interacted with the noise in the signal emitted from the PZT–polymer sensor at a frequency of 350 kHz. The A2 waveform interacted with the higher frequency noise at 900 kHz, resulting in a larger amplitude signal. The A0 waveform was at a frequency below the sensitivity bandwidth of both the PZT/polymer and the PICO sensor and, as such, did not result in an amplitude response. Similarly, the third order waveforms and above were present at a frequency too high to result in an amplitude response. The smaller amplitude signal characteristics were masked by the larger amplitude noise component.
The low stiffness of the polymer matrix meant that acoustic energy was attenuated rapidly in the film, further reducing reflections off the back face of the sensor. In the case of a commercial sensor, a dedicated acoustic attenuator is mounted behind the active piezoelectric element to eliminate these reflections.
Conclusion
The surface modification of the PZT powder through the addition of ZnO was shown to increase the adhesion between the PZT and the PVOH matrix phase. This allowed a flexible piezoelectric film to be created that could readily be applied to a variety of surfaces and used as a means of detecting AE events. The piezoelectric response of the films was increased from 8 to 15 pC/N, indicating that the ZnO coating enhanced the bonding between the ceramic and polymer phase thereby increasing the transfer of stress across the interface.
It was found that the (ZnO)PZT–polymer device was capable of detecting AE generated by Hsu–Nelson testing at frequencies between 200 kHz and 1·2 MHz. This indicates that the PZT–polymer device will act as a wideband sensor more suited to the characterisation of the AE and the structural element but less suited for use in locating of AE sources than a resonant type device such as the PICO sensor. However, the signal/noise ratio and the deformation of the signal was such that the PZT–polymer sensor was found to be incapable of identifying characteristics of the AE source and the steel test plate through which the AE was transmitted.