- Email: [email protected]
1–3 Connectivity lithium niobate composites for high temperature operation
Available online at www.sciencedirect.com
Ultrasonics 47 (2007) 15–22 www.elsevier.com/locate/ultras
1–3 Connectivity lithium niobate composites for high temperature operation N. Schmarje, K.J. Kirk *, S. Cochran Microscale Sensors, School of Computing, University of Paisley, High Street, Paisley PA1 2BE, UK Received 19 November 2004; received in revised form 10 November 2005; accepted 7 June 2007 Available online 14 June 2007
Abstract Lithium niobate, LiNbO3, is a piezoelectric material well known for its high Curie temperature. However, it has often been neglected for use in ultrasonic transducers because of its low electro-mechanical coupling coefficients. Recent advances in signal processing have made this disadvantage less significant and we now report an investigation of the potential of LiNbO3 composites for use in high temperature transducers for non-destructive testing (NDT). LiNbO3 composites of 1–3 connectivity in room temperature vulcanising (RTV) sealant and cement matrices were fabricated by the dice and fill method. The RTV and the cement are specified to withstand temperatures up to 350 C and 1600 C, respectively. The composites have been characterized by electrical impedance measurement at ambient and elevated temperatures. In array configuration, transmit–receive signals from the back wall of a steel specimen were obtained at room temperature with good signal to noise ratio. High temperature measurements were made at temperatures up to 180 C for the RTV composite and 360 C for the cement composite configured as single element transducers. Temperature cycling has also been investigated and the new composite materials have been demonstrated to withstand several cycles without deterioration. It is concluded that they may contribute toward a solution to presently unsolved problems in NDT at elevated temperatures. 2007 Elsevier B.V. All rights reserved. Keywords: Lithium niobate; High temperature; Non-destructive testing; Ultrasonic transducers
1. Introduction Because of the expense of plant shutdown, non-destructive testing (NDT) techniques which can be applied at elevated temperatures are of significant interest to industry. One of the most popular NDT techniques at ambient temperature is ultrasonic inspection. However, piezoelectric ultrasonic transducers are regarded as suitable for operation at temperatures only up to half of the Curie temperature of the piezoelectric material [1]. Typical Curie temperatures for conventional piezoceramics based on lead zirconate titanate (PZT) range from 200 to 400 C. Ultrasonic transducers for NDT applications up to 350 C are widely available from a range of suppliers. However, a temperature range up to 500 C is often desirable but much *
Corresponding author. Tel.: +44 141 848 3409; fax: +44 141 848 3663. E-mail address: [email protected] (K.J. Kirk).
0041-624X/$ - see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.ultras.2007.06.001
harder to achieve [2], and it is towards this aim that this paper is focused. Various adaptations of conventional NDT configurations have been developed for high temperature applications. Buffer rods have been introduced to allow a temperature gradient between the surface of the test object and the piezoelectric material [3]. Fully non-contacting ultrasonic methods, such as electromagnetic acoustic transducers (EMATs) [4], laser generated ultrasound [5] and air coupled ultrasound [6], have also been developed. However, it is a common disadvantage of the non-contact methods that the fabrication of arrays is difficult and scarcely reported to this date. A lot of research has been focused on piezoceramic ultrasonic transducers for direct contact, stretching the boundaries of use for piezoceramic based devices. Vermon (Tours, France) has developed an array probe based on a PZT/epoxy composite that is chemical and pressure
16
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22
resistant but operates up to only 100 C [7]. Piezoceramic composite transducers for operation in harsh environments have been developed and immersion tests have been carried out in Canola oil at 125 C [8]. An ultrasonic array has been developed for under sodium viewing using separate piezoceramic elements and data acquisition system and tests visualising targets in liquid sodium at 200 C have been reported [9]. More relevant to the higher temperature target considered here is the work by Kobayashi et al. [10,11] at McGill University, Montreal, Canada. They are working on thick film bismuth titanate/bismuth titanate composites with a possible operating temperature of 440 C. The high Curie temperature of LiNbO3, TC 1210 C, makes it a suitable material for this application, with an upper operating temperature limit known to be 650 C [12], beyond which the material suffers oxygen loss and shows low electrical resistivity. Therefore, LiNbO3 composites filled with cement have the potential to operate up to 650 C. The z-cut of crystalline LiNbO3 has an electro-mechanical coupling coefficient, kt, reported to be only 0.17. However, it is preferred over the y/36-cut in arrays because of its isotropic properties [13] whilst the y/36cut finds use in high frequency applications [14]. Previously, arrays made with monolithic plates of LiNbO3 have been fabricated and tested [15,16]. B-scans of defects were obtained; however it was observed that crystalline LiNbO3 tends to crack during cooling. This phenomenon is frequently observed in bulk crystalline LiNbO3. Therefore 1–3 connectivity LiNbO3 cement composites were proposed. Monolithic LiNbO3 has also been used in brazed-type transducer and tested at 550 C for 400 days [17]. The fabrication methods of 1–3 connectivity composites are well established [18] and benefits include an improvement in kt, broader bandwidth, reduced coupling of laterally propagating waves and the possibility to tailor the acoustic impedance. In a first attempt at a practical realisation, LiNbO3 cement composites were manufactured and tested experimentally [19]. A-scans were obtained at temperatures up to 336 C showing the reflected compression wave from the back wall of a test block. In this case, however, during cooling the sample detached from the test block to which it had been coupled with cement and the individual pillars showed evidence of the shattering previously seen in monolithic bulk material. In further, more recent tests [20] LiNbO3/epoxy composites were manufactured, characterized experimentally and modelled using the PZ Flex finite element analysis code (Weidlinger Associates, Los Altos, CA, USA). In both the experiments and the modelled results we observed an improvement in the thickness mode electro-mechanical coupling coefficient, kt, of 75% over the value of bulk material. This previous work suggests that there are two important requirements for the active material for a transducer to be used in condition monitoring at high temperatures. Firstly, the sensitivity of the material has to be sufficient for it to be
configured as an array. This is crucial in order to cover a large inspection region using beam steering with a fixed transducer. Secondly, the material has not only to withstand the high temperatures but also temperature cycling. In the remainder of this paper, we first introduce the RTV sealant and the cement used as passive fillers, then discuss composite parameters and present a B-scan of the back wall of a steel specimen at room temperature to show the sensitivity of the composite material. Furthermore, we demonstrate piezoelectric activity at elevated temperatures via electrical impedance measurements, which have also been used to validate the behaviour after temperature cycling. Finally, we present results from tests with the composite materials in transmit–receive mode at temperatures up to 360 C. 2. High temperature composites Composites of 1–3 connectivity were fabricated by the well known dice and fill method [21]. A saw pitch of
Fig. 1. Composite diced at 45 to ensure that electrodes cover equal areas of active material and that ‘‘dead’’ elements are avoided. LiNbO3 pillars are shown in light grey and outline of PCB tracks in dark grey. (a) Uniform coverage of active material. (b) How ‘‘dead’’ elements covering no active material can occur when the material is diced parallel to the edges. The illustrations are drawn to scale.
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22
Fig. 2. Cement (left) and RTV (right) lithium niobate composites. The pillar width is 0.7 mm and the volume fraction 45%. The composites are 10 mm · 10 mm in size.
approximately 1.1 mm and a pillar width of 0.7 mm resulted in a LiNbO3 volume fraction of 45%. The thickness of the material was approximately 1.5 mm. For single element transducer tests the material was cut to a size of 10 · 10 mm and for tests in an array configuration 10 · 35 mm material was used. In both cases the dicing direction was aligned at an angle of 45 to the crystal edges to ensure that an approximately equal area of active material would be covered by each electrode when a printed circuit board (PCB) with linear electrodes was applied for tests in array configuration (see Fig. 1). For the work reported here, z-cut LiNbO3 was used, despite its relatively low thickness mode electro-mechanical coupling factor as its properties are closer to isotropic than for other common cuts. This is important in consideration of future use in beam steered arrays; previous work [15] showed that the more sensitive y/36-cut material introduced minima in the beam profile of both single array elements and steered beam configurations. A high temperature composite also requires that the passive filler material can withstand the operating conditions. Two passive materials were used in the present work, RTV sealant suitable for use at temperatures up to 350 C (Cu371, Intel Adhesive Ltd., Blyth, Northumberland, UK) and high temperature cement for use up to 1600 C (C920, Cotronics, Brooklyn, NY, USA). The RTV itself is a flexible material; therefore the RTV composites are flexible, whilst the cement composites are rigid. Photographs of the composites are shown in Fig. 2. In this paper each material is termed according to its passive phase, i.e. ‘‘RTV composite’’ and ‘‘cement composite’’. 3. Experimental setup For sensitivity tests in transmit–receive mode in an array configuration at room temperature, an epoxy-glass PCB was clamped onto a cement composite to achieve electrical excitation and sensing. The PCB had rectangular electrodes 0.25 mm wide, with a spacing of 0.635 mm, which were long enough to overhang the composite at each side. Each
17
electrode was connected to an external connector pin. Due to the materials of the PCB, this setup can operate only at room temperature. However, thick film gold electrodes on alumina substrates have been used previously [15] and it is known that they can withstand the temperatures cited here. Both RTV and cement composites were characterized by electrical impedance measurements and tested in transmit– receive mode. For electrical impedance characterization the composite was electroded on both major surfaces with conductive silver paint and clamped between two steel blocks with thin steel plates used to connect to cables, electrically and acoustically insulated from the clamping blocks by cement plates as shown in Fig. 3a. The assembly was set on a hot plate for heating. At intervals the electrical impedance was measured as a function of frequency, with an HP 4192 A impedance analyser (Agilent Technologies, South Queensferry, UK). For transmit–receive testing, the composite was coupled on to a 37.5 mm thick steel block with UCA-HT couplant (Ely Chemicals, Ely, Cambridgeshire, UK). The top of the composite was again electroded with silver paint and a small steel plate and electrical insulator were used to clamp it onto the specimen as shown in Fig. 3b. Again, this assembly was set on a hot plate. The transducer was excited by applying a broadband electrical signal in the form of a negative going step with an amplitude of approximately 200 V and a fall time of approximately 20 ns, generated by a DPR300 pulser-receiver (JSR Ultrasound, Pittsford, NY, USA). The steel specimen itself served as the ground electrode and provided some acoustic damping. Return signals were amplified with the DPR300 and recorded with a digital oscilloscope.
4. Results To test the cement composite in the array configuration at room temperature, transmit–receive data were acquired from each of the 36 elements formed by the individual PCB tracks. A typical signal for this array configuration at room temperature, from element 13, is shown in Fig. 4a. Each linear electrode of the PCB excites parts of 6–7 pillars with a total electrode area 10 mm · 0.25 mm, thus covering only a relatively small area of active material as illustrated previously in Fig. 1. The relative peak-topeak (p–p) amplitudes of the back wall signals for all elements are shown in Fig. 4b, normalized to a maximum of 0 dB. Apart from one element with poor sensitivity, the amplitudes of the signal remain the same to within much less than 10 dB for all but one of the elements, which suggests good electrical coupling between the PCB and the pillars throughout the composite. Additionally, all signals are easily distinguished from the noise level, as shown in Fig. 4c, where the SNR was calculated from the peak to peak (p–p) amplitude of the back wall signal divided by the maximum p–p amplitude of the voltage recorded between the excitation ring down and the occurrence of
18
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22
Fig. 3. Illustration of experimental setup for (a) electrical impedance measurements at high temperature and (b) transmit–receive tests at high temperatures.
the back wall signal, hence producing conservative values for SNR. Electrical impedance as a function of frequency was measured at room temperature and during heating for an RTV composite. The temperature was stabilized at 250 C and the electrical impedance was measured at hourly intervals for 5 h. Fig. 5 shows impedance graphs at room temperature and one of five very similar graphs at 250 C, which were taken at hourly intervals. The absolute impedance magnitude decreased slightly with increasing temperature, as did the perturbation due to the piezoelectric resonance. The decease in impedance magnitude with temperature can be attributed to the positive
temperature coefficient of permittivity of LiNbO3. However, the measurements taken at 250 C during 5 h showed almost identical graphs in terms of perturbation and absolute impedance magnitude. These results were reproduced up to four times on the same piece of composite. The material regained the same absolute impedance magnitude after each cooling and still showed piezoelectric activity, indicating the absence of any curing effects on the composite during the first cycles and its viability for temperature cycling. The same impedance test has been carried out with a piece of bulk RTV material to investigate any variation in the relative permittivity, er, as a function of temperature. The very soft RTV material was clamped between two steel
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22 10000
0.005
0 12
13
14
15
16
-0.005
RT
Impedance magnitude [Ω]
Signal Amplitude [V]
0.01
19
-0.01
1000 1.5
Time [µs]
2
HT250
2.5
3
Frequency [MHz] Array Element 1
6
11
16
21
26
31
36
-75
RT
Impedance phase [˚]
Relative Amplitude [dB]
0
-5
-10
HT250
-80
-85
-90
-15
20 -95 1.5
SNR (dB)
2
2.5
3
Frequency [MHz]
15
Fig. 5. Electrical impedance as a function of frequency of RTV composite at room temperature (RT) and 250 C (HT250). (a) Impedance magnitude and (b) impedance phase.
10
5
0 1
6
11
16
21
26
31
36
Array Element
Fig. 4. For the cement composite in array configuration in (a) a typical pulse-echo signal from the back wall of the test block at room temperature, recorded with array element 13, in (b) relative signal amplitudes for all array elements and in (c) the signal to noise ratio for each element is shown.
blocks making it difficult to determine the precise thickness and therefore a precise value for er. However, the impedance of the RTV did not vary with temperature. This indicates a constant relative permittivity over the temperature range discussed in this paper. We therefore attribute the reduction in the impedance response not to changes in permittivity but to increased acoustic damping in the RTV as the temperature is increased. In transmit–receive tests with the RTV composite configured as a single element transducer, the signal from the back wall arrives after a 12.7 ls delay at room temperature and a slightly longer delay at 180 C, as expected [22]. Fig. 6 shows the data taken at room temperature and 180 C. The soft RTV is rather difficult to machine. With the present method of manual finishing with abrasive
paper, the RTV emerges slightly thicker than the LiNbO3 leaving an undulating surface profile, with the RTV as the peaks and the LiNbO3 as the troughs as illustrated in Fig. 7a. This causes problems in trying to achieve good coupling between the transducer and the steel test block, leading to relatively long signals and narrow bandwidth. The latter can be observed in Fig. 6b and d, where the presence of the 3rd and 5th harmonics of the fundamental frequency is also clear. These harmonics are the reason for the appearance of higher frequency operation near the beginning of each pulse in the time domain in Fig. 6a and b. Higher frequency operation does not occur near the end of the pulses because higher frequencies are more strongly attenuated. Furthermore, we observed a decrease in the amplitude of the back wall signal as the temperature increased and above 180 C it could no longer be distinguished from the noise. For the cement composite, the electrical impedance was measured at room temperature and up to 350 C. Again, the temperature was stabilized and kept at 350 C for 5 h and the impedance was measured at hourly intervals. This was repeated up to three times on the same piece of composite, which still showed piezoelectric activity after cooling. The data are presented in Fig. 8. It can be seen that the absolute impedance magnitude decreases with increas-
20
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22 0.8
Signal [V]
0.4
0.0 12
14
16
18
-0.4
Fig. 7. Schematic of the surface profile of (a) RTV composite and (b) cement composite. Light grey is LiNbO3 and dark grey is the passive material.
-0.8
Time [µs] 50 45
100000
35 30 25 20 15 10 5 0 0
5
10
15
Impedance magnitude [Ω]
Signal [arb. units]
40
RT HT350
10000
Frequency [MHz]
1000 1.5
2
0.8
0.4
3
-65
0.0 12
13
14
15
16
17
18
-0.4
Impedance phase [˚]
Signal [V]
2.5
Frequency [MHz]
RT
-75
HT350
-85
-0.8
Time [µs]
-95 1.5
Signal [arb. units]
20
2
2.5
3
Frequency [MHz]
Fig. 8. Electrical impedance as a function of frequency of cement composite at room temperature (RT) and 350 C (HT350). (a) Impedance magnitude and (b) impedance phase.
15
10
5
0 0
5
10
15
Frequency [MHz]
Fig. 6. Results from single element RTV composite. (a) The back wall signal at room temperature, (b) the frequency spectrum of the signal in (a), (c) the back wall signal at 180 C and (d) the frequency spectrum of the signal in (c).
ing temperature. Furthermore, we observe a shift of this resonance to lower frequencies with increasing temperature
and an increase in the perturbation of the impedance graph around the thickness mode resonant frequency during the first cycle, indicating curing effects on the composite. er as a function of temperature for the cement was determined in the same way as described for the RTV sealant. In this case, it was found to be stable at er = 7.2 ± 0.5 over the temperature range discussed here, and can therefore again be neglected as a reason for changes in electrical impedance with temperature. Tests in transmit–receive mode were conducted with the composite material as a single element transducer on the same specimen as before. Fig. 9 shows data taken at room temperature and 360 C, and corresponding frequency
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22 0.050
Signal [V]
0.025
0.000 12
13
14
15
16
17
18
-0.025
-0.050
Time [µs] 2.5
Signal [arb. units]
2.0
1.5
1.0
0.5
0.0
0
5
10
21
by manually removing the excess passive material and the LiNbO3 stock. Usually we would use precision lapping equipment for this operation; in this case however, the cement is water soluble and would be damaged. In addition the cement is much softer than the LiNbO3 and therefore abrades much faster. This leaves again an undulating surface profile, but this time with LiNbO3 as the thicker material (see Fig. 7b). Thus the cement composite is rather fragile; however, it does not experience the same problem in coupling as the RTV composite. Signals from the back wall were detected after a 12.7 ls delay at room temperature and after a slightly longer time difference at 360 C than at 180 C, because the higher temperature leads to a larger acoustic velocity reduction in the steel than at 180 C [22,23]. With the cement composite the signal amplitude did not decrease as significantly with temperature as for the RTV composite and the signal was still clearly visible at 360 C, which was the maximum temperature that could be achieved with the current experimental setup. However, the effects of slower ring down after the excitation signal obscure the signal to a greater extent at this temperature. The slower ring down is attributed to degradation of the couplant.
15
Frequency [MHz]
5. Conclusions
0.050
Signal [V]
0.025
0.000 12
13
14
15
16
17
18
-0.025
-0.050
Time [µs] 2.5
Signal [arb. units]
2.0
1.5
1.0
0.5
0.0 0
5
10
15
Frequency [MHz]
Fig. 9. Results from single element cement composite; (a) the back wall signal from a single element cement composite at room temperature, (b) the frequency spectrum of the signal in (a), (c) the back wall signal from the RTV composite at 360 C and (d) the frequency spectrum of the signal in (c).
spectra indicating broader bandwidth than for the RTV composite. Our cement composites were also fabricated
This paper has described the fabrication, characterization and testing of 1–3 connectivity LiNbO3 composites at room temperature as well as at elevated temperatures. The piezocomposite material showed good potential for implementation as the active material in high temperature ultrasound trandsucers. To test the sensitivity and suitability for imaging of LiNbO3 cement composites the material was successfully configured as an array at room temperature with elements of 10 mm · 0.25 mm. Echoes from the back wall of a steel specimen were obtained with each element of the array with satisfactory SNR. Both the RTV and the cement composites showed piezoelectric activity after several temperature cycles up to 250 and 350 C, respectively. Difficulties were experienced in the use of the RTV composites for transmit–receive tests because of poor acoustic coupling and consequent poor SNR. There is potential to improve this by using more advanced fabrication methods such as low temperature machining. However, signals from the back wall of a steel block were detected with the RTV composite prototypes at temperatures up to 180 C. In transmit–receive tests with the cement composite, signals from the back wall of a steel block were detected at temperatures up to 360 C, the maximum set by the experimental setup. Both the RTV and cement composites were thermally cycled without problems. As the materials have been demonstrated to be both functional and robust, we conclude that LiNbO3 composites are promising for the active elements in high temperature ultrasonic transducers and arrays.
22
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22
References [1] R.C. Turner, P.A. Fuierer, R.E. Newnham, T.R. Shrout, Material for high temperature acoustic and vibration sensors: a review, Appl. Acoust. 41 (1994) 299–324. [2] Private communications, Doosan Babcock, Renfrew, UK. [3] L.C. Lynnworth, G. Jossinet, E. Cherifi, 300 Clamp-on ultrasonic transducer for measuring water flow and level, in: Procceedings of the IEEE Ultrasonics Symposium, 1996, pp. 407–412. [4] H.M. Frost, Electromagnetic ultrasonic transducer: principles, practice and applications, in: W.P Mason, R.N. Thurston (Eds.), Physical Acoustics, vol. XIV, 1985, pp. 179–275. [5] C.B. Scruby, R.J. Dewhurst, D.A. Hutchins, S.B. Palmer. Laser Generation of Ultrasound in Metals, in: R.S. Sharpe (Ed.), Research Techniques in Nondestructive Testing, vol. 5, 1982, pp. 281– 327. [6] R. Farlow, G. Hayward, Real-time ultrasonic techniques suitable for implementing non-contact NDT systems employing piezoceramic composite transducers, Br. J. Nondestr. Test. 36 (1994) 926–935. [7] C. Devallencourt, S. Michau, C. Bantignies N. Felix, A 5 MHz piezocomposite ultrasound array for operation in high temperature and harsh environment, in: IEEE Ultrasonics Symposium, 2004, pp. 1294–1297. [8] Q. Xue, M. Stanton, G. Elfbaum, A high temperature and broadband immersion 1–3 piezocomposite transducer for accurate inspection in harsh environments, in: IEEE Ultrasonics Symposium, 2003, pp. 1372–1375. [9] H. Karasawa, M. Izumi, T. Suzuki, S. Nagai, M. Tamura, S. Fijumori, Development of under-sodium three-dimensional visual inspection technique using matrix-arrayed ultrasonic transducer, J. Nucl. Sci. Technol. (2000) 769–779. [10] M. Kobayashi, C.-K. Jen Y. Ono, Lead free thick piezoelectric film as miniature high temperature ultrasonic transducer, in: IEEE Ultrasonics Symposium, 2004, pp. 910–913. [11] M. Kobayashi, C-K Jen, Piezoelectric thick bismuth titanate/lead zirconate titanate composite film transducers for smart NDE of metals, Smart Mater. Struct. 13 (2004) 951–956.
[12] BRITE/EURAM project BREU/CT92-0254, Project report: New piezoelectric ceramic with Tc > 1000C for operation up to 800C, Project coordinator: W.Wolny, Ferroperm A/S, Kvistgard, Denmark, 1992–1995. [13] I.B. Stumpf, Investigation of high density phased arrays for ultrasonic contact testing in steel, PhD Thesis, Paisley College of Technology (1982). [14] J.M. Cannata, T.A. Ritter, W.-H. Chen, R.H. Silverman, K. Kirk Shung, Design of efficient, broadband single-element (20–80 MHz) ultrasonic transducers for medical imaging applications, IEEE Trans. Ultrason. Ferroelect. Freq. Contr. 50 (2003) 1548–1557. [15] K. Kirk, A. McNab, A. Cochran, I. Hall, G. Hayward, Ultrasonic array for monitoring cracks in an industrial plant at high temperatures, IEEE Trans. Ultrason. Ferroelect. Freq. Contr. 46 (1999) 311– 319. [16] A. McNab, K.J. Kirk, A. Cochran, Ultrasonic transducers for high temperature applications, IEE Proc. Sci. Meas. Technol. 145 (5) (1998) 229–236. [17] T. Arakawa, K. Yoshikawa, S. Chiba, K. Muto, Y. Atsuta, Appliactions of brazed-type ultrasonic probes for high and low temperature uses, Nondestr. Test Eval. 7 (1992) 263–272. [18] W.A. Smith, The role of piezocomposites in ultrasonic transducers, in: Proceedings of the IEEE Ultrasonics Symposium, 1989, pp. 755– 766. [19] G. Shepherd, A. Cochran, K.J. Kirk, A. McNab, 1–3 connectivity composite material made from lithium niobate and cement for ultrasonic condition monitoring at elevated temperatures, Ultrasonics 40 (1–8) (2002) 223–226. [20] N. Schmarje, J.-F. Saillant, K. Kirk, S. Cochran, Imaging with lithium niobate/epoxy composites, Ultrasonics 42 (2004) 439–442. [21] H.P. Savakus, K.A. Klicker, R.E. Newnham, PZT-Epoxy piezoelectric transducer: a simplified fabrication procedure, Mater. Res. Bull. 16 (1981) 677–680. [22] J. Krautkra¨mer, H. Krautkra¨mer, Ultrasonic Testing of Materials, Springer-Verlag, Berlin, 1990, ISBN 0-387-51231-4, p. 497. [23] J.D.N. Cheeke, Fundamentals and Applications of Ultrasonic Waves, CRC Press, Boca Raton, FL, 2002, ISBN 0-8493-0130-0, pp. 86–87.
Ultrasonics 47 (2007) 15–22 www.elsevier.com/locate/ultras
1–3 Connectivity lithium niobate composites for high temperature operation N. Schmarje, K.J. Kirk *, S. Cochran Microscale Sensors, School of Computing, University of Paisley, High Street, Paisley PA1 2BE, UK Received 19 November 2004; received in revised form 10 November 2005; accepted 7 June 2007 Available online 14 June 2007
Abstract Lithium niobate, LiNbO3, is a piezoelectric material well known for its high Curie temperature. However, it has often been neglected for use in ultrasonic transducers because of its low electro-mechanical coupling coefficients. Recent advances in signal processing have made this disadvantage less significant and we now report an investigation of the potential of LiNbO3 composites for use in high temperature transducers for non-destructive testing (NDT). LiNbO3 composites of 1–3 connectivity in room temperature vulcanising (RTV) sealant and cement matrices were fabricated by the dice and fill method. The RTV and the cement are specified to withstand temperatures up to 350 C and 1600 C, respectively. The composites have been characterized by electrical impedance measurement at ambient and elevated temperatures. In array configuration, transmit–receive signals from the back wall of a steel specimen were obtained at room temperature with good signal to noise ratio. High temperature measurements were made at temperatures up to 180 C for the RTV composite and 360 C for the cement composite configured as single element transducers. Temperature cycling has also been investigated and the new composite materials have been demonstrated to withstand several cycles without deterioration. It is concluded that they may contribute toward a solution to presently unsolved problems in NDT at elevated temperatures. 2007 Elsevier B.V. All rights reserved. Keywords: Lithium niobate; High temperature; Non-destructive testing; Ultrasonic transducers
1. Introduction Because of the expense of plant shutdown, non-destructive testing (NDT) techniques which can be applied at elevated temperatures are of significant interest to industry. One of the most popular NDT techniques at ambient temperature is ultrasonic inspection. However, piezoelectric ultrasonic transducers are regarded as suitable for operation at temperatures only up to half of the Curie temperature of the piezoelectric material [1]. Typical Curie temperatures for conventional piezoceramics based on lead zirconate titanate (PZT) range from 200 to 400 C. Ultrasonic transducers for NDT applications up to 350 C are widely available from a range of suppliers. However, a temperature range up to 500 C is often desirable but much *
Corresponding author. Tel.: +44 141 848 3409; fax: +44 141 848 3663. E-mail address: [email protected] (K.J. Kirk).
0041-624X/$ - see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.ultras.2007.06.001
harder to achieve [2], and it is towards this aim that this paper is focused. Various adaptations of conventional NDT configurations have been developed for high temperature applications. Buffer rods have been introduced to allow a temperature gradient between the surface of the test object and the piezoelectric material [3]. Fully non-contacting ultrasonic methods, such as electromagnetic acoustic transducers (EMATs) [4], laser generated ultrasound [5] and air coupled ultrasound [6], have also been developed. However, it is a common disadvantage of the non-contact methods that the fabrication of arrays is difficult and scarcely reported to this date. A lot of research has been focused on piezoceramic ultrasonic transducers for direct contact, stretching the boundaries of use for piezoceramic based devices. Vermon (Tours, France) has developed an array probe based on a PZT/epoxy composite that is chemical and pressure
16
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22
resistant but operates up to only 100 C [7]. Piezoceramic composite transducers for operation in harsh environments have been developed and immersion tests have been carried out in Canola oil at 125 C [8]. An ultrasonic array has been developed for under sodium viewing using separate piezoceramic elements and data acquisition system and tests visualising targets in liquid sodium at 200 C have been reported [9]. More relevant to the higher temperature target considered here is the work by Kobayashi et al. [10,11] at McGill University, Montreal, Canada. They are working on thick film bismuth titanate/bismuth titanate composites with a possible operating temperature of 440 C. The high Curie temperature of LiNbO3, TC 1210 C, makes it a suitable material for this application, with an upper operating temperature limit known to be 650 C [12], beyond which the material suffers oxygen loss and shows low electrical resistivity. Therefore, LiNbO3 composites filled with cement have the potential to operate up to 650 C. The z-cut of crystalline LiNbO3 has an electro-mechanical coupling coefficient, kt, reported to be only 0.17. However, it is preferred over the y/36-cut in arrays because of its isotropic properties [13] whilst the y/36cut finds use in high frequency applications [14]. Previously, arrays made with monolithic plates of LiNbO3 have been fabricated and tested [15,16]. B-scans of defects were obtained; however it was observed that crystalline LiNbO3 tends to crack during cooling. This phenomenon is frequently observed in bulk crystalline LiNbO3. Therefore 1–3 connectivity LiNbO3 cement composites were proposed. Monolithic LiNbO3 has also been used in brazed-type transducer and tested at 550 C for 400 days [17]. The fabrication methods of 1–3 connectivity composites are well established [18] and benefits include an improvement in kt, broader bandwidth, reduced coupling of laterally propagating waves and the possibility to tailor the acoustic impedance. In a first attempt at a practical realisation, LiNbO3 cement composites were manufactured and tested experimentally [19]. A-scans were obtained at temperatures up to 336 C showing the reflected compression wave from the back wall of a test block. In this case, however, during cooling the sample detached from the test block to which it had been coupled with cement and the individual pillars showed evidence of the shattering previously seen in monolithic bulk material. In further, more recent tests [20] LiNbO3/epoxy composites were manufactured, characterized experimentally and modelled using the PZ Flex finite element analysis code (Weidlinger Associates, Los Altos, CA, USA). In both the experiments and the modelled results we observed an improvement in the thickness mode electro-mechanical coupling coefficient, kt, of 75% over the value of bulk material. This previous work suggests that there are two important requirements for the active material for a transducer to be used in condition monitoring at high temperatures. Firstly, the sensitivity of the material has to be sufficient for it to be
configured as an array. This is crucial in order to cover a large inspection region using beam steering with a fixed transducer. Secondly, the material has not only to withstand the high temperatures but also temperature cycling. In the remainder of this paper, we first introduce the RTV sealant and the cement used as passive fillers, then discuss composite parameters and present a B-scan of the back wall of a steel specimen at room temperature to show the sensitivity of the composite material. Furthermore, we demonstrate piezoelectric activity at elevated temperatures via electrical impedance measurements, which have also been used to validate the behaviour after temperature cycling. Finally, we present results from tests with the composite materials in transmit–receive mode at temperatures up to 360 C. 2. High temperature composites Composites of 1–3 connectivity were fabricated by the well known dice and fill method [21]. A saw pitch of
Fig. 1. Composite diced at 45 to ensure that electrodes cover equal areas of active material and that ‘‘dead’’ elements are avoided. LiNbO3 pillars are shown in light grey and outline of PCB tracks in dark grey. (a) Uniform coverage of active material. (b) How ‘‘dead’’ elements covering no active material can occur when the material is diced parallel to the edges. The illustrations are drawn to scale.
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22
Fig. 2. Cement (left) and RTV (right) lithium niobate composites. The pillar width is 0.7 mm and the volume fraction 45%. The composites are 10 mm · 10 mm in size.
approximately 1.1 mm and a pillar width of 0.7 mm resulted in a LiNbO3 volume fraction of 45%. The thickness of the material was approximately 1.5 mm. For single element transducer tests the material was cut to a size of 10 · 10 mm and for tests in an array configuration 10 · 35 mm material was used. In both cases the dicing direction was aligned at an angle of 45 to the crystal edges to ensure that an approximately equal area of active material would be covered by each electrode when a printed circuit board (PCB) with linear electrodes was applied for tests in array configuration (see Fig. 1). For the work reported here, z-cut LiNbO3 was used, despite its relatively low thickness mode electro-mechanical coupling factor as its properties are closer to isotropic than for other common cuts. This is important in consideration of future use in beam steered arrays; previous work [15] showed that the more sensitive y/36-cut material introduced minima in the beam profile of both single array elements and steered beam configurations. A high temperature composite also requires that the passive filler material can withstand the operating conditions. Two passive materials were used in the present work, RTV sealant suitable for use at temperatures up to 350 C (Cu371, Intel Adhesive Ltd., Blyth, Northumberland, UK) and high temperature cement for use up to 1600 C (C920, Cotronics, Brooklyn, NY, USA). The RTV itself is a flexible material; therefore the RTV composites are flexible, whilst the cement composites are rigid. Photographs of the composites are shown in Fig. 2. In this paper each material is termed according to its passive phase, i.e. ‘‘RTV composite’’ and ‘‘cement composite’’. 3. Experimental setup For sensitivity tests in transmit–receive mode in an array configuration at room temperature, an epoxy-glass PCB was clamped onto a cement composite to achieve electrical excitation and sensing. The PCB had rectangular electrodes 0.25 mm wide, with a spacing of 0.635 mm, which were long enough to overhang the composite at each side. Each
17
electrode was connected to an external connector pin. Due to the materials of the PCB, this setup can operate only at room temperature. However, thick film gold electrodes on alumina substrates have been used previously [15] and it is known that they can withstand the temperatures cited here. Both RTV and cement composites were characterized by electrical impedance measurements and tested in transmit– receive mode. For electrical impedance characterization the composite was electroded on both major surfaces with conductive silver paint and clamped between two steel blocks with thin steel plates used to connect to cables, electrically and acoustically insulated from the clamping blocks by cement plates as shown in Fig. 3a. The assembly was set on a hot plate for heating. At intervals the electrical impedance was measured as a function of frequency, with an HP 4192 A impedance analyser (Agilent Technologies, South Queensferry, UK). For transmit–receive testing, the composite was coupled on to a 37.5 mm thick steel block with UCA-HT couplant (Ely Chemicals, Ely, Cambridgeshire, UK). The top of the composite was again electroded with silver paint and a small steel plate and electrical insulator were used to clamp it onto the specimen as shown in Fig. 3b. Again, this assembly was set on a hot plate. The transducer was excited by applying a broadband electrical signal in the form of a negative going step with an amplitude of approximately 200 V and a fall time of approximately 20 ns, generated by a DPR300 pulser-receiver (JSR Ultrasound, Pittsford, NY, USA). The steel specimen itself served as the ground electrode and provided some acoustic damping. Return signals were amplified with the DPR300 and recorded with a digital oscilloscope.
4. Results To test the cement composite in the array configuration at room temperature, transmit–receive data were acquired from each of the 36 elements formed by the individual PCB tracks. A typical signal for this array configuration at room temperature, from element 13, is shown in Fig. 4a. Each linear electrode of the PCB excites parts of 6–7 pillars with a total electrode area 10 mm · 0.25 mm, thus covering only a relatively small area of active material as illustrated previously in Fig. 1. The relative peak-topeak (p–p) amplitudes of the back wall signals for all elements are shown in Fig. 4b, normalized to a maximum of 0 dB. Apart from one element with poor sensitivity, the amplitudes of the signal remain the same to within much less than 10 dB for all but one of the elements, which suggests good electrical coupling between the PCB and the pillars throughout the composite. Additionally, all signals are easily distinguished from the noise level, as shown in Fig. 4c, where the SNR was calculated from the peak to peak (p–p) amplitude of the back wall signal divided by the maximum p–p amplitude of the voltage recorded between the excitation ring down and the occurrence of
18
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22
Fig. 3. Illustration of experimental setup for (a) electrical impedance measurements at high temperature and (b) transmit–receive tests at high temperatures.
the back wall signal, hence producing conservative values for SNR. Electrical impedance as a function of frequency was measured at room temperature and during heating for an RTV composite. The temperature was stabilized at 250 C and the electrical impedance was measured at hourly intervals for 5 h. Fig. 5 shows impedance graphs at room temperature and one of five very similar graphs at 250 C, which were taken at hourly intervals. The absolute impedance magnitude decreased slightly with increasing temperature, as did the perturbation due to the piezoelectric resonance. The decease in impedance magnitude with temperature can be attributed to the positive
temperature coefficient of permittivity of LiNbO3. However, the measurements taken at 250 C during 5 h showed almost identical graphs in terms of perturbation and absolute impedance magnitude. These results were reproduced up to four times on the same piece of composite. The material regained the same absolute impedance magnitude after each cooling and still showed piezoelectric activity, indicating the absence of any curing effects on the composite during the first cycles and its viability for temperature cycling. The same impedance test has been carried out with a piece of bulk RTV material to investigate any variation in the relative permittivity, er, as a function of temperature. The very soft RTV material was clamped between two steel
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22 10000
0.005
0 12
13
14
15
16
-0.005
RT
Impedance magnitude [Ω]
Signal Amplitude [V]
0.01
19
-0.01
1000 1.5
Time [µs]
2
HT250
2.5
3
Frequency [MHz] Array Element 1
6
11
16
21
26
31
36
-75
RT
Impedance phase [˚]
Relative Amplitude [dB]
0
-5
-10
HT250
-80
-85
-90
-15
20 -95 1.5
SNR (dB)
2
2.5
3
Frequency [MHz]
15
Fig. 5. Electrical impedance as a function of frequency of RTV composite at room temperature (RT) and 250 C (HT250). (a) Impedance magnitude and (b) impedance phase.
10
5
0 1
6
11
16
21
26
31
36
Array Element
Fig. 4. For the cement composite in array configuration in (a) a typical pulse-echo signal from the back wall of the test block at room temperature, recorded with array element 13, in (b) relative signal amplitudes for all array elements and in (c) the signal to noise ratio for each element is shown.
blocks making it difficult to determine the precise thickness and therefore a precise value for er. However, the impedance of the RTV did not vary with temperature. This indicates a constant relative permittivity over the temperature range discussed in this paper. We therefore attribute the reduction in the impedance response not to changes in permittivity but to increased acoustic damping in the RTV as the temperature is increased. In transmit–receive tests with the RTV composite configured as a single element transducer, the signal from the back wall arrives after a 12.7 ls delay at room temperature and a slightly longer delay at 180 C, as expected [22]. Fig. 6 shows the data taken at room temperature and 180 C. The soft RTV is rather difficult to machine. With the present method of manual finishing with abrasive
paper, the RTV emerges slightly thicker than the LiNbO3 leaving an undulating surface profile, with the RTV as the peaks and the LiNbO3 as the troughs as illustrated in Fig. 7a. This causes problems in trying to achieve good coupling between the transducer and the steel test block, leading to relatively long signals and narrow bandwidth. The latter can be observed in Fig. 6b and d, where the presence of the 3rd and 5th harmonics of the fundamental frequency is also clear. These harmonics are the reason for the appearance of higher frequency operation near the beginning of each pulse in the time domain in Fig. 6a and b. Higher frequency operation does not occur near the end of the pulses because higher frequencies are more strongly attenuated. Furthermore, we observed a decrease in the amplitude of the back wall signal as the temperature increased and above 180 C it could no longer be distinguished from the noise. For the cement composite, the electrical impedance was measured at room temperature and up to 350 C. Again, the temperature was stabilized and kept at 350 C for 5 h and the impedance was measured at hourly intervals. This was repeated up to three times on the same piece of composite, which still showed piezoelectric activity after cooling. The data are presented in Fig. 8. It can be seen that the absolute impedance magnitude decreases with increas-
20
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22 0.8
Signal [V]
0.4
0.0 12
14
16
18
-0.4
Fig. 7. Schematic of the surface profile of (a) RTV composite and (b) cement composite. Light grey is LiNbO3 and dark grey is the passive material.
-0.8
Time [µs] 50 45
100000
35 30 25 20 15 10 5 0 0
5
10
15
Impedance magnitude [Ω]
Signal [arb. units]
40
RT HT350
10000
Frequency [MHz]
1000 1.5
2
0.8
0.4
3
-65
0.0 12
13
14
15
16
17
18
-0.4
Impedance phase [˚]
Signal [V]
2.5
Frequency [MHz]
RT
-75
HT350
-85
-0.8
Time [µs]
-95 1.5
Signal [arb. units]
20
2
2.5
3
Frequency [MHz]
Fig. 8. Electrical impedance as a function of frequency of cement composite at room temperature (RT) and 350 C (HT350). (a) Impedance magnitude and (b) impedance phase.
15
10
5
0 0
5
10
15
Frequency [MHz]
Fig. 6. Results from single element RTV composite. (a) The back wall signal at room temperature, (b) the frequency spectrum of the signal in (a), (c) the back wall signal at 180 C and (d) the frequency spectrum of the signal in (c).
ing temperature. Furthermore, we observe a shift of this resonance to lower frequencies with increasing temperature
and an increase in the perturbation of the impedance graph around the thickness mode resonant frequency during the first cycle, indicating curing effects on the composite. er as a function of temperature for the cement was determined in the same way as described for the RTV sealant. In this case, it was found to be stable at er = 7.2 ± 0.5 over the temperature range discussed here, and can therefore again be neglected as a reason for changes in electrical impedance with temperature. Tests in transmit–receive mode were conducted with the composite material as a single element transducer on the same specimen as before. Fig. 9 shows data taken at room temperature and 360 C, and corresponding frequency
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22 0.050
Signal [V]
0.025
0.000 12
13
14
15
16
17
18
-0.025
-0.050
Time [µs] 2.5
Signal [arb. units]
2.0
1.5
1.0
0.5
0.0
0
5
10
21
by manually removing the excess passive material and the LiNbO3 stock. Usually we would use precision lapping equipment for this operation; in this case however, the cement is water soluble and would be damaged. In addition the cement is much softer than the LiNbO3 and therefore abrades much faster. This leaves again an undulating surface profile, but this time with LiNbO3 as the thicker material (see Fig. 7b). Thus the cement composite is rather fragile; however, it does not experience the same problem in coupling as the RTV composite. Signals from the back wall were detected after a 12.7 ls delay at room temperature and after a slightly longer time difference at 360 C than at 180 C, because the higher temperature leads to a larger acoustic velocity reduction in the steel than at 180 C [22,23]. With the cement composite the signal amplitude did not decrease as significantly with temperature as for the RTV composite and the signal was still clearly visible at 360 C, which was the maximum temperature that could be achieved with the current experimental setup. However, the effects of slower ring down after the excitation signal obscure the signal to a greater extent at this temperature. The slower ring down is attributed to degradation of the couplant.
15
Frequency [MHz]
5. Conclusions
0.050
Signal [V]
0.025
0.000 12
13
14
15
16
17
18
-0.025
-0.050
Time [µs] 2.5
Signal [arb. units]
2.0
1.5
1.0
0.5
0.0 0
5
10
15
Frequency [MHz]
Fig. 9. Results from single element cement composite; (a) the back wall signal from a single element cement composite at room temperature, (b) the frequency spectrum of the signal in (a), (c) the back wall signal from the RTV composite at 360 C and (d) the frequency spectrum of the signal in (c).
spectra indicating broader bandwidth than for the RTV composite. Our cement composites were also fabricated
This paper has described the fabrication, characterization and testing of 1–3 connectivity LiNbO3 composites at room temperature as well as at elevated temperatures. The piezocomposite material showed good potential for implementation as the active material in high temperature ultrasound trandsucers. To test the sensitivity and suitability for imaging of LiNbO3 cement composites the material was successfully configured as an array at room temperature with elements of 10 mm · 0.25 mm. Echoes from the back wall of a steel specimen were obtained with each element of the array with satisfactory SNR. Both the RTV and the cement composites showed piezoelectric activity after several temperature cycles up to 250 and 350 C, respectively. Difficulties were experienced in the use of the RTV composites for transmit–receive tests because of poor acoustic coupling and consequent poor SNR. There is potential to improve this by using more advanced fabrication methods such as low temperature machining. However, signals from the back wall of a steel block were detected with the RTV composite prototypes at temperatures up to 180 C. In transmit–receive tests with the cement composite, signals from the back wall of a steel block were detected at temperatures up to 360 C, the maximum set by the experimental setup. Both the RTV and cement composites were thermally cycled without problems. As the materials have been demonstrated to be both functional and robust, we conclude that LiNbO3 composites are promising for the active elements in high temperature ultrasonic transducers and arrays.
22
N. Schmarje et al. / Ultrasonics 47 (2007) 15–22
References [1] R.C. Turner, P.A. Fuierer, R.E. Newnham, T.R. Shrout, Material for high temperature acoustic and vibration sensors: a review, Appl. Acoust. 41 (1994) 299–324. [2] Private communications, Doosan Babcock, Renfrew, UK. [3] L.C. Lynnworth, G. Jossinet, E. Cherifi, 300 Clamp-on ultrasonic transducer for measuring water flow and level, in: Procceedings of the IEEE Ultrasonics Symposium, 1996, pp. 407–412. [4] H.M. Frost, Electromagnetic ultrasonic transducer: principles, practice and applications, in: W.P Mason, R.N. Thurston (Eds.), Physical Acoustics, vol. XIV, 1985, pp. 179–275. [5] C.B. Scruby, R.J. Dewhurst, D.A. Hutchins, S.B. Palmer. Laser Generation of Ultrasound in Metals, in: R.S. Sharpe (Ed.), Research Techniques in Nondestructive Testing, vol. 5, 1982, pp. 281– 327. [6] R. Farlow, G. Hayward, Real-time ultrasonic techniques suitable for implementing non-contact NDT systems employing piezoceramic composite transducers, Br. J. Nondestr. Test. 36 (1994) 926–935. [7] C. Devallencourt, S. Michau, C. Bantignies N. Felix, A 5 MHz piezocomposite ultrasound array for operation in high temperature and harsh environment, in: IEEE Ultrasonics Symposium, 2004, pp. 1294–1297. [8] Q. Xue, M. Stanton, G. Elfbaum, A high temperature and broadband immersion 1–3 piezocomposite transducer for accurate inspection in harsh environments, in: IEEE Ultrasonics Symposium, 2003, pp. 1372–1375. [9] H. Karasawa, M. Izumi, T. Suzuki, S. Nagai, M. Tamura, S. Fijumori, Development of under-sodium three-dimensional visual inspection technique using matrix-arrayed ultrasonic transducer, J. Nucl. Sci. Technol. (2000) 769–779. [10] M. Kobayashi, C.-K. Jen Y. Ono, Lead free thick piezoelectric film as miniature high temperature ultrasonic transducer, in: IEEE Ultrasonics Symposium, 2004, pp. 910–913. [11] M. Kobayashi, C-K Jen, Piezoelectric thick bismuth titanate/lead zirconate titanate composite film transducers for smart NDE of metals, Smart Mater. Struct. 13 (2004) 951–956.
[12] BRITE/EURAM project BREU/CT92-0254, Project report: New piezoelectric ceramic with Tc > 1000C for operation up to 800C, Project coordinator: W.Wolny, Ferroperm A/S, Kvistgard, Denmark, 1992–1995. [13] I.B. Stumpf, Investigation of high density phased arrays for ultrasonic contact testing in steel, PhD Thesis, Paisley College of Technology (1982). [14] J.M. Cannata, T.A. Ritter, W.-H. Chen, R.H. Silverman, K. Kirk Shung, Design of efficient, broadband single-element (20–80 MHz) ultrasonic transducers for medical imaging applications, IEEE Trans. Ultrason. Ferroelect. Freq. Contr. 50 (2003) 1548–1557. [15] K. Kirk, A. McNab, A. Cochran, I. Hall, G. Hayward, Ultrasonic array for monitoring cracks in an industrial plant at high temperatures, IEEE Trans. Ultrason. Ferroelect. Freq. Contr. 46 (1999) 311– 319. [16] A. McNab, K.J. Kirk, A. Cochran, Ultrasonic transducers for high temperature applications, IEE Proc. Sci. Meas. Technol. 145 (5) (1998) 229–236. [17] T. Arakawa, K. Yoshikawa, S. Chiba, K. Muto, Y. Atsuta, Appliactions of brazed-type ultrasonic probes for high and low temperature uses, Nondestr. Test Eval. 7 (1992) 263–272. [18] W.A. Smith, The role of piezocomposites in ultrasonic transducers, in: Proceedings of the IEEE Ultrasonics Symposium, 1989, pp. 755– 766. [19] G. Shepherd, A. Cochran, K.J. Kirk, A. McNab, 1–3 connectivity composite material made from lithium niobate and cement for ultrasonic condition monitoring at elevated temperatures, Ultrasonics 40 (1–8) (2002) 223–226. [20] N. Schmarje, J.-F. Saillant, K. Kirk, S. Cochran, Imaging with lithium niobate/epoxy composites, Ultrasonics 42 (2004) 439–442. [21] H.P. Savakus, K.A. Klicker, R.E. Newnham, PZT-Epoxy piezoelectric transducer: a simplified fabrication procedure, Mater. Res. Bull. 16 (1981) 677–680. [22] J. Krautkra¨mer, H. Krautkra¨mer, Ultrasonic Testing of Materials, Springer-Verlag, Berlin, 1990, ISBN 0-387-51231-4, p. 497. [23] J.D.N. Cheeke, Fundamentals and Applications of Ultrasonic Waves, CRC Press, Boca Raton, FL, 2002, ISBN 0-8493-0130-0, pp. 86–87.